Researcher basic information

• Doctor of Engineering, Toyohashi University of Technology, Mar. 2012

• https://researchmap.jp/hiroki_sakaji?lang=en

• https://jglobal.jst.go.jp/detail?JGLOBAL_ID=201501074215806080
• https://orcid.org/0000-0001-5030-625X
• https://www.tetsuwaka.net

• 70722809

• https://orcid.org/0000-0001-5030-625X

• 201501074215806080

• Information Extraction
• Text Mining
• Financial Informatics
• Natural Language Processing
• Artificial Intelligence

• Informatics, Intelligent informatics, Natural Language Processing
• Humanities & Social Sciences, Library and information science, humanistic and social informatics

• Bachelor's degree program, School of Engineering
• Master's degree program, Graduate School of Information Science and Technology
• Doctoral (PhD) degree program, Graduate School of Information Science and Technology

Research activity information

• Jul. 2023, 14th IIAI International Congress on Advanced Applied Informatics, Competitive Paper Award
  Policy Gradient Stock GAN for Realistic Discrete Order Data Generation in Financial Markets
  Masanori Hirano;Hiroki Sakaji;Kiyoshi Izumi
• Jul. 2022, 10th International Congress on Advanced Applied Informatics, Outstanding Paper Award
  Investment Strategy via Lead Lag Effect using Economic Causal Chain and SSESTM Model
  Kei Nakagawa;Shingo Sashida;Hiroki Sakaji
• May 2022, The 2nd Workshop on Financial Technology on the Web, Best Paper Award
  Graph Representation Learning of Banking TransactionNetwork with Edge Weight-Enhanced Attention and Textual Information
  Naoto Minakawa;Kiyoshi Izumi;Hiroki Sakaji;Hitomi Sano
• 2022, Japanese Society for Artificial Intelligence, JSAI Incentive Award
  因果情報を用いた経済数値予測
  和泉潔;坂地泰紀;佐野仁美
• 2020, Japanese Society for Artificial Intelligence, JSAI Annual Conference Award
  STBM+: Advanced Stochastic Trading Behavior Model for Financial Markets based on Residual Blocks or Transformers
  Masanori Hirano;Kiyoshi Izumi;Hiroki Sakaji
• Aug. 2019, The First Workshop on Financial Technology and Natural Language Processing (FinNLP), Best Paper Award
  Economic Causal-Chain Search using Text Mining Technology
  Kiyoshi Izumi;Hiroki Sakaji
• 2019, Japanese Society for Artificial Intelligence, JSAI Incentive Award
  高頻度電力需要データを用いた製造業活動のナウキャスティングモデルの構築
  水門善之;和泉潔;坂地泰紀;島田尚;松島裕康
• 2019, Japanese Society for Artificial Intelligence, JSAI Annual Conference Award
  経済因果チェーン検索のシステム紹介と応用
  和泉潔;坂地泰紀
• Dec. 2017, IDRユーザフォーラム2017, 企業賞「Sansan株式会社」
  動画サイトのコメントを用いたタグ推定に関する研究
  坂地泰紀;小林暁雄;小花聖輝
• Oct. 2014, 金融情報学研究会, 優秀論文賞
  坂地 泰紀

• Efficient Assignment of Immediate Tasks Using Deep Reinforcement Learning in Multi-Agent Pickup and Delivery
  Taisei Hirayama; Kohei Yoshida; Hiroki Sakaji; Itsuki Noda
  23rd International Conference on Practical Applications of Agents and Multi-Agent Systems (PAAMS), 2026, [Peer-reviewed]
  English, International conference proceedings
• LDAS: Proposal for Semi-optimal Storage Locations Using LDA and SA
  Tatsuto Ito; Naoki Hattori; Hiroki Skaji; Itsuki Noda
  23rd International Conference on Practical Applications of Agents and Multi-Agent Systems (PAAMS), 2026, [Peer-reviewed]
  English, International conference proceedings
• CausalEnhance: Knowledge-Enhanced Pre-training for Causality Identification and Extraction
  Meiyun Wang; Kiyoshi Izumi; Hiroki Sakaji
  Knowledge-Based Systems, Elsevier BV, Sep. 2025, [Peer-reviewed]
  English, Scientific journal
• Genetic Optimization of Item Multiplicity and Location Assignments in Automated Warehouses
  Kosei Uemura; Kazuya Okamoto; Hiroki Sakaji; Itsuki Noda
  the 17th International Conference on Smart Computing and Artificial Intelligence (SCAI 2025) in 18th IIAI International Congress on Advanced Applied Informatics (IIAI AAI 2025), Jul. 2025, [Peer-reviewed]
  English, International conference proceedings
• Sentiment works in small-cap stocks: Japanese stock’s sentiment with language models
  Masahiro Suzuki; Yasushi Ishikawa; Masayuki Teraguchi; Hiroki Sakaji
  International Journal of Information Management Data Insights, Jun. 2025, [Peer-reviewed]
  English, Scientific journal
• Economy Watchers Survey Provides Datasets and Tasks for Japanese Financial Domain.
  Masahiro Suzuki; Hiroki Sakaji
  The ACM Web Conference 2025, 805, 808, May 2025, [Peer-reviewed]
  English, International conference proceedings
• Attempt to Extract and Analyse Causality from Agricultural Newspapers
  KATSURAGI Tetsuo; SAKAJI HIROKI; KOBAYASHI Akio; MORI Shotaro; OTOMO Masahiro; ISHIHARA Junichi; SUZUKI Masahiro; KAWAMURA Takahiro; NODA Itsuki
  Proceedings of the Annual Conference of JSAI, JSAI2025, 2Win582, 2Win582, The Japanese Society for Artificial Intelligence, 2025
  Japanese, In the Cabinet Office's Project of ``programs for Bridging the gap between R & d and the IDeal society (society 5.0) and Generating Economic and social value,'' the AI Agricultural Society Implementation Project aims to establish AI technologies to compensate for the decrease in agricultural labor force caused by the declining number of farmers. Among the challenges, the reduction of prefectural extension workers, who directly instruct farmers, is significant. This project is working on creating AI systems capable of answering questions like an extension agent, utilizing Large Language Models (LLM). At the National Agriculture and Food Research Organization (NARO), we are working to develop a system that can effectively respond to queries related to farming management—a key support need for extension agents. Our approach involves constructing datasets that represent the impact of weather on market conditions and aligning these data with their corresponding natural language expressions. Previous research has highlighted a critical shortage of such natural language data. In this study, we expand the existing dataset and investigate the correspondence between natural language expressions and actual market data, thereby establishing a foundational dataset for the future development of a Q &A system to support farming management guidance.
• Construction and Validation of a QA Dataset Specialized in Local Agriculture
  ITAKURA Ryoma; SUZUKI Masahiro; SAKAJI Hiroki; NODA Itsuki; KOBAYASHI Akio; OHTOMO Masahiro; ISHIHARA Junichi; KATSURAGI Tetsuo
  Proceedings of the Annual Conference of JSAI, JSAI2025, 1Win477, 1Win477, The Japanese Society for Artificial Intelligence, 2025
  Japanese, 本研究では、地方農業に特有の知識を含んだQAを構築し、その知識に関して既存の大規模言語モデル（LLM）がどのように理解しているか検証する。農業の栽培技術や取り組みは、国内の各自治体や公共団体、農協、その他の民間企業などがそれぞれ独自に取りまとめたデータがそれぞれのサイト上などから公開されている。しかしながら、これらの知識は機械可読な形式で提供されていることは稀で、現状のLLMなどでうまく活用できているとは言い難い。本研究では、長崎県の農業に関するデータから、LLMで利用しやすいQAデータセットの半自動構築を行った。その結果、既存のLLMには地方特有の農業知識が含まれていないことが明らかになった。
• Performance Comparison and Evaluation of Few-shot Example Selection for Generating Background, Causes, and Improvement Measures of Medical Incidents
  HASEYAMA Yuna; ITO Tomoki; SAKAJI Hiroki; NODA Itsuki
  Proceedings of the Annual Conference of JSAI, JSAI2025, 2L1OS2504, 2L1OS2504, The Japanese Society for Artificial Intelligence, 2025
  Japanese, In recent years, research has been conducted on methods for sample selection for few-shot learning, but there has been limited research related to medical incidents and patient safety. This study utilizes the Japanese Medical Incident Dataset (JMID), a dataset of medical incidents and near-miss cases collected and provided by the Japan Council for Quality Health Care and evaluates the generated results using BERTScore. Near-miss cases (referred to as ”hiyari-hatto” in Japanese) are incidents where accidents were narrowly avoided but could have potentially occurred. The JMID contains descriptions of various types of medical incidents and near-miss cases. In this study, these cases are classified into 18 categories based on similar content. In this paper, while presenting the results of zero-shot learning as a baseline, few-shot learning with similar cases and few-shot learning with randomly selected cases are compared and analyzed.
• Testing the Effectiveness of Knowledge Editing in Financial Polarity Editing
  HIRAMA Taiki; YUKI Ito; SAKAKI Hiroki; NODA Itsuki
  Proceedings of the Annual Conference of JSAI, JSAI2025, 3Win595, 3Win595, The Japanese Society for Artificial Intelligence, 2025
  Japanese, A major challenge for large language models (LLMs) is adapting to domain-specific knowledge and context efficiently. In this study, we propose a domain adaptation method using knowledge editing for sentiment analysis, where the criteria for positive and negative classifications change over time and across contexts (e.g., before and after the COVID-19 pandemic). While previous studies have demonstrated the effectiveness of knowledge editing, its use in sentiment classification remains underexplored. We explore a novel training approach leveraging knowledge editing and evaluate its effectiveness in a Japanese sentiment analysis task. Specifically, we propose an enhanced training method applying Rank-One Model Editing (ROME) to the LLaMA model and assess its performance in a zero-shot setting. The results show our approach achieves higher accuracy in sentiment classification than conventional methods. This study provides the first empirical validation of knowledge editing in sentiment analysis and highlights its potential as an efficient domain adaptation technique.
• Trade-off Analysis in Multi-objective Optimization of Storage Location Assignment using Genetic Algorithms
  UEMURA Kosei; SAKAJI Hiroki; NODA Itsuki; OKAMOTO Hironobu
  Proceedings of the Annual Conference of JSAI, JSAI2025, 3J1GS505, 3J1GS505, The Japanese Society for Artificial Intelligence, 2025
  Japanese, In warehouse optimization, it is crucial to balance not only improving processing speed but also ensuring smooth coordination with subsequent processes. In this paper, we focuse on the problem of optimization while balancing the two indices of delivery efficiency and order delay. To achive this, we use a genetic algorithm to analyze two indices in terms of the multiple placement of the same items. By formulating the problem as a multi-objective optimization problem and revealing its Pareto solution, we perform various analyses on delivery efficiency and order delay.
• Simulation Evaluation of the Introduction of Multiple Operation Modes for AI On-demand Transportation Service using Real Demand Data
  UNO Chiharu; OCHIAI Junichi; SAKAJI Hiroki; NODA Itsuki
  Proceedings of the Annual Conference of JSAI, JSAI2025, 3J1GS501, 3J1GS501, The Japanese Society for Artificial Intelligence, 2025
  Japanese, 本研究では、AI便乗サービスにおける車両ごとに異なる運行形態を混在させた配車方式を提案し、その効果を実運行データに基づくデマンド分布を用いたシミュレーションで評価することを目的とした。デマンドが多く発生する市街地エリアに専用車両を配置することで、デマンドの少ない郊外エリアに車両が流れることを防ぎ、運行効率の向上を目指した。運行形態の評価には交通シミュレータを使用し、異なる運行形態の効率性を比較分析した。シミュレーションでは、実地域のマップと、実データに基づいて作成したデマンドデータを使用した。
• SL(Storage Location)-LDA: Proposal for optimal Storage Location in Automated Warehouse using LDA and Simulated Annealing
  TATSUTO Ito; NAOKI Hattori; HIROKI Sakaji; ITSUKI Noda
  Proceedings of the Annual Conference of JSAI, JSAI2025, 3J1GS502, 3J1GS502, The Japanese Society for Artificial Intelligence, 2025
  Japanese, 自動倉庫における荷物配置の最適化は、実世界において運搬の効率化という側面で重要な問題である。しかし、MAPD問題はNP困難として知られており、最適な荷物配置を見つけるのは現実的ではない。荷物配置に関する手法は数多く存在し、SLAP(Storage Location Assignment Problems)として研究されている。私たちは新規手法として、LDAとSAを組み合わせたSL(Storage Location)-LDAを提案する。実際のデータに対して行った実験では、提案手法は既存手法と比べて優位な性能を持つことを示した。
• Narrowing-Down of Vehicles by Surrogate Model for Assignment Process in AI On-demand Transportation Service
  TAKASE Shohei; SAKAJI Hiroki; NODA Itsuki
  Proceedings of the Annual Conference of JSAI, JSAI2025, 2I5GS1004, 2I5GS1004, The Japanese Society for Artificial Intelligence, 2025
  Japanese, 本研究では，AI便乗サービスの最適配車計算のアルゴリズムについて，配車対象車両の絞り込みをサロゲートモデルにより行う方法を提案する．AI便乗サービスにおいて，最適配車計算のスケーラビリティを担保する技術開発は，サービスの大規模化を可能にするための重要な課題である．AI便乗サービスでは車両の運行台数が増加すると利便性が向上し，利用者数が増加すると採算性が向上するため，大規模な運行を行って多くの人が利用することで高い利便性を低価格で提供できる．しかし，AI便乗サービスにおける現行の最適配車計算のアルゴリズムには運行規模の拡大に伴って計算量が増大し，ユーザへのレスポンスが遅くなるという問題が存在する．AI便乗サービスの大規模運行に向けて，本研究は配車対象車両を絞り込むことで最適配車計算を効率化し，スケーラビリティを向上させることを目的とする．提案手法ではAttention機構を用いて当該車両による配車可能性を出力するサロゲートモデルを構成し，その出力に応じて絞り込みを行う．提案手法の有効性を確認するため，大規模な運行を想定したシミュレーション実験を行い，有用性を示した．
• Proposal and Evaluation of the BDI Logic Inference System GrAssBox Prover
  GONDO Hiraku; SAKAJI Hiroki; NODA Itsuki
  Proceedings of the Annual Conference of JSAI, JSAI2025, 2L5GS102, 2L5GS102, The Japanese Society for Artificial Intelligence, 2025
  Japanese, In this study, we propose a new BDI logic inference system called GrassBox Prover, which improves the efficiency of BDI operator computations through approximation using the hill-climbing method, bitwise operations, and setbased pruning. Until now, no systems have been proposed to handle predicate logic in BDI logic due to the significant computational time required for BDI operator processing. To address this issue, our research tackles this challenge by developing a new execution system for BDI logic. Furthermore, we evaluate the proposed GrassBox Prover in terms of the accuracy of approximate solutions related to BDI operators and its performance in proving BDI logical formulas.
• Optimal Assignment of Immediate Tasks in Multi-agent Pickup and Delivery
  Taisei Hirayama; Itsuki Noda; Hiroki Sakaji; Norihiko Kato
  The 29th International Conference on Technologies and Applications of Artificial Intelligence (TAAI 2024), 2025, [Peer-reviewed]
  English, International conference proceedings
• The attempt to improve the accuracy of predicting DPC codes for discharge summaries using causal information
  HATAKEYAMA Takuto; SAKAJI Hiroki; NODA Itsuki; TSUMOTO Shusaku; KIMURA Tomohiro
  JSAI Technical Report, Type 2 SIG, 2024, AIMED-014, 01, The Japanese Society for Artificial Intelligence, 21 Dec. 2024
  Japanese, This paper proposes a method to improve the prediction accuracy of Diagnosis Procedure Combination (DPC) codes in discharge summaries. First, the discharge summary data are formatted uniformly using LLM. Second, Causal Extraction is applied to extract causal information. Third, morphological analysis is performed on the original discharge summary data to create a vector of word features. Fourth, the causal information is used to highlight the features. Fifth, training examples of the classifier are generated. Finally, machine learning methods are applied to the training examples. Experimental validation results show that causal information is effective in improving the prediction accuracy of DPC codes.
• 自動倉庫における並列タスクキュー効率予測を行う代理モデル
  西澤 匠; 坂地 泰紀; 野田 五十樹; 服部 直樹; 岡本 和也
  人工知能学会第二種研究会資料, 2024, DOCMAS-026, 02, 一般社団法人 人工知能学会, 20 Dec. 2024
  Japanese, 本研究では、自動倉庫における並列タスクキューの評価のための代理モデルを提案する。自動倉庫の最適化は現実の物流における重要な課題であり、一般的にMAPD問題として定式化される。本研究では、MAPD問題へのタスク追加の方式を制御するため、並列タスクキューという枠組みを定義している。この並列タスクキューの最適化により、自動倉庫の効率化が行えることを期待している。提案する代理モデルはニューラルネットワークを用いてこの並列タスクキューから自動倉庫の性能を評価するものである。実験によって代理モデルは一定の精度が示され、並列タスクキューを用いた自動倉庫効率化に活用できることが期待される。
• 山登り法とビット演算を用いたBDI Proverの構築および評価
  権藤 拓; 坂地 泰紀; 野田 五十樹
  人工知能学会第二種研究会資料, 2024, DOCMAS-026, 01, 一般社団法人 人工知能学会, 20 Dec. 2024
  Japanese
• Refined and Segmented Price Sentiment Indices from Survey Comments
  Masahiro Suzuki; Hiroki Sakaji
  2024 IEEE International Conference on Big Data (BigData), 6642, 6650, IEEE, 15 Dec. 2024, [Peer-reviewed]
  English, International conference proceedings
• 多次元ガウス混合モデルによる携帯GPSデータの滞在目的分類
  音喜多 俊平; 坂地 泰紀; 野田 五十樹
  人工知能学会第二種研究会資料, 2024, SAI-051, 03, 一般社団法人 人工知能学会, 05 Dec. 2024
  Japanese
• Is ChatGPT the Future of Causal Text Mining? A Comprehensive Evaluation and Analysis.
  Takehiro Takayanagi; Masahiro Suzuki 0004; Ryotaro Kobayashi; Hiroki Sakaji; Kiyoshi Izumi
  IEEE Big Data, 6651, 6660, Dec. 2024, [Peer-reviewed]
  English, International conference proceedings
• Extraction of Research Objectives, Machine Learning Model Names, and Dataset Names from Academic Papers and Analysis of Their Interrelationships Using LLM and Network Analysis.
  S. Nishio; Hirofumi Nonaka; N. Tsuchiya; A. Migita; T. Banno; Teruaki Hayashi; Hiroki Sakaji; Takeshi Sakumoto; Kohei Watabe
  IEEM, 1377, 1381, Dec. 2024, [Peer-reviewed]
  International conference proceedings
• Enhancing Financial Domain Adaptation of Language Models via Model Augmentation.
  Kota Tanabe; Masanori Hirano 0001; Kazuki Matoya; Kentaro Imajo; Hiroki Sakaji; Itsuki Noda
  IEEE Big Data, 6661, 6669, Dec. 2024, [Peer-reviewed]
  English, International conference proceedings
• Knowledge Management for Automobile Failure Analysis Using Graph RAG.
  Yuta Ojima; Hiroki Sakaji; Tadashi Nakamura; Hiroaki Sakata; Kazuya Seki; Yuu Teshigawara; Masami Yamashita; Kazuhiro Aoyama
  IEEE Big Data, 6624, 6631, Dec. 2024, [Peer-reviewed]
  English, International conference proceedings
• Metadata-based Data Exploration with Retrieval-Augmented Generation for Large Language Models.
  Teruaki Hayashi; Hiroki Sakaji; Jiayi Dai; Randy Goebel
  IEEE Big Data, 6574, 6583, Dec. 2024, [Peer-reviewed]
  English, International conference proceedings
• Indexing Economic Fluctuation Narratives from Keiki Watchers Survey.
  Eriko Shigetsugu; Hiroki Sakaji; Itsuki Noda
  IEEE Big Data, 3871, 3879, Dec. 2024, [Peer-reviewed]
  English, International conference proceedings
• JaFIn: Japanese Financial Instruction Dataset
  Kota Tanabe; Masahiro Suzuki; Hiroki Sakaji; Itsuki Noda
  2024 IEEE Symposium on Computational Intelligence for Financial Engineering and Economics (CIFEr), 1, 10, IEEE, 22 Oct. 2024, [Peer-reviewed]
  English, International conference proceedings
• A Proposal of Score Distribution Predictive Model in Self-Play Deep Reinforcement Learning
  Kazuya Kagoshima; Hiroki Sakaji; Itsuki Noda
  Transactions of the Japanese Society for Artificial Intelligence, 39, 5, G, O63_1, Japanese Society for Artificial Intelligence, 01 Sep. 2024, [Peer-reviewed]
  Japanese, Scientific journal
• PGSGAN: Policy Gradient Stock GAN
  Masanori HIRANO; Hiroki SAKAJI; Kiyoshi IZUMI
  International Journal of Smart Computing and Artificial Intelligence, 8, 1, 1, 1, International Institute of Applied Informatics, Aug. 2024, [Peer-reviewed]
  English, Scientific journal
• Firm Default Prediction by GNN with Gravity-Model Informed Neighbor Node Sampling
  Naoto Minakawa; Kiyoshi Izumi; Yuri Murayama; Hiroki Sakaji
  The Review of Socionetwork Strategies, Springer Science and Business Media LLC, 29 Jul. 2024, [Peer-reviewed]
  English, Scientific journal, Abstract
  
  Firm default prediction is important in credit risk management and understanding economic trends. Both practitioners and academic researchers have long studied it. While traditional statistical methods such as discriminant analysis and logistic regression have been used recently, machine learning and deep learning methods have been widely applied. The graph neural network (GNN) is one of the latest applications of deep-learning approaches. With the use of GNNs, it is possible to reflect the non-linear relationships of features among neighboring companies around the target company, whereas ordinary machine learning and deep learning methods focus only on the features of the target company. However, when handling large-scale graphs such as inter-firm networks, it is difficult to apply vanilla GNNs naively. Although uniform neighbor node sampling is commonly used for large-scale graphs, to the best of our knowledge, no research has focused on better sampling methods for GNN applications for default prediction. From the practical viewpoint, it means which companies should be considered with priority for firm default prediction. In this study, we propose a novel gravity model-informed neighbor sampling method based on the estimated transaction volume by utilizing knowledge from econophysics. The scope of this research is to determine whether we can improve default predictions by considering neighboring companies with larger transaction amounts compared to ordinary uniform sampling. We also verified that the proposed method improves the prediction performance and stability compared to GNNs with other sampling techniques and other machine learning methods using real large-scale inter-firm network data.
• FinDeBERTaV2: Word-Segmentation-Free Pre-trained Language Model for Finance
  Masahiro Suzuki; Hiroki Sakaji; Masanori Hirano; Kiyoshi Izumi
  Transactions of the Japanese Society for Artificial Intelligence, 39, 4, FIN23, G_1, Japanese Society for Artificial Intelligence, 01 Jul. 2024, [Peer-reviewed]
  Scientific journal
• Development and analysis of medical instruction-tuning for Japanese large language models
  Issey Sukeda; Masahiro Suzuki; Hiroki Sakaji; Satoshi Kodera
  Artificial Intelligence in Health, 1, 2, 107, 107, AccScience Publishing, 08 Apr. 2024, [Peer-reviewed]
  English, Scientific journal, In the ongoing wave of impact driven by large language models (LLMs) like ChatGPT, the adaptation of LLMs to the medical domain has emerged as a crucial research frontier. Since mainstream LLMs tend to be designed for general-purpose applications, constructing a medical LLM through domain adaptation is a huge challenge. While instruction-tuning, particularly based on low-rank adaptation (LoRA), has become a frequently employed strategy to fine-tune LLMs recently, its precise roles in domain adaptation remain unknown. Here, we investigated how LoRA-based instruction-tuning improves the performance of Japanese medical question-answering tasks by employing a multifaceted evaluation of multiple-choice questions, including scoring based on “Exact match” and “Gestalt distance” in addition to the conventional accuracy. Our findings suggest that LoRA-based instruction-tuning can partially incorporate domain-specific knowledge into LLMs, with larger models demonstrating more pronounced effects. Furthermore, our results underscore the potential of adapting English-centric models for Japanese applications in domain adaptation, while also highlighting the persisting limitations of Japanese-centric models. This initiative represents a pioneering effort in enabling medical institutions to fine-tune and operate models without relying on external services.
• Enhancing risk analysis with GNN: Edge classification in risk causality from securities reports
  Hajime Sasaki; Motomasa Fujii; Hiroki Sakaji; Shigeru Masuyama
  International Journal of Information Management Data Insights, 4, 1, 100217, 100217, Elsevier BV, Apr. 2024, [Peer-reviewed]
  Scientific journal
• Metadata-based Clustering and Selection of Metadata Items for Similar Dataset Discovery and Data Combination Tasks
  Takeshi Sakumoto; Teruaki Hayashi; Hiroki Sakaji; Hirofumi Nonaka
  IEEE Access, 1, 1, Institute of Electrical and Electronics Engineers (IEEE), 11 Mar. 2024, [Peer-reviewed]
  Scientific journal
• Transit Allocation Method for Area-Divided Smart Access Vehicle Service
  UNO Chiharu; SAKAJI Hiroki; NODA Itsuki
  Proceedings of the Annual Conference of JSAI, JSAI2024, 3A5GS501, 3A5GS501, The Japanese Society for Artificial Intelligence, 2024
  Japanese, In this paper, we propose a transit allocation method for Smart Access Vehicle Sercvice. This method is devised to enable efficient allocation of vehicles to demand that extends over multiple business areas. There are two types of proposed methods: the Last method and the Next method. Simulation experiments were conducted to compare the two proposed methods with the current allocation method. The results of the experiments show that the proposed methods have advantages in several criterion of usabilities. From experiment results, we could find that the Next method was superior than the Last method.
• Trade-off Analysis between Complexity and Efficiency of Evacuation Plan in Artificial Maps
  UEMURA Kosei; SAKAJI Hiroki; NODA Itsuki
  Proceedings of the Annual Conference of JSAI, JSAI2024, 3A5GS502, 3A5GS502, The Japanese Society for Artificial Intelligence, 2024
  Japanese, Evacuation planning during disasters is considered to be a trade-off between complexity and efficiency. In this paper, we formulate the efficiency and complexity as a multi-objective optimization problem and show that various analyses can be performed by clarifying the Pareto solution.
• Verification of Reasoning Ability using BDI Logic and Large Language Model in AIWolf
  GONDO Hiraku; SAKAJI Hiroki; NODA Itsuki
  Proceedings of the Annual Conference of JSAI, JSAI2024, 2F6GS504, 2F6GS504, The Japanese Society for Artificial Intelligence, 2024
  Japanese, We attempt to improve the reasoning capability of LLM in werewolf game by combining BDI logic with LLM. LLM such as ChatGPT has been developed and used for various tasks. However, LLM has many challenges, and logical reasoning is one of them. Therefore, the purpose of this study was to verify the logical reasoning ability of LLM in dialogue of werewofl game by using BDI logic-based prompts. Experiments and evaluations were conducted using ``Werewolf,'' a communication game with incomplete information. From the results of the game played by five agents, we compare the logical reasoning ability of LLM by using the win rate and the vote rate against werewolf. We conducted a discussion using the results.
• Indexing Economic Fluctuation Narratives from Keiki Watchers Survey.
  Eriko Shigetsugu; Hiroki Sakaji; Itsuki Noda
  CoRR, abs/2412.01265, 2024
  Scientific journal
• Enhancing Financial Domain Adaptation of Language Models via Model Augmentation.
  Kota Tanabe; Masanori Hirano 0001; Kazuki Matoya; Kentaro Imajo; Hiroki Sakaji; Itsuki Noda
  CoRR, abs/2411.09249, 2024
  Scientific journal
• Advances in Language Processing in the Financial and Economic Domain
  Hiroki Sakaji; Kei Nakagawa
  Journal of Natural Language Processing, 31, 2, 763, 768, Association for Natural Language Processing, 2024
  Scientific journal
• LLMFactor: Extracting Profitable Factors through Prompts for Explainable Stock Movement Prediction.
  Meiyun Wang; Kiyoshi Izumi; Hiroki Sakaji
  ACL (Findings), 3120, 3131, 2024, [Peer-reviewed]
  English, International conference proceedings
• Indexing and Visualization of Climate Change Narratives Using BERT and Causal Extraction.
  Hiroki Sakaji; Noriyasu Kaneda
  2023 IEEE International Conference on Big Data (BigData), 5674, 5683, 15 Dec. 2023, [Peer-reviewed], [Lead author, Corresponding author]
  English, International conference proceedings
• From Base to Conversational: Japanese Instruction Dataset and Tuning Large Language Models
  Masahiro Suzuki; Masanori Hirano; Hiroki Sakaji
  2023 IEEE International Conference on Big Data (BigData), IEEE, 15 Dec. 2023, [Peer-reviewed]
  English, International conference proceedings
• Discovering new applications: Cross-domain exploration of patent documents using causal extraction and similarity analysis
  Meiyun Wang; Hiroki Sakaji; Hiroaki Higashitani; Mitsuhiro Iwadare; Kiyoshi Izumi
  World Patent Information, 75, Dec. 2023, [Peer-reviewed]
  Scientific journal
• Financial Causality Extraction Based on Universal Dependencies and Clue Expressions.
  Hiroki Sakaji; Kiyoshi Izumi
  New Generation Computing, 41, 4, 839, 857, Nov. 2023, [Peer-reviewed], [Lead author, Corresponding author]
  Scientific journal
• Constructing Sentiment Signal-Based Asset Allocation Method with Causality Information
  Rei Taguchi; Hiroki Sakaji; Kiyoshi Izumi; Yuri Murayama
  New Generation Computing, Springer Science and Business Media LLC, 11 Sep. 2023, [Peer-reviewed]
  Scientific journal, Abstract
  
  This study demonstrates whether financial text is useful for the tactical asset allocation method using stocks. This can be achieved using natural language processing to create polarity indexes in financial news. We perform clustering of the created polarity indexes using the change point detection algorithm. In addition, we construct a stock portfolio and rebalanced it at each change point using an optimization algorithm. Consequently, the proposed asset allocation method outperforms the comparative approach. This result suggests that the polarity index is useful for constructing the equity asset allocation method.
• llm-japanese-dataset v0: Construction of Japanese Chat Dataset for Large Language Models and Its Methodology
  Masanori Hirano; Masahiro Suzuki; Hiroki Sakaji
  Advances in Networked-based Information Systems, 442, 454, Springer Nature Switzerland, 24 Aug. 2023, [Peer-reviewed]
  In book
• Constructing and analyzing domain-specific language model for financial text mining
  Masahiro Suzuki; Hiroki Sakaji; Masanori Hirano; Kiyoshi Izumi
  Information Processing & Management, 60, 2, 103194, 103194, Elsevier BV, Mar. 2023, [Peer-reviewed]
  English, Scientific journal
• JMedLoRA: Medical Domain Adaptation on Japanese Large Language Models using Instruction-tuning.
  Issey Sukeda; Masahiro Suzuki 0004; Hiroki Sakaji; Satoshi Kodera
  CoRR, abs/2310.10083, 2023
  Scientific journal
• Policy Gradient Stock GAN for Realistic Discrete Order Data Generation in Financial Markets.
  Masanori Hirano 0001; Hiroki Sakaji; Kiyoshi Izumi
  IIAI-AAI, 361, 368, 2023
  International conference proceedings
• Gradual Further Pre-training Architecture for Economics/Finance Domain Adaptation of Language Model
  Hiroki Sakaji; Masahiro Suzuki; Kiyoshi Izumi; Hiroyuki Mitsugi
  2022 IEEE International Conference on Big Data (Big Data), IEEE, 17 Dec. 2022, [Peer-reviewed], [Lead author, Corresponding author]
  International conference proceedings
• A Model of Pricing Data and Their Constituent Variables Traded in Two-Sided Markets with Resale: A Subject Experiment
  Toshihiko Nanba; Kazuhito Ogawa; Naoki Watanabe; Teruaki Hayashi; Hiroki Sakaji
  2022 IEEE International Conference on Big Data (Big Data), IEEE, 17 Dec. 2022, [Peer-reviewed]
  International conference proceedings
• Extraction and classification of risk-related sentences from securities reports
  Motomasa Fujii; Hiroki Sakaji; Shigeru Masuyama; Hajime Sasaki
  International Journal of Information Management Data Insights, 2, 2, 100096, 100096, Elsevier BV, Nov. 2022, [Peer-reviewed]
  English, Scientific journal
• Models of Exchanged Datasets and Interactions of Buyers in the Data Market: Toward Multi-Agent Simulators for System Design
  Teruaki Hayashi; Hiroyasu Matsushima; Hiroki Sakaji; Yoshiaki Fukami; Takumi Shimizu
  Procedia Computer Science, 207, 1695, 1704, Elsevier BV, Sep. 2022, [Peer-reviewed]
  English, Scientific journal, 32117891
• Investment Strategy via Lead Lag Effect using Economic Causal Chain and SSESTM Model
  Kei Nakagawa; Shingo Sashida; Hiroki Sakaji
  10th International Congress on Advanced Applied Informatics (SCAI2022), 287, 292, Jul. 2022, [Peer-reviewed]
  English, International conference proceedings
• Forecasting Stock Price Trends by Analyzing Economic Reports With Analyst Profiles
  Masahiro Suzuki; Hiroki Sakaji; Kiyoshi Izumi; Yasushi Ishikawa
  Frontiers in Artificial Intelligence, 5, Frontiers Media SA, 07 Jun. 2022, [Peer-reviewed]
  English, Scientific journal, This article proposes a methodology to forecast the movements of analysts' estimated net income and stock prices using analyst profiles. Our methodology is based on applying natural language processing and neural networks in the context of analyst reports. First, we apply the proposed method to extract opinion sentences from the analyst report while classifying the remaining parts as non-opinion sentences. Then, we employ the proposed method to forecast the movements of analysts' estimated net income and stock price by inputting the opinion and non-opinion sentences into separate neural networks. In addition to analyst reports, we input analyst profiles to the networks. As analyst profiles, we used the name of an analyst, the securities company to which the analyst belongs, the sector which the analyst covers, and the analyst ranking. Consequently, we obtain an indication that the analyst profile effectively improves the model forecasts. However, classifying analyst reports into opinion and non-opinion sentences is insignificant for the forecasts.
• Attempt to Develop An Approach Based on BERT for Task of NTCIR-16 QA Lab-Poliinfo-3 Budget Argument Mining
  Akio Kobayashi; Hiroki Sakaji
  the 16th NTCIR Conference on Evaluation of Information Access Technologies, 199, 200, Jun. 2022, [Peer-reviewed]
  English, International conference proceedings
• Constructing Equity Investment Strategies Using Analyst Reports and Regime Switching Models
  Rei Taguchi; Hikaru Watanabe; Hiroki Sakaji; Kiyoshi Izumi; Kenji Hiramatsu
  Frontiers in Artificial Intelligence, 5, Frontiers Media SA, 18 May 2022, [Peer-reviewed]
  English, Scientific journal, This study demonstrates whether analysts' sentiments toward individual stocks are useful for stock investment strategies. This is achieved by using natural language processing to create a polarity index from textual information in analyst reports. In this study, we performed time series forecasting for the created polarity index using deep learning, and clustered the forecasted values by volatility using a regime switching model. In addition, we constructed a portfolio from stock data and rebalanced it at each change point of the regime. Consequently, the investment strategy proposed in this study outperforms the benchmark portfolio in terms of returns. This suggests that the polarity index is useful for constructing stock investment strategies.
• Concept and Practice of Artificial Market Data Mining Platform
  Masanori Hirano; Hiroki Sakaji; Kiyoshi Izumi
  2022 IEEE Symposium on Computational Intelligence for Financial Engineering and Economics (CIFEr), IEEE, May 2022, [Peer-reviewed]
  English, International conference proceedings
• Implementation of Actual Data for Artificial Market Simulation
  Masanori Hirano; Kiyoshi Izumi; Hiroki Sakaji
  The 21st International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2022), May 2022, [Peer-reviewed]
  English, International conference proceedings
• Graph Representation Learning of Banking TransactionNetwork with Edge Weight-Enhanced Attention and Textual Information
  Naoto Minakawa; Kiyoshi Izumi; Hiroki Sakaji; Hitomi Sano
  The 2nd Workshop on Financial Technology on the Web (FinWeb), May 2022, [Peer-reviewed]
  English, International conference proceedings
• STBM+: Advanced Stochastic Trading Behavior Model for Financial Markets using Residual Blocks or Transformers
  Masanori Hirano; Kiyoshi Izumi; Hiroki Sakaji
  New Generation Computing, 40, 1, 7, 24, Springer Science and Business Media LLC, Apr. 2022, [Peer-reviewed]
  English, Scientific journal
• SETN: Stock Embedding Enhanced with Textual and Network Information.
  Takehiro Takayanagi; Hiroki Sakaji; Kiyoshi Izumi
  IEEE International Conference on Big Data, 2377, 2382, IEEE, 2022, [Peer-reviewed]
  International conference proceedings
• SSAAM: Sentiment Signal-based Asset Allocation Method with Causality Information.
  Rei Taguchi; Hiroki Sakaji; Kiyoshi Izumi
  IEEE International Conference on Big Data, 2373, 2376, IEEE, 2022, [Peer-reviewed]
  International conference proceedings
• Bilateral Trade Flow Prediction by Gravity-informed Graph Auto-encoder.
  Naoto Minakawa; Kiyoshi Izumi; Hiroki Sakaji
  IEEE International Conference on Big Data, 2327, 2332, IEEE, 2022, [Peer-reviewed]
  International conference proceedings
• Growing Process of Communities on Data Platforms: Case Analysis of a COVID-19 Dataset
  Teruaki Hayashi; Takumi Shimizu; Yoshiaki Fukami; Hiroki Sakaji; Hiroyasu Matsushima
  2021 IEEE International Conference on Big Data (Big Data), IEEE, 15 Dec. 2021, [Peer-reviewed]
  English, International conference proceedings
• Retrieving of Data Similarity using Metadata on a Data Analysis Competition Platform
  Hiroki Sakaji; Teruaki Hayashi; Yoshiaki Fukami; Takumi Shimizu; Hiroyasu Matsushima; Kiyoshi Izumi
  2021 IEEE International Conference on Big Data (Big Data), IEEE, 15 Dec. 2021, [Peer-reviewed], [Lead author, Corresponding author]
  English, International conference proceedings
• Market Trend Analysis Using Polarity Index Generated from Analyst Reports
  Rei Taguchi; Hikaru Watanabe; Masanori Hirano; Masahiro Suzuki; Hiroki Sakaji; Kiyoshi Izumi; Kenji Hiramatsu
  2021 IEEE International Conference on Big Data (Big Data), IEEE, 15 Dec. 2021, [Peer-reviewed]
  English, International conference proceedings
• Data Combination for Problem-Solving: A Case of an Open Data Exchange Platform
  Teruaki Hayashi; Hiroki Sakaji; Hiroyasu Matsushima; Yoshiaki Fukami; Takumi Shimizu; Yukio Ohsawa
  The Review of Socionetwork Strategies, 15, 2, 521, 534, Springer Science and Business Media LLC, Nov. 2021, [Peer-reviewed]
  English, Scientific journal, Abstract
  
  In recent years, rather than enclosing data within a single organization, exchanging and combining data from different domains has become an emerging practice. Many studies have discussed the economic and utility value of data and data exchange, but the characteristics of data that contribute to problem-solving through data combination have not been fully understood. In big data and interdisciplinary data combinations, large-scale data with many variables are expected to be used, and value is expected to be created by combining data as much as possible. In this study, we conducted three experiments to investigate the characteristics of data, focusing on the relationships between data combinations and variables in each dataset, using empirical data shared by the local government. The results indicate that even datasets that have a few variables are frequently used to propose solutions for problem-solving. Moreover, we found that even if the datasets in the solution do not have common variables, there are some well-established solutions to these problems. The findings of this study shed light on the mechanisms behind data combination for solving problems involving multiple datasets and variables.
• Economic Causal-Chain Search and Economic Indicator Prediction using Textual Data
  Kiyoshi Izumi; Hitomi Sano; Hiroki Sakaji
  Proceedings of the 3rd Financial Narrative Processing Workshop, FNP 2021, 19, 25, 2021, [Peer-reviewed]
  International conference proceedings
• Predictive Uncertainty in Neural Network-Based Financial Market Forecasting
  Iwao Maeda; Hiroyasu Matsushima; Hiroki Sakaji; Kiyoshi Izumi; David deGraw; Atsuo Kato; Michiharu Kitano
  International Journal of Smart Computing and Artificial Intelligence, 5, 1, 1, 18, International Institute of Applied Informatics, 2021, [Peer-reviewed], [Invited]
  English, Scientific journal
• STBM: Stochastic Trading Behavior Model for Financial Markets
  Masanori Hirano; Hiroyasu Matsushima; Kiyoshi Izumi; Hiroki Sakaji
  Advances in Intelligent Systems and Computing, 157, 165, Springer International Publishing, 2021, [Peer-reviewed]
  In book
• Implementation of Real Data for Financial Market Simulation Using Clustering, Deep Learning, and Artificial Financial Market
  Masanori Hirano; Hiroyasu Matsushima; Kiyoshi Izumi; Hiroki Sakaji
  PRIMA 2020: Principles and Practice of Multi-Agent Systems, 3, 18, Springer International Publishing, 2021, [Peer-reviewed]
  In book
• Verification of Data Similarity using Metadata on a Data Exchange Platform
  Hiroki Sakaji; Teruaki Hayashi; Kiyoshi Izumi; Yukio Ohsawa
  2020 IEEE International Conference on Big Data (Big Data), IEEE, 10 Dec. 2020, [Peer-reviewed], [Lead author, Corresponding author]
  English, International conference proceedings
• Latent Segmentation of Stock Trading Strategies Using Multi-Modal Imitation Learning
  Iwao Maeda; David deGraw; Michiharu Kitano; Hiroyasu Matsushima; Kiyoshi Izumi; Hiroki Sakaji; Atsuo Kato
  JOURNAL OF RISK AND FINANCIAL MANAGEMENT, 13, 11, Nov. 2020, [Peer-reviewed]
  English, Scientific journal
• Estimating Manufacturing Activity via Machine Learning Analysis of High-frequency Electricity Demand Patterns
  Yoshiyuki Suimon; Kiyoshi Izumi; Hiroki Sakaji; Takashi Shimada; Hiroyasu Matsushima
  2020 9th International Congress on Advanced Applied Informatics (IIAI-AAI), IEEE, Sep. 2020, [Peer-reviewed]
  International conference proceedings
• Comparing Actual and Simulated HFT Traders' Behavior for Agent Design
  Masanori Hirano; Kiyoshi Izumi; Hiroyasu Matsushima; Hiroki Sakaji
  The Journal of Artificial Societies and Social Simulation, 23, 3, 6, JASSS, 30 Jun. 2020, [Peer-reviewed], [International Magazine]
  English, Scientific journal, Recently financial markets have shown significant risks and levels of volatility. Understanding the sources of these risks require simulation models capable of representing adequately the real mechanisms of markets. In this paper, we compared data of the high-frequency-trader market-making (HFT-MM) strategy from both the real financial market and our simulation. Regarding the former, we extracted trader clusters and identified one cluster whose statistical indexes indicated HFT-MM features. We then analyzed the difference between these traders' orders and the market price. In our simulation, we built an artificial market model with a continuous double auction system, stylized trader agents, and HFT-MM trader agents based on prior research. As an experiment, we compared the distribution of the order placements of HFT-MM traders in the real and simulated financial data. We found that the order placement distribution near the market or best price in both the real data and the simulations were similar. However, the orders far from the market or best price differed significantly when the real data exhibited a wider range of orders. This indicates that in order to build more realistic simulation of financial markets, integrating fine-grained data is essential.
• Forecasting Net Income Estimate and Stock Price Using Text Mining from Economic Reports
  Masahiro Suzuki; Hiroki Sakaji; Kiyoshi Izumi; Hiroyasu Matsushima; Yasushi Ishikawa
  Information, 11, 6, 292, 30 May 2020, [Peer-reviewed]
  English, Scientific journal
• Autoencoder-Based Three-Factor Model for the Yield Curve of Japanese Government Bonds and a Trading Strategy
  Yoshiyuki Suimon; Hiroki Sakaji; Kiyoshi Izumi; Hiroyasu Matsushima
  Journal of Risk and Financial Management, 13, 4, 82, 23 Apr. 2020, [Peer-reviewed], [International Magazine]
  English, Scientific journal
• Construction of Macroeconomic Uncertainty Indices for Financial Market Analysis Using a Supervised Topic Model
  Kyoto Yono; Hiroki Sakaji; Hiroyasu Matsushima; Takashi Shimada; Kiyoshi Izumi
  Journal of Risk and Financial Management, 13, 4, 79, 19 Apr. 2020, [Peer-reviewed], [International Magazine]
  English, Scientific journal
• Impact Analysis of Financial Regulation on Multi-Asset Markets Using Artificial Market Simulations
  Masanori Hirano; Kiyoshi Izumi; Takashi Shimada; Hiroyasu Matsushima; Hiroki Sakaji
  Journal of Risk and Financial Management, 13, 4, 75, 17 Apr. 2020, [Peer-reviewed]
  English, Scientific journal
• Deep Reinforcement Learning in Agent Based Financial Market Simulation
  Iwao Maeda; David deGraw; Michiharu Kitano; Hiroyasu Matsushima; Hiroki Sakaji; Kiyoshi Izumi; Atsuo Kato
  Journal of Risk and Financial Management, 13, 4, 71, Apr. 2020, [Peer-reviewed], [International Magazine]
  English, Scientific journal
• Concept Cloud-Based Sentiment Visualization for Financial Reviews
  Tomoki Ito; Kota Tsubouchi; Hiroki Sakaji; Tatsuo Yamashita; Kiyoshi Izumi
  Advances in Intelligent Systems and Computing, 1009 AISC, 183, 191, 2020, [Peer-reviewed]
  International conference proceedings
• Japanese Interest Rate Forecast Considering the Linkage of Global Markets Using Machine Learning Methods
  Yoshiyuki Suimon; Hiroki Sakaji; Kiyoshi Izumi; Takashi Shimada; Hiroyasu Matsushima
  International Journal of Smart Computing and Artificial Intelligence, 4, 1, 1, 17, International Institute of Applied Informatics, 2020, [Peer-reviewed]
  Scientific journal
• Value Creation through Digital Platforms in Data Exchange Ecosystems
  Shimizu Takumi; Hayashi Teruaki; Fukami Yoshiaki; Matsushima Hiroyasu; Sakaji Hiroki; Ohsawa Yukio
  Journal of Information and Communications Policy, 4, 1, 103, 123, Institute for Information and Communications Policy, 2020, [Peer-reviewed]
  Japanese,

  Value creation with digital data receives much attention as a source of innovation as the digitization of industries and economic systems advances. Data exchange ecosystems have emerged to facilitate data exchange and trade among heterogeneous actors, including private firms, non-profit organizations, public sectors, and individuals. Although the importance of nurturing data exchange ecosystems is well recognized, there are limited studies on developing and designing such ecosystems since there are many complex problems that need to be solved, such as legal, policy, and technology issues. Moreover, little is known about mechanisms behind data exchange platforms including the characteristics of data and the relationships between data while many discussions on business and technological aspects of data exchange ecosystems exist. Therefore, this paper unpacks data exchange mechanisms by analyzing the data combination networks on a digital platform. Our network analysis identifies 1) the characteristics of highly linkable data, 2) the conditions that facilitate data combination, and 3) structural characteristics of data networks on a digital platform. The paper demonstrates that variables related to "time" and "place" in the data facilitate data combination between diverse datasets. The paper also shows that the mixture of sensitive and shareable data increases the likelihood of link between the data from different fields. Furthermore, the structural characteristics of the data network (locally dense and global sparse) indicate the possibility of combining diverse data by filling the structural holes of sparse networks. Implications for developing data exchange platforms and promoting the data economy are discussed based on the findings of the paper.

• SSNN: Sentiment Shift Neural Network.
  Tomoki Ito; Kota Tsubouchi; Hiroki Sakaji; Tatsuo Yamashita; Kiyoshi Izumi
  Proceedings of the 2020 SIAM International Conference on Data Mining(SDM), 262, 270, SIAM, 2020, [Peer-reviewed]
  International conference proceedings
• Word-Level Contextual Sentiment Analysis with Interpretability.
  Tomoki Ito; Kota Tsubouchi; Hiroki Sakaji; Tatsuo Yamashita; Kiyoshi Izumi
  The Thirty-Fourth AAAI Conference on Artificial Intelligence(AAAI), 4231, 4238, AAAI Press, 2020, [Peer-reviewed]
  International conference proceedings
• Contextual Sentiment Neural Network for Document Sentiment Analysis.
  Tomoki Ito; Kota Tsubouchi; Hiroki Sakaji; Tatsuo Yamashita; Kiyoshi Izumi
  Data Sci. Eng., 5, 2, 180, 192, 2020, [Peer-reviewed]
  Scientific journal
• Measuring the Macroeconomic Uncertainty Based on the News Text by Supervised LDA for Investor’s Decision Making
  Kyoto Yono; Kiyoshi Izumi; Hiroki Sakaji; Takashi Shimada; Hiroyasu Matsushima
  Advances in Intelligent Systems and Computing, 125, 133, Springer International Publishing, 2020, [Peer-reviewed]
  In book
• Causal Sentence Extraction from Multilingual Texts Using End-to-End LSTM Models
  Yuta Niki; Hiroki Sakaji; Kiyoshi Izumi; Hiroyasu Matsushima
  The 2nd International Workshop on Cross-disciplinary Data Exchange and Collaboration (CDEC 2019), 17, 23, IEEE, Nov. 2019, [Peer-reviewed]
  English, International conference proceedings
• Extraction of Volitional Utterances from Japanese Local Political Corpus
  Hiroki Sakaji; Yasutomo Kimura; Izumi Kiyoshi; Hiroyasu Matsushima
  The 2nd International Workshop on Cross-disciplinary Data Exchange and Collaboration (CDEC 2019), 24, 29, IEEE, Nov. 2019, [Peer-reviewed]
  English, International conference proceedings
• Chain Bankruptcy Simulation Considering Investment Banks to Companies
  Ryo Hamawaki; Junichi Ozaki; Kiyoshi Izumi; Takashi Shimada; Hiroki Sakaji; Hiroyasu Matsushima
  IEEE International Conference on System, Man, and Cybernetics, 3784, 3790, IEEE, Oct. 2019, [Peer-reviewed]
  English, International conference proceedings
• Card Price Prediction of Trading Cards Using Machine Learning Methods
  Hiroki Sakaji; Akio Kobayashi; Masaki Kohana; Yasunao Takano; Kiyoshi Izumi
  The 8th International Workshop on Web Services and Social Media (WSSM-2019), in conjunction with the 22th International Conference on Network-Based Information Systems, (NBiS-2019), 705, 714, Springer, Sep. 2019, [Peer-reviewed]
  English, International conference proceedings
• Financial Text Data Analytics Framework for Business Confidence Indices and Inter-Industry Relations
  Hiroki Sakaji; Ryota Kuramoto; Hiroyasu Matsushima; Kiyoshi Izumi; Takashi Shimada; Keita Sunakawa
  Proceedings of the First Workshop on Financial Technology and Natural Language Processing, 40, 46, Aug. 2019, [Peer-reviewed]
  English, International conference proceedings
• mhirano at the FinSBD Task: Pointwise Prediction Based on Multi-layer Perceptron for Sentence Boundary Detection
  Masanori Hirano; Hiroki Sakaji; Kiyoshi Izumi; Hiroyasu Matsushima
  Proceedings of the First Workshop on Financial Technology and Natural Language Processing, 102, 107, Aug. 2019, [Peer-reviewed]
  English, International conference proceedings
• Analysis of Investment Effect using Multiplex Network Simulation of Banks and Firms
  Ryo Hamawaki; Junichi Ozaki; Kiyoshi Izumi; Takashi Shimada; Hiroyasu Matsushima; Hiroki Sakaji
  5th International Conference on Computational Social Science, Jul. 2019, [Peer-reviewed]
  English, International conference proceedings
• Effectiveness of Uncertainty Consideration in Neural-Network-Based Financial Forecasting
  Iwao Maeda; Hiroyasu Matsushima; Hiroki Sakaji; Kiyoshi Izumi; David Degraw; Atsuo Kato; Michiharu Kitano
  7th International Conference on Smart Computing and Artificial Intelligence (SCAI 2019), 673, 678, IEEE, Jul. 2019, [Peer-reviewed]
  English, International conference proceedings
• Analysis of Macro Economic Uncertainty from News Text with financial market
  Kyoto Yono; Kiyoshi Izumi; Hiroki Sakaji; Takashi Shimada; Hiroyasu Matsushima
  7th International Conference on Smart Computing and Artificial Intelligence (SCAI 2019), 661, 666, IEEE, Jul. 2019, [Peer-reviewed]
  English, International conference proceedings
• Extraction of relationship between Japanese and US interest rates using machine learning methods
  Yoshiyuki Suimon; Hiroki Sakaji; Kiyoshi Izumi; Takashi Shimada; Hiroyasu Matsushima
  7th International Conference on Smart Computing and Artificial Intelligence (SCAI 2019), 649, 654, IEEE, Jul. 2019, [Peer-reviewed]
  English, International conference proceedings
• Learning Uncertainty in Market Trend Forecast Using Bayesian Neural Networks
  Iwao Maeda; Hiroyasu Matsushima; Hiroki Sakaji; Kiyoshi Izumi; David Degraw; Hirokazu Tomioka; Atsuo Kato; Michiharu Kitano
  The International Conference on Decision Economics (DECON 2019), 210, 218, Springer International Publishing, Jun. 2019, [Peer-reviewed]
  English, International conference proceedings
• Extraction of Focused Topic and Sentiment of Financial Market by using Supervised Topic Model for Price Movement Prediction
  Kyoto Yono; Kiyoshi Izumi; Hiroki Sakaji; Hiroyasu Matsushima; Takashi Shimada
  IEEE Computational Intelligence for Financial Engineering & Economics (CIFEr) 2019, 149, 155, May 2019, [Peer-reviewed]
  English, International conference proceedings
• Short-term Stock Price Prediction by Analysis of Order Pattern Images
  Atsuki Nakayama; Kiyoshi Izumi; Hiroki Sakaji; Hiroyasu Matsushima; Takashi Shimada; Kenta Yamada
  IEEE Computational Intelligence for Financial Engineering & Economics (CIFEr) 2019, 84, 89, May 2019, [Peer-reviewed]
  English, International conference proceedings
• Japanese long-term interest rate forecast considering the connection between the Japanese and US yield curve
  Yoshiyuki Suimon; Hiroki Sakaji; Kiyoshi Izumi; Takashi Shimada; Hiroyasu Matsushima
  IEEE Computational Intelligence for Financial Engineering & Economics (CIFEr) 2019, 34, 40, May 2019, [Peer-reviewed]
  English, International conference proceedings
• Estimation of Tags Using Various Data for Online Videos
  Hiroki Sakaji; Akio Kobayashi; Masaki Kohana; Yasunao Takano; Kiyoshi Izumi
  The 33rd International Conference on Advanced Information Networking and Applications (AINA-2019), 301, 312, Springer, Mar. 2019, [Peer-reviewed]
  English, International conference proceedings
• Improving Document Similarity Calculation Using Cosine-Similarity Graphs
  Yasunao Takano; Yusuke Iijima; Kou Kobayashi; Hiroshi Sakuta; Hiroki Sakaji; Masaki Kohana; Akio Kobayashi
  The 33rd International Conference on Advanced Information Networking and Applications (AINA-2020), 512, 522, Springer, Mar. 2019, [Peer-reviewed]
  English, International conference proceedings
• Forecasting Crypto-Asset Price using Influencer Tweet
  Hirofumi Yamamoto; Hiroki Sakaji; Hiroyasu Matsushima; Yuki Yamashita; Kyohei Osawa; Kiyoshi Izumi; Takashi Shimada
  The 33-rd International Conference on Advanced Information Networking and Applications (AINA-2019), 940, 951, Springer, Mar. 2019, [Peer-reviewed]
  English, International conference proceedings
• Encoding of High-frequency Order Information and Prediction of Short-term Stock Price by Deep Learning
  Daigo Tashiro; Hiroyasu Matsushima; Kiyoshi Izumi; Hiroki Sakaji
  Quantitative Finance, 19, 9, 1499, 1506, 2019, [Peer-reviewed]
  English, Scientific journal
• Chain Bankruptcy Size in Inter-bank Network: the Effects of Asset Price Volatility and the Network Structure
  Ryo Hamawaki; Kiyoshi Izumi; Hiroki Sakaji; Takashi Shimada; Hiroyasu Matsushima
  Journal of Computational Social Science, 2, 1, 53, 66, 2019, [Peer-reviewed]
  English, Scientific journal
• Related Stocks Selection with Data Collaboration using Text Mining
  Masanori Hirano; Hiroki Sakaji; Shoko Kimura; Kiyoshi Izumi; Hiroyasu Matsushima; Shintaro Nagao; Atsuo Kato
  Information, 10, 3, 102, 102, 2019, [Peer-reviewed]
  English, Scientific journal
• Word-level Sentiment Visualizer for Financial Documents
  Tomoki Ito; Kota Tsubouchi; Hiroki Sakaji; Tatsuo Yamashita; Kiyoshi Izumi
  2019 IEEE CONFERENCE ON COMPUTATIONAL INTELLIGENCE FOR FINANCIAL ENGINEERING & ECONOMICS (CIFER 2019), 27, 33, 2019, [Peer-reviewed]
  English, International conference proceedings
• CSNN: Contextual Sentiment Neural Network
  Tomoki Ito; Kota Tsubouchi; Hiroki Sakaji; Kiyoshi Izumi; Tatsuo Yamashita
  2019 19TH IEEE INTERNATIONAL CONFERENCE ON DATA MINING (ICDM 2019), 1126, 1131, 2019, [Peer-reviewed]
  English, International conference proceedings
• Segment Information Extraction from Financial Annual Reports Using Neural Network.
  Tomoki Ito; Hiroki Sakaji; Kiyoshi Izumi
  Advances in Artificial Intelligence - Selected Papers from the Annual Conference of Japanese Society of Artificial Intelligence (JSAI 2019)(JSAI), 215, 226, Springer, 2019, [Peer-reviewed]
  International conference proceedings
• Economic Causal-Chain Search Using Text Mining Technology.
  Kiyoshi Izumi; Hiroki Sakaji
  Artificial Intelligence. IJCAI 2019 International Workshops - Macao(IJCAI), 23, 35, Springer, 2019, [Peer-reviewed]
  International conference proceedings
• The Extraction of the Future-Oriented Sentences from Annual Reports.
  Yoshinori Tanaka; Syunya Kodera; Fumihito Sato; Hiroki Sakaji; Kiyoshi Izumi
  8th International Congress on Advanced Applied Informatics(IIAI-AAI), 679, 684, IEEE, 2019, [Peer-reviewed]
  International conference proceedings
• Stock Price Analysis Using Combination of Analyst Reports and Several Documents.
  Masahiro Suzuki; Toshiya Katagi; Hiroki Sakaji; Kiyoshi Izumi; Yasushi Ishikawa
  2019 International Conference on Data Mining Workshops, 30, 36, IEEE, 2019, [Peer-reviewed]
  International conference proceedings
• Economic Causal Chain and Predictable Stock Returns.
  Kei Nakagawa; Shingo Sashida; Hiroki Sakaji; Kiyoshi Izumi
  8th International Congress on Advanced Applied Informatics(IIAI-AAI), 655, 660, IEEE, 2019, [Peer-reviewed]
  International conference proceedings
• Automatic Generation of Regional Business Sentiment Indexes using Contact Histories
  SAKAJI Hiroki; IZUMI Kiyoshi; MATSUSHIMA Hiroyasu; KAWASE Kazunari; HAYASHI Hiroshi
  Journal of Japan Society for Fuzzy Theory and Intelligent Informatics, 31, 2, 626, 635, Japan Society for Fuzzy Theory and Intelligent Informatics, 2019, [Peer-reviewed]
  Japanese, Scientific journal,

  In this paper, we propose a method to generate regional business sentiment indexes by using text of regional banks for business use. There are various textual data in the bank. In this study, we focused on contact histories from within those textual data. Contact histories are data recorded when employees communicate something with a customer, and various things are described. By analyzing contact histories, we thought that we can understand business sentiments of the area. Therefore, in this research, we generate regional sentiment business indexes from the contact histories. First, we selected the optimal model for business index creation using economy watchers survey. After that, we created regional sentiment business indexes using the model and compared it with the existing indicators.

• Enriching folksonomy for online videos
  Hiroki Sakaji; Masaki Kohana; Akio Kobayashi; Hiroyuki Sakai
  International Journal of Grid and Utility Computing, 10, 3, 258, 264, 2019, [Peer-reviewed]
  English, Scientific journal
• File Assignment Control for a Web System of Contents Categorization
  Kohana Masaki; Hiroki Sakaji; Akio Kobayashi; Shusuke Okamoto
  Transactions on Computational Collective Intelligence,Lecture Notes in Computer Science, 11610, 89, 102, 2019, [Peer-reviewed]
  English, Scientific journal
• 地方議会会議録における発言文推定手法の性能評価
  桧森 拓真; 木村 泰知; 坂地 泰紀; 荒木 健治
  日本知能情報ファジィ学会誌, 31, 2, 645, 652, 2019, [Peer-reviewed]
• 複数企業からの関連企業の抽出と事業内容に基づく分類
  田中 瑞竜; 酒井 浩之; 坂地 泰紀; 北島 良三
  日本知能情報ファジィ学会誌, 31, 1, 546, 562, 2019, [Peer-reviewed]
  Japanese, Scientific journal
• Selection of Related Stocks using Financial Text Mining
  Masanori Hirano; Hiroki Sakaji; Shoko Kimura; Kiyoshi Izumi; Hiroyasu Matsushima; Shintaro Nagao; Atsuo Kato
  The 1st International Workshop on Cross-disciplinary Data Exchange and Collaboration (CDEC) in 18th IEEE International Conference on Data Mining Workshops (ICDMW2018), 191, 198, IEEE, Nov. 2018, [Peer-reviewed]
  English, International conference proceedings
• Impact Assessments of the CAR Regulation using Artificial Markets
  Masahiro Hirano; Kiyoshi Izumi; Hiroki Sakaji; Takashi Shimada; Hiroyasu Matsushima
  International Workshop on Artificial Market 2018 (IWAM2018) in The 21st International Conference on Principles and Practice of Multi-Agent Systems (PRIMA2018), 43, 58, Oct. 2018, [Peer-reviewed]
  English, International conference proceedings
• Impact on Financial Markets of Dark Pools, Large Investor, and HFT
  Shin Nishioka; Kiyoshi Izumi; Wataru Matsumoto; Takashi Shimada; Hiroki Sakaji; Hiroyasu Matsushima
  International Workshop on Artificial Market 2018 (IWAM2018) in The 21st International Conference on Principles and Practice of Multi-Agent Systems (PRIMA2018), 4, 16, Oct. 2018, [Peer-reviewed]
  English, International conference proceedings
• Estimation of Cross-Lingual News Similarities Using Text-Mining Methods
  Wang Zhouhao; Liu Enda; Sakaji Hiroki; Ito Tomoki; Izumi Kiyoshi; Tsubouchi Kota; Yamashita Tatsuo
  JOURNAL OF RISK AND FINANCIAL MANAGEMENT, 11, 1, Mar. 2018, [Peer-reviewed]
  English, Scientific journal
• Extraction of sentences concerning business performance forecast and economic forecast from summaries of financial statements by deep learning
  Shiori Kitamori; Hiroyuki Sakai; Hiroki Sakaji
  2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings, 2018-, 1, 7, Institute of Electrical and Electronics Engineers Inc., 02 Feb. 2018, [Peer-reviewed]
  English, International conference proceedings
• Development of sentiment indicators using both unlabeled and labeled posts
  Tomoki Ito; Hiroki Sakaji; Kiyoshi Izumi; Kota Tsubouchi; Tatsuo Yamashita
  2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings, 2018-, 1, 8, Institute of Electrical and Electronics Engineers Inc., 02 Feb. 2018, [Peer-reviewed]
  English, International conference proceedings
• Discovery of rare causal knowledge from financial statement summaries
  Hiroki Sakaji; Risa Murono; Hiroyuki Sakai; Jason Bennett; Kiyoshi Izumi
  2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings, 2018-, 1, 7, Institute of Electrical and Electronics Engineers Inc., 02 Feb. 2018, [Peer-reviewed]
  English, International conference proceedings
• Extracting Laboratory Front Pages from University Websites
  Hiroki Sakaji; Atsuya Miyazaki; Hiroyuki Sakai; Kiyoshi Izumi
  ADVANCES IN NETWORK-BASED INFORMATION SYSTEMS, NBIS-2017, 7, 1117, 1125, 2018, [Peer-reviewed]
  English, International conference proceedings
• GINN: Gradient Interpretable Neural Networks for Visualizing Financial Texts
  Tomoki Ito; Hiroki Sakaji; Kiyoshi Izumi; Kota Tsubouchi; Tatsuo Yamashita
  International Journal of Data Science and Analytics, 9, 4, 431, 445, 2018, [Peer-reviewed]
  English, Scientific journal
• Web-based system for Japanese local political documents
  Ototake Hokuto; Sakaji Hiroki; Takamaru Keiichi; Kobayashi Akio; Uchida Yuzu; Kimura Yasutomo
  INTERNATIONAL JOURNAL OF WEB INFORMATION SYSTEMS, 14, 3, 357, 371, 2018, [Peer-reviewed]
• A Parallel Calculation Method on Web Browser for Contents Categorization.
  Masaki Kohana; Hiroki Sakaji; Akio Kobayashi; Shusuke Okamoto
  32nd International Conference on Advanced Information Networking and Applications Workshops, AINA 2018 workshops, Krakow, Poland, May 16-18, 2018, 40, 44, IEEE Computer Society, 2018, [Peer-reviewed]
• Text-Visualizing Neural Network Model: Understanding Online Financial Textual Data.
  Tomoki Ito; Hiroki Sakaji; Kota Tsubouchi; Kiyoshi Izumi; Tatsuo Yamashita
  Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Melbourne, VIC, Australia, June 3-6, 2018, Proceedings, Part III, 247, 259, Springer, 2018, [Peer-reviewed]
  International conference proceedings
• Application of Artificial Intelligence in Asset Management
  和泉 潔; 坂地 泰紀
  年金と経済, 36, 4, 23, 27, 年金総合研究センター, Jan. 2018, [Invited]
  Japanese
• A System for Classifying Proposals and Estimating Start Pages Stated in Notice of Annual Meeting of Shareholders
  Takano Kaito; Sakai Hiroyuki; Sakaji Hiroki; Izumi Kiyoshi; Okada Nana; Mizuuchi Toshikazu
  Journal of Natural Language Processing, 25, 1, 3, 31, 一般社団法人 言語処理学会, 2018, [Peer-reviewed]
  Japanese, <p>In this paper, we describe research on applied systems for realizing efficiency of work to store information of notice of annual meeting of shareholders in the database by using text mining technology. We aim to estimate start pages of proposals stated in notice of the meeting of shareholders and classify which proposal the page is. And we developed a system that automatically performs these tasks using text information of the notice of convocation of shareholders, and actually operates it. As a result of comparative experiment between our implemented system and conventional manual work, the working time was shortened to about 1/10. We propose three methods for classifying proposals. The first method classifies proposals by specialized terms extracted from training data. The second method classifies proposals by using deep learning. The final method classifies proposals by extracted proposal title. We evaluated our methods, and the effectiveness of each method was verified. </p>
• Development of an interpretable neural network model for creation of polarity concept dictionaries
  Tomoki Ito; Hiroki Sakaji; Kiyoshi Izumi; Kota Tsubouchi; Tatsuo Yamashita
  IEEE International Conference on Data Mining Workshops, ICDMW, 2017-, 1122, 1131, IEEE Computer Society, 15 Dec. 2017, [Peer-reviewed]
  English, International conference proceedings
• 金融テキストマイニングの最新技術動向 (特集 AIの金融応用(実践編))
  和泉 潔; 坂地 泰紀; 伊藤 友貴; 伊藤 諒
  証券アナリストジャーナル = Securities analysts journal, 55, 10, 28, 36, 日本証券アナリスト協会, Oct. 2017, [Invited]
  Japanese
• A Method for Extracting Correct Links from Automatic Created Links on Folksonomy
  Akio Kobayashi; Hiroki Sakaji; Masaki Kohana
  The 6th International Workshop on Web Services and Social Media (WSSM-2017), in conjunction with the 20th International Conference on Network-Based Information Systems (NBiS-2018), 1144, 1150, Springer, Aug. 2017, [Peer-reviewed]
  English, International conference proceedings
• A Topic Trend on P2P Based Social Media
  M. Kohana; H. Sakaji; A. Kobayashi; S. Okamoto
  2017 6th International Workshop on Web Services and Social Media, in conjunction with 2017 20th International Conference on Network-Based Information Systems, 1136, 1143, Aug. 2017, [Peer-reviewed]
  English, International conference proceedings
• A Web-Based Visualization System for Interdisciplinary Research Using Japanese Local Political Corpus
  Hokuto Ototake; Hiroki Sakaji; Keiichi Takamaru; Akio Kobayashi; Yuzu Uchida; Yasutomo Kimura
  Advances in Network-Based Information Systems, The 20th International Conference on Network-Based Information Systems (NBiS-2017), Aug. 2017, [Peer-reviewed]
• 決算短信PDFからの業績予測文の抽出
  北森詩織; 酒井浩之; 坂地泰紀
  電子情報通信学会論文誌 D, J100-D, 2, 150‐161, Feb. 2017, [Peer-reviewed]
  Japanese
• A Distributed Calculation Scheme for Contents Categorization
  Masaki Kohana; Hiroki Sakaji; Akio Kobayashi; Shusuke Okamoto
  2017 IEEE 31ST INTERNATIONAL CONFERENCE ON ADVANCED INFORMATION NETWORKING AND APPLICATIONS (AINA), 614, 620, 2017, [Peer-reviewed]
  English, International conference proceedings
• Lexicon Creation for Financial Sentiment Analysis using Network Embedding
  Ryo Ito; Kiyoshi Izumi; Hiroki Sakaji; Shintaro Suda
  Journal of Mathematical Finance, 7, 4, 869, 907, 2017, [Peer-reviewed]
  English, Scientific journal
• Investigation and construction of dictionary for analysis of comments in a video sharing site
  Yousuke Kawamoto; Yasuki Nishiyama; Akio Kobayashi; Hiroki Sakaji; Shigeru Masuyama
  4th IGNITE Conference and 2016 International Conference on Advanced Informatics: Concepts, Theory and Application, ICAICTA 2016, Institute of Electrical and Electronics Engineers Inc., 30 Dec. 2016, [Peer-reviewed]
  English, International conference proceedings
• Creating Japanese Political Corpus from Local Assembly Minutes of 47 Prefectures
  Yasutomo Kimura; Keiichi Takamaru; Takuma Tanaka; Akio Kobayashi; Hiroki Sakaji; Yuzu Uchida; Hokuto Ototake; Shigeru Masuyama
  Proceedings of the 12th Workshop on Asian Language Resources (ALR12)Proceedings of the 12th Workshop on Asian Language Resources (ALR12), The COLING 2016 Organizing Committee, 78, 85, Dec. 2016, [Peer-reviewed]
  English
• Estimation of Tags via Comments on Nico Nico Douga
  Hiroki Sakaji; Masaki Kohana; Akio Kobayashi; Hiroyuki Sakai
  PROCEEDINGS OF 2016 19TH INTERNATIONAL CONFERENCE ON NETWORK-BASED INFORMATION SYSTEMS (NBIS), 550, 553, 2016, [Peer-reviewed]
  English, International conference proceedings
• Extraction of polarity comments from Nico Nico Douga.
  Hiroki Sakaji; Junya Ishibuchi; Hiroyuki Sakai
  IJSSC, 6, 3, 165, 172, 2016, [Peer-reviewed]
• Extracting Causal Expressions from PDF Files of Summary of Financial Statements
  Hiroki Sakaji; Hiroyuki Sakai; Shigeru Masuyama
  電子情報通信学会論文誌D, J98-D, 5, 811, 822, May 2015, [Peer-reviewed]
  Japanese, Scientific journal
• Extraction of causal information from PDF files of the summary of financial statements of companies
  Hiroyuki Sakai; Hiroko Nishizawa; Shogo Matsunami; Hiroki Sakaji
  Transactions of the Japanese Society for Artificial Intelligence, 30, 1, 172, 182, Japanese Society for Artificial Intelligence, 06 Jan. 2015, [Peer-reviewed]
  Japanese, Scientific journal
• Extracting Polarity Comments from Nico Nico Douga
  Hiroki Sakaji; Junya Ishibuchi; Hiroyuki Sakai
  PROCEEDINGS 2015 18TH INTERNATIONAL CONFERENCE ON NETWORK-BASED INFORMATION SYSTEMS (NBIS 2015), 669, 672, 2015, [Peer-reviewed]
  English, International conference proceedings
• A Method for Extracting Sentences Including Causal Relations from Newspaper Articles
  SAKAJI Hiroki; MASUYAMA Shigeru
  The IEICE transactions on information and systems (Japanese edetion), 94, 8, 1496, 1506, The Institute of Electronics, Information and Communication Engineers, Aug. 2011, [Peer-reviewed]
  Japanese, Scientific journal, 本論文では,新聞記事から因果関係を含む文を自動的に抽出する手法を提案する.現在,ウェブページや新聞記事を含む大規模な機械可読文書が入手可能であり,その中には実アプリケーションに役立つ様々な情報が存在し,テキストマイニング技術を用いることで獲得することが可能である.そのような情報の一つに因果関係があり,本研究では因果関係の存在を示す手掛りとなる表現に基づいた因果関係を含む文の抽出を行った.その結果,人手により作られた辞書やパターンを用いず,自動的に因果関係を含む文を抽出することができた.本手法は,素性として構文的な素性と,意味的な素性を用いた.また,追加学習データを自動的に獲得することができる.その結果,性能が向上し,F値0.797を達成した.
• A Polarity Assignment Method to Basis Expressions of Economic Trends Extracted from Economic Newspaper Articles
  TANIGUCHI Shota; SAKAJI Hiroki; SAKAI Hiroyuki; MASUYAMA Shigeru
  The IEICE transactions on information and systems (Japanese edetion), 94, 6, 1039, 1043, The Institute of Electronics, Information and Communication Engineers, Jun. 2011, [Peer-reviewed]
  Japanese, Scientific journal, 経済新聞記事から抽出した景気動向を示唆する表現を対象として,その表現が景気の回復を示唆する表現なのか,景気の悪化を示唆する表現なのかのいずれかに自動的に分類する研究を行った.
• Cross-Bootstrapping: An Automatic Extraction Method of Solution-Effect Expressions from Patent Documents
  SAKAJI Hiroki; NONAKA Hirofumi; SAKAI Hiroyuki; MASUYAMA Shigeru
  The IEICE transactions on information and systems (Japanese edetion), 93, 6, 742, 755, The Institute of Electronics, Information and Communication Engineers, Jun. 2010, [Peer-reviewed]
  Japanese, Scientific journal, 特許文書から直接的なユーザの便益に相当する表現と,技術上の解決課題を示す表現を自動的に抽出するアルゴリズム「Cross-Bootstrapping」を提案する.特許出願件数は年間40万件にものぼり,1文書当りの文章量も膨大であるため,出願動向調査に有用なパテントマップ(特許出願動向を可視化したもの)を手作業で作成するには多大な時間とコストを要するため,その作成に役立つ情報を自動的に抽出する技術が求められている.そこで,本研究ではパテントマップの作成に役立つ「直接的なユーザの便益に相当する表現」と「技術上の解決課題を示す表現」を自動的に抽出する.本手法は,二つの手がかりと統計随報を用いて,ブートストラップ的に表現対を抽出する.また,辞書や人手により作成したパターンを用いず,自動的に表現を抽出することができる.最後に本手法の評価実験を行い,F値0.89と高い性能を達成したことを確認した.
• Extraction of the effect and the technology terms from a patent document
  Hirofumi Nonaka; Akio Kobayahi; Hiroki Sakaji; Yusuke Suzuki; Hiroyuki Sakai; Shigeru Masuyama
  40th International Conference on Computers and Industrial Engineering: Soft Computing Techniques for Advanced Manufacturing and Service Systems, CIE40 2010, 63, 105, 111, 2010, [Peer-reviewed]
  English, International conference proceedings
• Extracting Causal Knowledge Using Clue Phrases and Syntactic Patterns
  Hiroki Sakaji; Satoshi Sekine; Shigeru Masuyama
  PRACTICAL ASPECTS OF KNOWLEDGE MANAGEMENT, PROCEEDINGS, 5345, 111, +, 2008, [Peer-reviewed]
  English, International conference proceedings
• Automatic Extraction of Basis Expressions That Indicate Economic Trends.
  Hiroki Sakaji; Hiroyuki Sakai; Shigeru Masuyama
  Advances in Knowledge Discovery and Data Mining, 12th Pacific-Asia Conference, PAKDD 2008, Osaka, Japan, May 20-23, 2008 Proceedings, 977, 984, Springer, 2008, [Peer-reviewed]

• Comparison of Encoding Methods for Deep Reinforcement Learning in Improving Storage Location Assignment Efficiency of Automated Warehouses
  Sorachi, Kurita; Naoki, Hattori; Hiroki, Sakaji; Itsuki, Noda, 行動変容と社会システム vol.11, 2025, 26 Feb. 2025
  本研究では, 自動倉庫の荷配置効率化問題に深層強化学習を導入する際のエンコード手法の比較を行う. 荷配置効率化問題とは, 自動倉庫の全体的性能が効率化出来るように荷物配置を決めることをいう. 本問題においては, ルールベースでも, 様々な倉庫の制約下では組み合せ爆発が起こり計算コストが高い. そこで, 深層強化学習を行ったニューラルネットで荷配置を決定することを目指し, 本研究ではその導入として倉庫内環境のエンコード手法を比較する.
  In this study, we compare different encoding methods for introducing deep reinforcement learning into the storage location assignment efficiency problem of automated warehouses. The storage location assignment efficiency problem involves determining how to place items so that the overall performance of an automated warehouse is optimized. Even rule-based approaches, under various warehouse constraints, can lead to combinatorial explosions and high computational costs. Therefore, with the aim of using a neural network trained by deep reinforcement learning for storage location assignment, we compare two warehouse environment encoding methods as an initial step in this research., 情報処理学会, Japanese
• Development of a Dataset Representing Agricultural Product Prices for Generative AI of Agricultural Extension Specialist.
  小林暁雄; 坂地泰紀; 桂樹哲雄; 森翔太郎; 橋本祥; 鈴木雅弘; 川村隆浩, 人工知能学会全国大会論文集(Web), 38th, 2024
• Extraction SDGs-related sentences from Sustainability Reports using BERT and ChatGPT
  SASHIDA Masaki, JSAI Technical Report, Type 2 SIG, 2023, FIN-031, 55, 60, 10 Oct. 2023
  責任投資の原則（PRI）のもと機関投資家は、ESGの取り組みを取り入れた投融資をすることが求められている。しかし、現在日本においてはESG及びSDGsの取り組みに関する開示は、主にPDFベースのサステナビリティレポートや統合報告書で行なわれており、統一的なフォームは存在していない。そのため、ESG投資の責任者は手作業で各企業の取り組みの状況を評価する必要が生じ、多くの手がかかっている。本論文では、サステナビリティレポート等のPDFベースの情報からSDG関連の文を自動で抽出し、17のゴールに分類を行なうモデルの構築した。分類モデルは、事前学習済みのsentence-BERTモデルをファインチューニングして作成した。作成の過程で、SDGsのゴールのうち、「貧困」等に取り組んでいる企業が日本では少なく、学習データとなる事例が少ないという問題が生じた。そこで、ChatGPTを活用し、「貧困」等に関する文章を生成しそれを学習データとして用いることで、その課題の解決を図った。, The Japanese Society for Artificial Intelligence, Japanese
• Automotive Supply Chains Turbulence Index with Foot Traffic Data
  UEDA Tsubasa; IZUMI Kiyoshi; SAKAJI Hiroki, JSAI Technical Report, Type 2 SIG, 2023, FIN-030, 40, 44, 04 Mar. 2023
  COVID-19の流行以降、サプライチェーンの混乱が経済や資産市場に大きな影響を及ぼしている。政策当局や金融市場関係者の間で、供給関連指標に対する関心は高まっているが、データの粒度や迅速性の点で課題が残る。そこで、本研究では、オルタナティブデータと深層学習手法を用いて、自動車サプライチェーンの異常度をリアルタイムで測定する指数の構築を試みた。構築した指数は、ミクロ的な生産障害を把握する上で有用であり、既存の統計指標とも一定の関係性があることを確認した。, The Japanese Society for Artificial Intelligence, Japanese
• Proposal for Turning Point Detection Method Using Financial Text and Transformer
  Rei Taguchi; Hikaru Watanabe; Hiroki Sakaji; Kiyoshi Izumi; Kenji Hiramatsu, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 13859 LNAI, 171, 181, 2023
• New Intellectual Property Management Method Aiming at Expanding Technology Applications and Secondary Development
  WANG Meiyun; SAKAJI Hiroki; HIGASHITANI Hiroaki; IWADARE Mitsuhiro; IZUMI Kiyoshi, Proceedings of the Annual Conference of JSAI, JSAI2023, 2N1GS1004, 2N1GS1004, 2023
  In this study, we propose a new approach to intellectual property management to enhance knowledge discovery by combining causal extraction and similarity analysis. Existing research is limited to mining patent text data and predicting technology trends in the same field. Our proposed method can broaden the potential application of technology in multiple areas and enable the secondary development of the intellectual property. Specifically, we implement an approach to extract information from patent text data in various domains. Then, patent text analysis is performed using a pre-trained language model to compare information, and ultimately a causal chain for knowledge discovery is constructed. Data based on expert evaluation demonstrates that our method is more robust than existing deep learning methods., The Japanese Society for Artificial Intelligence, Japanese
• Proposing task to extract differences from time series financial documents
  SUZUKI Masahiro; SAKAJI Hiroki; IZUMI Kiyoshi, Proceedings of the Annual Conference of JSAI, JSAI2023, 1E3GS602, 1E3GS602, 2023
  In the natural language processing in the financial domain, many studies focus on the analysis of documents at a certain point in time, while various documents such as financial statements and financial results are published regularly.For investors with many stocks, it is not easy to read every detail of documents that exist at two points in time about each company continuously.It is also difficult to find out what has changed since the last time published.In this study, we propose a task to extract differences from two similar sentences in the two financial results written about the same company.We automatically extract similar parts from the two documents.For the extracted parts, we manually extract the differences between the two sentences.In addition, we conduct an evaluation experiment with a pre-trained language model., The Japanese Society for Artificial Intelligence, Japanese
• Proposed asset allocation based on sentiment signals considering causal relationships
  TAGUCHI Rei; SAKAJI Hiroki; IZUMI Kiyoshi, Proceedings of the Annual Conference of JSAI, JSAI2023, 2T1GS1001, 2T1GS1001, 2023
  This study demonstrates whether the financial text is useful for using stocks in the tactical asset allocation method. This can be achieved using natural language processing to create polarity indexes in financial news. We perform clustering of the created polarity indexes using the change point detection algorithm. In addition, we construct a stock portfolio and rebalanced it at each change point using an optimization algorithm. Consequently, the proposed asset allocation method outperforms the comparative approach. This result suggests that the polarity index is useful for constructing the equity asset allocation method., The Japanese Society for Artificial Intelligence, Japanese
• Financial Causality Extraction using Graph Neural Networks
  SAKAJI Hiroki; IZUMI Kiyoshi, Proceedings of the Annual Conference of JSAI, JSAI2023, 3A5GS601, 3A5GS601, 2023
  In this paper, we propose a method for extracting financial causal knowledge from multilingual text data. In the financial field, fund managers and financial analysts need causal knowledge in their work. Existing language processing techniques are very effective in extracting causal knowledge recognized by humans, but existing methods have two major problems. First, multilingual causality extraction has not been established so far. Second, the technology for extracting complex causal structures, such as nested causal knowledge, is insufficient. To solve these problems, we propose a method to extract nested causal knowledge based on clues (because, due to, etc.) and syntactic information. As a result of evaluating the proposed financial causal knowledge extraction method with multilingual text data in the financial field, it was demonstrated that the proposed model outperforms existing models., The Japanese Society for Artificial Intelligence, Japanese
• Prediction of China Market Index Using Bilingual Sentiment Analysis
  CONG Liu; SAKAJI Hiroki; IZUMI Kiyoshi; HAYAKAWA Tadaaki; TSUKAMOTO Kazuya; KATO Daisuke, JSAI Technical Report, Type 2 SIG, 2022, FIN-029, 28, 31, 08 Oct. 2022
  近年、中国経済の躍進に伴い、中国の各国経済に与える影響が高まっている。そのため、米国経済を中心に把握するだけではなく、中国経済の動向を把握することがより重要になっている。しかしながら、中国経済に言及した英語記事は中国語媒体よりも少なく、また、中国語で記載された中国経済に関する膨大な記事から選別してトピックを抽出することは現状難しい。そこで本研究では、中国語記事と英語記事の両方からセンチメントを獲得し、これらを合わせて利用することで、中国市場インデックスを予測する新たなモデルを提案する。, The Japanese Society for Artificial Intelligence, Japanese
• Framework for Predicting Monetary Policy of Central Bank Based on Text Mining
  SAKAJI Hiroki; KATO Daisuke; YOSHIDA Yusuke; WATANABE Tsuyoshi; HAYAKAWA Tadaaki; IZUMI Kiyoshi, JSAI Technical Report, Type 2 SIG, 2022, FIN-028, 113, 12 Mar. 2022
  本研究では，テキストマイニング技術を用いた中央銀行の金融政策変更を予測するフレームワークを提案する．中央銀行による金融政策の変更は為替市場や株式市場，国債市場などにも影響を与えることから，企業や銀行は金融政策変更の予兆を捉えようと試みているが，中央銀行によって公開される情報の量が限られることから非常に困難である．そこで，我々はテキストマイニング技術を用いて，金融政策変更の予兆に関する情報をニュース記事から抽出することを試みる．この問題を解決するために，我々は因果関係に基づく金融政策変更を予測する新たなフレームワークを提案する．本フレームワークはトピックモデル，機械学習に基づく文選択，機械学習に基づく期待値予測から構成される．, The Japanese Society for Artificial Intelligence, Japanese
• Analysis of the Relationship between Rainfall Observation Data and ATM Statistics Data using Location Information
  SANO Hitomi; MINAKAWA Naoto; SAKAJI Hiroki; IZUMI Kiyoshi, JSAI Technical Report, Type 2 SIG, 2022, FIN-028, 105, 12 Mar. 2022
  近年，局地的大雨の事象は年々増加し，水災害により甚大な経済的ダメージを被る事例が増えている．特に，被災地では当座の生活における安定的な決済手段が必要であり，これまでは現金決済が最も有用な手段の一つと考えられてきた．しかし，現在は，国内でキャッシュレス決済が推進され，都心部を中心に現金決済の比率が減少傾向であり，災害時における現金需要の行動様式にも変化をもたらす可能性が考えられる．このため，本研究では，自然災害のうち大雨の事例にフォーカスし，降雨情報（国土交通省のXRAIN）と金融情報（銀行ATM 統計情報）の異分野データを互いの位置情報を利用して融合する新たな手法により，降雨発生時における現金需要の動向を分析した．, The Japanese Society for Artificial Intelligence, Japanese
• Construction and Validation of a Pre-Training and Additional Pre-Training Financial Language Model
  SUZUKI Masahiro; SAKAJI Hiroki; HIRANO Masanori; IZUMI Kiyoshi, JSAI Technical Report, Type 2 SIG, 2022, FIN-028, 132, 12 Mar. 2022
  本研究では決算短信や有価証券報告書を用い，言語モデルのBERT とELECTRA について，事前学習や追加で事前学習(追加事前学習) を行いモデルを構築する．構築したモデルについて，金融ドメインのタスクによって汎用コーパスを用いたモデルとの性能を比較する．その際，ファインチューニングを行う層の数などパラメーターによる性能についても比較を行う．構築した一部のモデルについては一般に公開する．, The Japanese Society for Artificial Intelligence, Japanese
• Transaction Prediction using Textual Industry Information and Graph Neural Network
  MINAKAWA Naoto; IZUMI Kiyoshi; SAKAJI Hiroki; SANO Hitomi, JSAI Technical Report, Type 2 SIG, 2022, FIN-028, 124, 12 Mar. 2022
  金融機関の取引データを活用すると，企業の活動をリアルタイムで把握することが可能である．これらのデータを有効活用すれば与信管理をはじめ，CRM 等にも活用使途が見出せる．一方，主要な取引先以外の企業については取引量が少ない等，データ活用上は注意を要する．すなわち，潜在的な取引を見逃している可能性がある．本問題に対し，取引主体をノード，取引有無をエッジとした企業間ネットワークを構成すれば，取引有無の予測は本ネットワーク上のリンク予測問題として定式化できる．昨今のグラフ上の深層学習の進展に伴い，より精度の高いリンク予測が可能となった．本研究では，最近のグラフ深層学習手法を使った潜在的な取引予測について紹介する．, The Japanese Society for Artificial Intelligence, Japanese
• Policy Gradient Stock GAN for Realistic Discrete Order Data Generation in Financial Markets.
  Masanori Hirano 0001; Hiroki Sakaji; Kiyoshi Izumi, CoRR, abs/2204.13338, 2022
• Extraction of theme stocks using individual stock information and cross sectional stock information
  TAKAYANAGI Takehiro Takehiro; SAKAJI Hiroki; IZUMI Kiyoshi, Proceedings of the Annual Conference of JSAI, JSAI2022, 2A6GS601, 2A6GS601, 2022
  In this research, we propose a method to extract stocks related to a certain theme by using stock embeddings. Various investment trusts treat thematic mutual funds. Thematic mutual funds are created and sold according to the trends of the market and have been increasing in popularity, especially among Japanese retail investors. However, the stocks related to the theme are manually selected and it is costly to build. Therefore, in this research, We propose a method to help extract thematic mutual funds by embedding stocks’ individual information and cross-sectional information. As a result of evaluation experiments, our proposed method is better than the comparative method. It was shown that stocks can be extracted with high performance., The Japanese Society for Artificial Intelligence, Japanese
• Analysis of Cash Withdrawal Trends in the Case of Localized Heavy Rain by Data Fusion using Location Information
  SANO Hitomi; MINAKAWA Naoto; SAKAJI Hiroki; IZUMI Kiyoshi, Proceedings of the Annual Conference of JSAI, JSAI2022, 4N1GS302, 4N1GS302, 2022
  In recent years, localized heavy rain has occurred frequently, and economic damage cases caused by water disasters are increasing. However, there is no practical method for assessing the impact of the economy and financial system on climate change. This study focuses on localized heavy rain and develops the methods by data fusion using location information of weather information (rainfall observation information) and financial information (statistical information on withdrawals from ATMs). This paper introduces the analysis results of ATM cash withdrawal trends during localized heavy rains., The Japanese Society for Artificial Intelligence, Japanese
• Transaction prediction by using graph neural network and textual industry information
  MINAKAWA Naoto; IZUMI Kiyoshi; SAKAJI Hiroki; SANO Hitomi, Proceedings of the Annual Conference of JSAI, JSAI2022, 4S1IS2f01, 4S1IS2f01, 2022
  Transaction data owned by financial institutions can be alternative source of information to comprehend real-time corporate activities. Such transaction data can be applied to predict stock price and macroeconomic indicator as well as to sophisticate credit management, customer relationship management, and etc. However, it needs attention when a financial institution uses transaction data for aforementioned applications since occurrence of transactions depends on miscellaneous factors including customer loyalty, implying missing potential transactions. To solve this issue, we can predict occurrence of transactions by formulating the problem as a link prediction task in a transaction network with bank accounts as nodes and transaction volume as edges. With the recent advances in deep learning on graphs, we can expect better link prediction. We introduce an approach to predict transaction occurrence by using graph neural network with a special attention mechanism and textual industry information, analyzing the effectiveness of the proposed attention mechanism., The Japanese Society for Artificial Intelligence, Japanese
• Construction and Validation of a Pre-Trained Language Model Using Financial Documents
  SUZUKI Masahiro; SAKAJI Hiroki; HIRANO Masanori; IZUMI Kiyoshi, JSAI Technical Report, Type 2 SIG, 2021, FIN-027, 05, 09 Oct. 2021
  BERT を始めとする事前学習言語モデルは，様々な自然言語処理のタスクにおいて成果を上げている．これらのモデルの多くはWikipedia やニュース記事などの一般的なコーパスを用いているため，専門的な単語が使用される金融分野においては十分な効果が得られない．本研究では決算短信や有価証券報告書から事前学習言語モデルを構築する．また金融ドメインのタスクによって汎用モデルとの性能を比較する．, The Japanese Society for Artificial Intelligence, Japanese
• Economic Forecasting Model Based on Machine Learning Using Government Bond Market Data
  SUIMON Yoshiyuki; IZUMI Kiyoshi; SAKAJI Hiroki, JSAI Technical Report, Type 2 SIG, 2021, FIN-027, 101, 09 Oct. 2021
  国債のイールドカーブ情報には中央銀行の金融政策の方針が反映されるほか，市場参加者の物価や景気の見通しが織り込まれている．また金利水準自体は各種経済主体にとっての借り入れコストとなることから，先行きの経済活動に影響を与えうる．これらを踏まえ，本研究では，企業や家計などの経済主体による景気に先行的な経済活動をとらえた各種経済統計に加えて，国債イールドカーブの情報を特徴量として用いた，ニューラルネットワークベースの機械学習手法に基づく短期経済予測モデルを構築した．その結果，深層学習手法の一種であり再帰的なネットワーク構造を持つRNN（リカレントニューラルネットワーク）ベースのモデルにおいて，相対的に高い予測精度を確認した．更に，経済統計のみをモデルの特徴量として用いた場合に比べて，イールドカーブの情報も学習に用いた場合に，先行きの経済予測の精度が改善する傾向を確認した．このことは，経済予測において，イールドカーブに織り込まれる情報の有用性を示す結果と考える．, The Japanese Society for Artificial Intelligence, Japanese
• 特集：「ファイナンスにおける人工知能応用」「ファイナンスにおける人工知能応用」にあたって
  和泉 潔; 坂地 泰紀, 人工知能, 36, 3, 260, 261, 01 May 2021
  一般社団法人 人工知能学会, Japanese
• Stock Price Prediction Using Limit Order Book Data with Data Stratification and Multi-Phase Pre-training
  MATSUBARA Fuyuki; IZUMI Kiyoshi; SAKAJI Hiroki, JSAI Technical Report, Type 2 SIG, 2021, FIN-026, 69, 06 Mar. 2021
  Predicting the movement of stock price is an important issue for market participants. Recently, there have been many attempts applying machine learning techniques in financial time series prediction. However, overfitting presents a huge challenge when machine learning approaches are used in financial time series prediction. In this paper, we propose a stock price prediction method utilizing limit order book data from stocks other than target stocks by stratifying the data and holding a multi-phase pre-training considering market liquidity. Experimental results shows that the proposed approach enhances prediction performance., The Japanese Society for Artificial Intelligence, Japanese
• Financial Applications Using Text Mining and Machine Learning
  鈴木智也; 鈴木智也; 中川慧; 伊藤友貴; 坂地泰紀, 人工知能, 36, 3, 2021
• Forecast of Economic Indicators using Causal Information
  和泉潔; 坂地泰紀; 佐野仁美, 人工知能学会全国大会(Web), 35th, 2021
• 金融・経済ドメインを対象とした言語処理
  坂地 泰紀; 和泉 潔; 酒井 浩之, 自然言語処理 = Journal of natural language processing, 27, 4, 951, 955, Dec. 2020
  言語処理学会, Japanese
• An Attempt to Generate Market Comments via Causal Extraction and Data Expansion Based on Tags
  SAKAJI Hiroki; KONDO Yohei; KIYOSHI Izumi; NAGAO Shintaro; KATO Atsuo, JSAI Technical Report, Type 2 SIG, 2020, FIN-025, 34, 10 Oct. 2020
  資産運用会社や証券会社などの多くの金融機関では，定期的な（月次，週次等）市況コメントを作成し，顧客に提供している．この市況コメントを作成する際には，ニュース記事や公的機関発表の統計など，様々な情報源にあたり事実関係を確認しながら，実務者の知識や経験に照らして原稿を作成することになる．一方で，こうした市況コメントの調査・執筆の作業は時間とコストが伴うものである．もし，主要な情報源から，自然言語処理をはじめとした機械的なプロセスによって，信頼できる市況コメントを作成することができれば，それは金融機関における当該業務のコストの削減，あるいは処理の迅速化やカバレッジ拡大による顧客サービス向上に資することになる．そこで，本研究では，まずニュース記事に含まれる原因・結果を含む因果文を抽出することで，市況コメントを自動的に生成することを試みた．, The Japanese Society for Artificial Intelligence, Japanese
• Analysis of Value and Momentum Factors in Japanese Government Bond and Stock Index Using Machine Learning
  Fuyuki Matsubara; Kiyoshi Izumi; Hiroki Sakaji; Hiroyasu Matsushima, The 34th Annual Conference of the Japanese Society for Artificial Intelligence, 2020, 1K4-ES-2-05, 09 Jun. 2020
  There have been many studies seeking to predict excess returns in financial time series data. Nevertheless, not many studies have focused on applying machine learning approaches among factors in different asset classes. The main objective of this paper is to analyze whether a predictability of return in financial products could be improved by considering factors obtained from other asset classes, and to indicate the effectiveness of machine learning in financial time series prediction. We targeted 10-year Japanese Government Bond(10-year JGB), and Nikkei Stock Average Index, implementing non-linear machine learning approaches as well as conventional multiple linear regression models to predict returns in both assets. The results suggest that considering factors from other asset classes could improve return prediction both in 10-year JGB and Nikkei Stock Average, especially when using non-linear approaches., The Japanese Society for Artificial Intelligence, English, Summary national conference
• STBM : Stochastic Trading Behavior Model for Financial Markets Based on Long Short-Term Memory
  Masanori HIRANO; Hiroyasu MATSUSHIMA; Kiyoshi IZUMI; Hiroki SAKAJI, The 34th Annual Conference of the Japanese Society for Artificial Intelligence, 2020, 1K4-ES-2-04, 09 Jun. 2020
  English, Summary national conference
• 金融文書のための別タスク学習による教師なし重要文判定
  平野 正徳; 坂地 泰紀; 松島 裕康; 和泉 潔, 言語処理学会 第26回年次大会, 26th, 569, 572, 16 Mar. 2020
  Japanese, Summary national conference
• 企業リスク分析のための重要単語抽出と因果関係ネットワークの構築
  五十嵐 光秋; 坂地 泰紀; 和泉 潔; 島田 尚; 松島 裕康; 須田 真太郎, 言語処理学会 第26回年次大会, 26th, 16 Mar. 2020
  Japanese, Summary national conference
• 因果判定データセットの構築と原因結果表現抽出への拡張
  仁木 裕太; 坂地 泰紀; 松島 裕康; 和泉 潔, 言語処理学会 第26回年次大会, 26th, 16 Mar. 2020
  Japanese, Summary national conference
• 経済レポートを対象とした因果関係に着目するクエリ志向型複数文書要約
  高嶺 航; 坂地 泰紀; 和泉 潔; 松島 裕康; 島田 尚, 言語処理学会 第26回年次大会, 26th, 16 Mar. 2020
  Japanese, Summary national conference
• Analyzing Basis of Stock Returns Prediction Model Based Deep Learning Using Stock Attributions and Market Information
  KODERA Shunya; SATO Fumihito; SAKAJI Hiroki; IZUMI Kiyoshi, JSAI Technical Report, Type 2 SIG, 2020, FIN-024, 50, 14 Mar. 2020
  近年，個別株式のリターン予測において，様々なファクターの中から予測に有用な特徴量を自動で抽出することのできる深層学習技術の応用研究がなされている．しかしながら，深層学習は計算の過程が複雑で，人間にはその予測根拠の把握が難しく，意思決定に理由が求められる実務での利用において，解釈の困難さが課題視されることがある．一方，深層学習の解釈手法についても研究が行われており，深層学習において研究が盛んな画像分類等のタスクだけでなく株価指数や株式個別銘柄等の資産価格リターン予測を行う深層学習モデルに対しても解釈を行う研究が行われている．本稿では，モデルの解釈に焦点を当て，個別銘柄のリターン予測をタスクとした深層学習モデルについて，LRP と呼ばれる深層学習の解釈手法を用いて，各入力値の寄与度を個別銘柄レベルで確認した．さらに，深層学習モデルの入力値に個別銘柄属性だけでなくマーケット指数等の市場情報を用いることで，銘柄属性と市場トレンドの2 つの側面での分析を行った．, The Japanese Society for Artificial Intelligence, Japanese
• Empirical Study on Lead-Lag Effect with Economic Casual Chain
  NAKAGAWA Kei; SASHIDA Shingo; SAKAJI Hiroki; IZUMI Kiyoshi, JSAI Technical Report, Type 2 SIG, 2020, FIN-024, 171, 14 Mar. 2020
  A lead-lag effect in stock markets describes the situation where one (leading) stock return is cross-correlated with another (lagging) stock return at later times. There are various methods for stock return forecasting based on such a lead-lag effect. One of the most representative methods is based on the supply chain network. In this research, we propose a stock return forecasting method with an economic causal chain. The economic causal chain refers to a cause and effect network structure constructed by extracting a description indicating a causal relationship from the texts of Japanese financial statement summaries. We examine the following lead-lag effect. (1) whether lead-lag effect spreads to the 'effect' stock group when there is a large stock uctuation in the 'cause' stock group in the causal chain. (2) whether lead-lag effect spreads to the 'cause' stock group when there is a large stock uctuation in the 'effect' stock group in the causal chain. We confirm the existence of the both side of lead-lag effect and the evidence of stock return predictability across causally linked firms in the Japanese stock market., The Japanese Society for Artificial Intelligence, Japanese
• Development of Search Tool for Listed Company Related to Themes Using Text Mining
  Masanori Hirano; Hiroki Sakaji; Syoko Kimura; Kiyoshi Izumi; Hiroyasu Matsushima; Shintaro Nagao; Atsuo Kato, 24th JSAI Special interest Group on Financial Informatics, 226, 233, 14 Mar. 2020
  JSAI Special interest Group on Financial Informatics, Japanese, Summary national conference
• Autoencoder-Based Three-Factor Model for Yield Curve of Japanese Government Bond
  Yoshiyuki Suimon; Hiroki Sakaji; Kiyoshi Izumi; Takashi Shimada; Hiroyasu Matsushima, 24th JSAI Special interest Group on Financial Informatics, 66, 69, 14 Mar. 2020
  JSAI Special interest Group on Financial Informatics, Japanese, Summary national conference
• Analysis of Local Business Confidence indices and Inter-Industry Relations using Contact Histories
  Hiroki Sakaji; Ryota Kuramoto; Kiyoshi Izumi; Hiroyasu Matsushima; Takashi Shimada; Keita Sunagawa, 24th JSAI Special interest Group on Financial Informatics, 98, 102, 14 Mar. 2020
  JSAI Special interest Group on Financial Informatics, Japanese, Summary national conference
• The Construction of High Frequency Trading Strategy via Reinforcement Learning
  Hiroyuki Kobayashi; Kiyoshi Izumi; Hiroyasu Matsushima; Hiroki Sakaji; Takashi Shimada, 24th JSAI Special interest Group on Financial Informatics, 129, 133, 14 Mar. 2020
  JSAI Special interest Group on Financial Informatics, Japanese, Summary national conference
• 解釈可能なニューラルネットワークによるレビュー可視化
  伊藤友貴; 坪内孝太; 山下達雄; 坂地泰紀; 和泉潔, 言語処理学会年次大会発表論文集(Web), 26th, 2020
• テキストマイニングによるアナリストレポートを用いた株価動向予測
  鈴木雅弘; 堅木聖也; 坂地泰紀; 和泉潔; 石川康, 言語処理学会年次大会発表論文集(Web), 26th, 2020
• Further Pretraining BERT for Causality Existence Classification in Financial Domain
  仁木裕太; 坂地泰紀; 和泉潔; 松島裕康, 人工知能学会全国大会(Web), 34th, 2020
• Learning stock trading strategy by fusion of artificial market and deep reinforcement learning
  前田巌; 松島裕康; 坂地泰紀; 和泉潔; DEGRAW David; 加藤惇雄; 北野道春, 人工知能学会全国大会(Web), 34th, 2020
• Impact Analysis of Equity Investment on Chain Bankruptcy Using Inter-bank Network Model
  若杉亮; 和泉潔; 松島裕康; 坂地泰紀, 人工知能学会全国大会(Web), 34th, 2020
• Extraction of causal and complementary information for generating market analysis comments by automatic generation of training data
  酒井浩之; 坂地泰紀; 和泉潔; 松井藤五郎; 入江圭太郎, 人工知能学会全国大会(Web), 34th, 2020
• Netincome Forecast from Analyst Reports by Text Mining
  鈴木雅弘; 坂地泰紀; 和泉潔; 松島裕康; 石川康, 人工知能学会全国大会(Web), 34th, 2020
• Real-time Economic Indicator of Manufacturing Industry based on High-frequency Electricity Demand Data
  Yoshiyuki Suimon; Kiyoshi Izumi; Hiroki Sakaji; Takashi Shimada; Hiroyasu Matsushima, 35th SIG-SAI, JSAI Joint SIGs 2019, 1, 5, Nov. 2019
  Japanese, Summary national conference
• Importance of Uncertainty Estimation in Deep Learning
  Iwao Maeda; Hiroyasu Matsushima; Hiroki Sakaji; Kiyoshi Izumi; David Degraw; Hirokazu Tomioka; Atsuo Kato; Michiharu Kitano, Proceedings of the Annual Conference of JSAI, JSAI2019, 0, 4I2J201, 4I2J201, Jun. 2019
  

  In recent years, predictions by machine learning and deep learning methods are utilized in various scenes of society. A model trained with deep learning methods can predict the target with high accuracy, but can not consider the predictive confidence sufficiently, and may predict high confident for extrapolated data which is hard to predict. In this study, we applied ordinary deep learning methods and methods considering predictive uncertainty, proposed in recent years, to an image classification task, and verified the robustness of trained models against extrapolated data. Models trained with the ordinary deep learning methods predicted high confidence values for data having characteristics not existing in the training data, but models trained with the methods considering uncertainty predicted low confidence values for such data. By using methods considering uncertainty, it is possible to avoid mispredictions for extrapolated data. Experimental results suggest the importance of uncertainty estimation in deep learning.

  , The Japanese Society for Artificial Intelligence, Japanese, Summary national conference
• Investigating the Effect of Index Investing on Stock Price Formation
  Izuru Matsuura; Kiyoshi Izumi; Hiroki Sakaji; Hiroyasu Matsushima; Takashi Shimada, Proceedings of the Annual Conference of JSAI, JSAI2019, 0, 1P2-J-13-01, 1P2J1301, Jun. 2019
  

  インデックス投資が市場の価格形成に与える影響を調べるため，証券市場とその参加者，価格決定をモデル化した．そのモデル上でインデックス投資が価格形成にほとんど影響を与えないことをシミュレーション実験により示した．

  , The Japanese Society for Artificial Intelligence, Japanese, Summary national conference
• Measuring Macro Economic Uncertainty from News Text by Supervised LDA
  Kyoto Yono; Kiyoshi Izumi; Hiroki Sakaji; Takashi Shimada; Hiroyasu Matsushima, Proceedings of the Annual Conference of JSAI, JSAI2019, 0, 4J3J1302, 4J3J1302, Jun. 2019
  

  For nancial market participants, uncertainty of macro economic events have crucial impact when they make decision to buy or sell nancial assets because macro economic event in uence the asset price. In this study, we aim to build a model to measure macro economic uncertainty from news text. We proposed extended topic model which using not only news text data but also numeric data as a supervised signal for each news articles. We constructed four macro economic uncertainty indexes by our proposed model. Each indexes matches with historical macro economic events and the correlation are higher with the volatility of market index related to the uncertainty index.

  , The Japanese Society for Artificial Intelligence, Japanese, Summary national conference
• Construction of Causal Networks for Risk Finding Considering Word Polarities
  Hiroaki Igarashi; Hiroki Sakaji; Kiyoshi Izumi; Takashi Shimada; Hiroyasu Matsushima; Shintaro Suda, Proceedings of the Annual Conference of JSAI, JSAI2019, 0, 2O3J1304, 2O3J1304, Jun. 2019
  

  In this research, we conduct an experiment of constructing a causal networks within settlement account briefs. We extracted the causal relations from settlement briefs, and construct a network by connecting them by judging similarities. For calculating the similarities, we use a word2vec model created from the Japanese Wikipedia corpus.We use a method based on a combination of idf values which representing importance of words. In addition, by giving the polarities of causal expression using a polar dictionary, and judgment of synonyms that word2vec can't detect, we define so to speak, "negative relation".

  , The Japanese Society for Artificial Intelligence, Japanese, Summary national conference
• Analysis of Limit Orders by High-Frequency Traders in Tokyo Stock Exchange
  Masanori Hirano; Kiyoshi Izumi; Hiroyasu Matsushima; Hiroki Sakaji; Takashi Shimada, Proceedings of the Annual Conference of JSAI, JSAI2019, 0, 2O1J1304, 2O1J1304, Jun. 2019
  

  This study aimed to analyze the order behavior by High-frequency traders (HFT) called market making (MM) strategy. We used the order data of Tokyo Stock Exchange provided by Japan Exchange Group, Inc.. Firstly, we preprocessed the order data for merging virtual server used by the same traders. Secondly, we did a cluster analysis of traders based on some indexes indicating features of their trading strategy and extracted HFT-MM traders. Then, we calculated how many ticks their ordering price is far from the last executed price. As a result, we found some of their orders were placed at quite far (5-10 ticks) price from the last executed price for HFT-MM. This result means they mixed some strategies other than market making strategy and the strategies, possibly, will cause the unstabilizing effect when the market price is very fluctuating.

  , The Japanese Society for Artificial Intelligence, Japanese, Summary national conference
• Analysis of Inter-bank Network's Temporal Change Using Clustering Structure Change Detection
  Yuta Niki; Kiyoshi Izumi; Hiroyasu Matsushima; Hiroki Sakaji; Takashi Shimada, Proceedings of the Annual Conference of JSAI, JSAI2019, 0, 2O1J1303, 2O1J1303, Jun. 2019
  

  In this research, a clustering structure change detection method is employed for analyzing temporal changes of the inter-bank network. The inter-bank network is constructed from data of Italian deposit trading system e-MID, and changes in the structure of the network that focused on the type of the bank are analyzed. By using clustering structure change detection, the change of the inter-bank network's structure after Lehman shock is detected, which depends on the type of bank.

  , The Japanese Society for Artificial Intelligence, Japanese, Summary national conference
• Automatic Summarization of Analyst Reports Based on Causal Relationships from News Articles
  Wataru Takamine; Kiyoshi Izumi; Hiroki Sakaji; Hiroyasu Matsushima; Takashi Shimada; Yasuhiro Shimizu, Proceedings of the Annual Conference of JSAI, JSAI2019, 0, 1P2J1305, 1P2J1305, Jun. 2019
  

  In this paper, we focused on the causal relationships in both of news articles and analyst reports. We proposed a novel approach for summarizing analyst reports automatically based on the causal relationships extracted from both text data. As a rst step toward summarization of analyst reports adequately, we analyzed the validity of the method in extracting causal relationships which can be evaluated from the analyst reports. As a result, the proposed method could extract basis information of analyst's opinions from analyst reports with some accuracy, and we could conrm the styles of analysts in expression of opinions and bases.

  , The Japanese Society for Artificial Intelligence, Japanese, Summary national conference
• Analysis of Chain Bankruptcy Using Multilayer Networks Simulation among Banks and Companies
  Ryo Hamawaki; Junichi Ozaki; Kiyoshi Izumi; Hiroki Sakaji; Takashi Shimada; Hiroyasu Matsushima, Proceedings of the Annual Conference of JSAI, JSAI2019, 0, 2O3J1302, 2O3J1302, Jun. 2019
  

  近年経済のグローバル化によりある金融機関の破綻の影響がその業界や国を超えて国際的なものとなっている。 関連する分野の先行研究は、銀行単体もしくは企業単体でのエージェントシミュレーションを行うものがほとんどである。 そこで、本研究では銀行と企業の相互作用に焦点を当てた多層ネットワークのモデルを提案し、銀行から企業への投資の有無が企業の成長率に与える影響を調べる。 具体的には、投資を受けているかどうかによって成長率の変動幅を変化させ、最終的な企業の規模の分布の変化について議論する。 銀行が持つ貸借対照表は全国銀行協会が公表している全国銀行財務諸表分析に基づいて作成した。 シミュレーションの結果として、投資は企業の成長の平均には影響を与えなかったが、企業間の格差が広がった。 これは投資を受けた企業のうち、投資を活かして成長することができたものと失敗したものに別れてしまったことが原因だと考えられる。

  , The Japanese Society for Artificial Intelligence, Japanese, Summary national conference
• 不動産分野へのデータ解析と人工知能技術の応用
  和泉潔; 諏訪博彦; 大西立顕; 坂地泰紀; 松島裕康, 不動産テックを考える，赤木正幸，浅見泰司，谷山智彦編，プログレス, 75, 89, May 2019
• Automatic generation of market analysis comments by using relevant articles
  SAKAI Hiroyuki; SAKAJI Hiroki; IZUMI Kiyoshi; MATSUI Tohgoroh; IRIE Keitaro, JSAI Technical Report, Type 2 SIG, 2019, FIN-022, 61, 03 Mar. 2019
  本研究では、経済新聞記事などの経済テキストから、日経平均株価などの市況について言及している記事を抽出し、それらの内容を自動的に要約することによりマーケットレポートにおける市況分析コメントを自動生成する手法の開発を行う。しかし，日経平均株価の市況について言及している記事のみでは，指定した期間において重要な内容について言及している文の数が少なく，そのために重要な内容が市況分析コメントに含まれないことがある．そこで本研究では，ある期間において日経平均株価に影響を与えたイベントを推定し，そのイベントについて述べた記事を関連記事として自動的に検索し，検索された関連記事をも使用することで，生成する市況分析コメントの精度を高める手法を提案する．, The Japanese Society for Artificial Intelligence, Japanese
• Stock Price Analysis in Mid-to-long Term Using Analyst Reports
  KATAGI Toshiya; SAKAJI Hiroki; IZUMI Kiyoshi; ISHIKAWA Yasushi; KASAOKA Kohei, JSAI Technical Report, Type 2 SIG, 2019, FIN-022, 42, 03 Mar. 2019
  In this paper, we propose a methodology to forecast the direction and extent of volatility in mid-to-long term excess return of stock price by applying natural language processing and neural networks on the context of analyst reports. Analyst reports are prepared by analysts in research department in stock brokerage firms and we consider the content of reports include usefull information to forecast movements in stock prices. First, our method extracts opinion sentences from analyst reports, while the remaining parts correspond to non-opinion sentences. Second, our method predicts stock price movements by inputting opinion sentences and non-opinion sentences to neural networks separately., The Japanese Society for Artificial Intelligence, Japanese
• Impact Analysis of Index Investments on Market Price Formation via Simulation
  MATSUURA Izuru; IZUMI Kiyoshi; SAKAJI Hiroki; MATSUSHIMA Hiroyasu; SHIMADA Takashi, JSAI Technical Report, Type 2 SIG, 2019, FIN-022, 116, 03 Mar. 2019
  In this paper, we modeled stock markets to investigate the effect of index investing on stock price formation. We showed that index investing has little effect on stock price formation in our stock markets model by analyzing results from experiments with various market settings., The Japanese Society for Artificial Intelligence, Japanese
• 日米イールドカーブの連動性を用いた機械学習に基づく日本国債の長期金利予測
  水門 善之; 坂地 泰紀; 和泉 潔; 島田 尚; 松島 裕康, 第22回金融情報学研究会資料, 81, 87, Mar. 2019
  Japanese, Summary national conference
• Prediction of Crypto-Asset Price using In uencer Tweets
  Hirofumi Yamamoto; Hiroki Sakaji; Hiroyasu Matsushima; Yuki Yamashita; Kyohei Osawa; Kiyoshi Izumi; Takashi Shimada, 22nd JSAI Special interest Group on Financial Informatics, 25, 30, Mar. 2019
  Japanese, Summary national conference
• Automatic Summarization of Analyst Reports Based on Causal Relationship Text-Mined from News Reports
  Wataru Takamine; Hiroki Sakaji; Kiyoshi Izumi; Hiroyasu Matsushima; Takashi Shimada; Yasuhiro Shimizu, 22nd JSAI Special interest Group on Financial Informatics, 48, 52, Mar. 2019
  Japanese, Summary national conference
• Analysis of Investment Effect Using Multilayer Networks Simulation among Banks and Companies
  Ryo Hamawaki; Kiyoshi Izumi; Hiroki Sakaji; Takashi Shimada; Hiroyasu Matsushima, 22nd JSAI Special interest Group on Financial Informatics, 104, 107, Mar. 2019
  Japanese, Summary national conference
• analyzing the Effect of Index Investing on Stock Price Formation
  Izuru Matsuura; Kiyoshi Izumi; Hiroki Sakaji; Hiroyasu Matsushima; Takashi Shimada, 22nd JSAI Special interest Group on Financial Informatics, 116, 119, Mar. 2019
  Japanese, Summary national conference
• Simulation Analysis of Bank Network Stabilization by Liquidity Risk Management
  Taihei Sone; Kiyoshi Izumi; Hiroki Sakaji; Hiroyasu Matsushima; Takashi Shimada, 22nd JSAI Special interest Group on Financial Informatics, 131, 138, Mar. 2019
  Japanese, Summary national conference
• Japanese Economic forecasting model based on the Cabinet Office Indexes of Business Conditions
  Yoshiyuki Suimon; Hiroki Sakaji; Kiyoshi Izumi; Takashi Shimada; Hiroyasu Matsushima, SIG-SAI, 34, 4, 1, 6, Mar. 2019
  Japanese, Summary national conference
• Economic Rule Design by the Integration of Data Analysis and Agent-based Simulation
  Kiyoshi Izumi; Hiroyasu Matsushima; Hiroki Sakaji, IPSJ SIG Technical Report, 2019-ICS-194, 4, 1, 4, Mar. 2019
  Japanese, Summary national conference
• Stability Evaluation of Value Chain Negotiation System by Simulation
  Kiyoshi Izumi; Hiroyasu Matsushima; Hiroki Sakaji, 18th SICE Technical Committee on Social System, 159, Mar. 2019
  Japanese, Summary national conference
• 文書内における単語の共起を利用した上位下位概念の推定
  平野 正徳; 坂地 泰紀; 木村 笙子; 和泉 潔; 松島 裕康; 長尾 慎太郎; 加藤 惇雄, 言語処理学会第25回年次大会(NLP2019), 25th, 597, 600, Mar. 2019
  Japanese, Summary national conference
• 日米イールドカーブの連動性に基づく機械学習を用いた日本の金利変動モデルの構築
  水門 善之; 坂地 泰紀; 和泉 潔; 島田 尚; 松島 裕康, 第50回 日本金融・証券計量・工学学会 大会, Feb. 2019
  Japanese, Summary national conference
• Economic Causal Chain Search System and its Application
  和泉潔; 坂地泰紀, 人工知能学会全国大会(Web), 33rd, 2019
• Short-time market trend prediction by considering time series of high frequency order
  Iwao Maeda; Hiroyasu Matsushima; Hiroki Sakaji; Kiyoshi Izumi; David deGraw; Hirokazu Tomioka; Atsuo Kato, JSAI Special interest Group on Financial Informatics, 21, 50, 52, Oct. 2018
  Japanese, Summary national conference
• Development of Platforms for Investors using Financial Text Mining
  Hiroki Sakaji; Kiyoshi Izumi; Hiroyasu Matsushima, 21st JSAI Special interest Group on Financial Informatics, 59, 60, Oct. 2018
  Japanese, Summary national conference
• Automatic generation of market analysis comments from financial articles
  SAKAI Hiroyuki; SAKAJI Hiroki; IZUMI Kiyoshi; MATSUI Tohgoroh; IRIE Keitaro, JSAI Technical Report, Type 2 SIG, 2018, FIN-020, 44, 20 Mar. 2018
  本研究では，経済新聞記事などの経済テキストから、日経平均株価などの市況について 言及している文書のみを抽出し，それらの内容を自動的に要約することにより，ファンドの運用報告 書における市況分析コメントを自動生成する手法の開発を行う．本手法では，まず経済新聞記事から 深層学習により日経平均株価の市況について言及している記事を抽出する．次に抽出された記事の 中から例えば「ギリシャへの金融支援協議が難航していることや、中国・上海株の値動きへの警戒感 から、投資家のリスクオフの動きが強まった。」のような日経平均が大幅に変動した理由について言 及している文を抽出する．そして，抽出された文を時系列順に並べることで市況分析コメントを自動 生成する．, The Japanese Society for Artificial Intelligence, Japanese
• Generation Support of Financial Reports by Textmining
  MARUZAWA Hidemasa; IZUMI Kiyoshi; SAKAJI Hiroki; TAMURA Hiromichi; MOTOHIRO Mamoru, JSAI Technical Report, Type 2 SIG, 2018, FIN-020, 74, 20 Mar. 2018
  Recently, with the increase of individual investors, the necessity of investment support technologies is increasing. Although analyst reports on which professional securities analysts forecast business performances or stock prices of companies are regarded as important investment decision materials, writing an analyst report is heavy burden. In this research, we summarize newspaper articles and support the generation of analyst reports by using knowledge of information features which are referred to as reasons for analysts' forecasts of business performances or stock prices in analyst reports., The Japanese Society for Artificial Intelligence, Japanese
• Validation of economic text visualization using deep learning
  ITO Tomoki; SAKAJI Hiroki; IZUMI Kiyoshi, JSAI Technical Report, Type 2 SIG, 2018, FIN-020, 61, 20 Mar. 2018
  経済文書のような専門的な文書は非専門家にとって読みにくい場合が多い．そのため，非専門家を対象に経済文書上のセンチメントを単語単位で可視化するようなサポートシステムを構築することには一定の需要があると思われる．経済文書上のセンチメントを可視化する手段の一つとして近年提案された「ニューラルネットワークモデルの解釈」に関する手法，LRP (Layer-wiseRepresentation Propagation) を用いるという手段がある．しかし現状 LRP が日本語の経済文書の可視化に有用かどうかは調査されておらず，その性質についての詳細な分析もあまりされていない．また，LRP の Attention RNN への適用方法は未だ提案されていない．本報告では LRP の AttentionRNN への適用方法を提案し，また，LRP が日本語金融テキストの可視化に有用かどうかを検証する．さらに，実データを用いた検証の中で LRP を用いた日本語文書可視化の性質について分析する．, The Japanese Society for Artificial Intelligence, Japanese
• Extraction of Causal Relation Chains using Vector Expressions
  NISHIMURA Kouhei; SAKAJI Hiroki; IZUMI Kiyoshi, JSAI Technical Report, Type 2 SIG, 2018, FIN-020, 50, 20 Mar. 2018
  複数のテキストデータから経済・金融事象を背景知識まで含めて可視化することは, 経済・金融事象の理解の助けになり有用である. しかしながら, 経済・金融事象の連鎖を手動で抽出することは非常に時間とコストがかかる. そこで, 本研究では経済・金融事象の連鎖を因果関係として扱い, 各事象を表したベクトル間の類似度を用いて因果関係の連鎖を構築する手法を提案する. また,提案手法における問題点から今後の提案をまとめる., The Japanese Society for Artificial Intelligence, Japanese
• Real time sentiment analysis of Bank of Japan using text of Financial report and macroeconomic index
  YONO Kyoto; IZUMI Kiyoshi; SAKAJI Hiroki, JSAI Technical Report, Type 2 SIG, 2018, FIN-020, 67, 20 Mar. 2018
  金融市場におけるテキストデータは、投資家にとって分析対象の一つであり、その重要性は日に日に増している。決算短信をはじめとする企業の業績について書かれたドキュメント、証券会社のアナリストが書いた個別企業についてのアナリストレポート、更にはツイッター等のSNSで発信している個人投資家のつぶやき等、多種多様なテキストデータが存在する。定性的な投資判断の材料になるとともに、これらのテキストを用いて、モデルを構築し、定量的な分析を行うこともある。本研究では、中央銀行が発行する議事録やステートメントなどの公的な文章を対象に、定量的な数値化を目的としている。具体的には、日本銀行の発行する金融政策決定会合の議事要旨から物価、生産、雇用等の各トピックに対するセンチメント指数の構築を目指す。, The Japanese Society for Artificial Intelligence, Japanese
• A Text Analysis Platform for Extracting Domain-Specific Logical Relations
  坂地 泰紀; 和泉 潔, 電子情報通信学会技術研究報告 = IEICE technical report : 信学技報, 117, 468, 21, 24, 02 Mar. 2018
  電子情報通信学会, Japanese
• 地方議会会議録における発言文の推定
  桧森拓真; 木村泰知; 坂地泰紀, 言語処理学会第24回年次大会(NLP2018), P1-3, 125, 128, Mar. 2018
  Japanese, Summary national conference
• Extraction of Causal Knowledge from Annual Securities Report
  SATO Fumihito; SAKUMA Hiroaki; KODERA Shunya; TANAKA Yoshinori; SAKAJI Hiroki; IZUMI Kiyoshi, Proceedings of the Annual Conference of JSAI, JSAI2018, 2O404, 2O404, 2018
  In annual securities report, various information such as corporate policy, risk management, R&D, and so on, is included other than business performance. Previous researches proposed the extraction methods of important sentences containing causal information from financial articles and texts but not annual financial reports. In this paper, we applied these extracting methods based on SVM discriminant model to annual securities reports in our original way. Our method indicated high performance and all evaluations, that were precision, recall and F-score, showed more than 0.8. By using our model, useful information from annual securities reports would be collected effectively, which allow us to make unique investment decisions., The Japanese Society for Artificial Intelligence, Japanese
• A Text Analysis Platform for Extracting Domain-Specific Logical Relations
  坂地泰紀; 和泉潔, 電子情報通信学会技術研究報告, 117, 468(AI2017 43-51), 2018
• 金融レポート,およびマクロ経済指数による日銀センチメント指数の構築
  余野京登; 和泉潔; 坂地泰紀, 人工知能学会全国大会論文集(CD-ROM), 32nd, 2018
• An Experiment of Causal Knowledge Extraction from Reuter News
  坂地泰紀; BENNETT Jason; 宮尾祐介; 和泉潔, 人工知能学会全国大会論文集(CD-ROM), 32nd, 2018
• Creation of Causal Relation Network using Semantic Similarity
  西村弘平; 坂地泰紀; 和泉潔, 人工知能学会全国大会論文集(CD-ROM), 32nd, 2018
• Effect of Asset Price Fluctuation and Inter-bank Lending and Borrowing Network on Chain Bankruptcy of Bank
  濱脇諒; 和泉潔; 坂地泰紀; 米納弘渡, 人工知能学会全国大会論文集(CD-ROM), 32nd, 2018
• Extraction of Newspaper Articles including Cause Informations Concerning Business Performances for each Industry
  MARUZAWA Hidemasa; IZUMI Kiyoshi; SAKAJI Hiroki; TAMURA Hiromichi, JSAI Technical Report, Type 2 SIG, 2017, FIN-019, 71, 14 Oct. 2017
  These days, a growing number of individual investors is attracting public attentions even in Japan, and securities companies are actively providing them with investment informations. Especially, analyst reports written by professional security analysts are important investment judgment materials, but their timing of publication varies by brands. In this paper, using the structures of causal relationships of sentences in analyst reports, the ways of security analysts paying attention to cause informations concerning business performances were learned, and newspaper articles including similar causal relationships were extracted. We aim to realize a real-time investment supporting system., The Japanese Society for Artificial Intelligence, Japanese
• Domain-specific dictionary construction method considering synonym and antonym
  ITO Ryo; HIROKI Sakaji; IZUMI Kiyoshi; SUDA Shintaro, JSAI Technical Report, Type 2 SIG, 2017, FIN-019, 78, 14 Oct. 2017
  In recent years, textual information, which is unstructured data attracts attention as new analytical data in the financial and economic fields and it is expected to structure knowledge on this domain. One such knowledge is a sentiment polarity dictionary in which each word is representing positive or negative. In building the dictionary, it is costly to add the polarity value to a vast number of words manually. Therefore, in this research, we propose a the dictionary construction model especially considering the synonymity and symmetry of words. As a result of the experiment, the proposed method is a more accurate than the model of the previous research. In addition, we extended the conventional dictionary using the proposed method, and we showed that the extended dictionary has higher accuracy than the dictionary which is not extended., The Japanese Society for Artificial Intelligence, Japanese
• Determination of unexpectedness of cause-result expressions extracted from summary of financial statements
  室野 莉沙; 酒井 浩之; 坂地 泰紀; ベネット ジェイスン, 電子情報通信学会技術研究報告 = IEICE technical report : 信学技報, 117, 207, 93, 98, 07 Sep. 2017
  電子情報通信学会, Japanese
• 地方議会の議案収集に向けた議案一覧抽出の試み
  田中琢真; 坂地泰紀; 小林暁雄; 木村泰知; 増山繁, 電子情報通信学会技術研究報告, 117, 207, 41, 46, Sep. 2017
• 議論の背景・過程・結果を関連づける地方政治コーパスの構築の試み
  木村 泰知; 小林 暁雄; 坂地 泰紀; 内田 ゆず; 高丸 圭一; 乙武 北斗; 吉田 光男; 荒木 健治, 第31回人工知能学会全国大会, May 2017
  一般社団法人 人工知能学会, Japanese, Summary national conference
• Estimation of Starting Pages for Each Agenda Item Stated in Notice of Annual Meeting of Shareholders
  TAKANO Kaito; SAKAI Hiroyuki; SAKAJI Hiroki; IZUMI Kiyoshi; OKADA Nana; MIZUUCHI Toshikazu, JSAI Technical Report, Type 2 SIG, 2017, FIN-018, 10, 10 Mar. 2017
  In this research, we aim to predict start pages of proposals stated in notice of the meeting of shareholders and classify which proposal the page is. We propose two methods that classification method of proposals. The first method heuristically predicts the page on which the proposal is described. Moreover our method extracts specialized terms of each proposal and assigns weights to them. After that, our method classifies proposals by specialized terms. The second method classifies proposals using deep learning. Each methods were evaluated, and the effectiveness of each methods was verified., The Japanese Society for Artificial Intelligence, Japanese
• 都道府県議会会議録からの意見や意志を表す発言の抽出
  坂地泰紀; 酒井浩之; 小林暁雄; 内田ゆず; 乙武北斗; 高丸圭一; 木村泰知, 言語処理学会第23回年次大会(NLP2017), 426, 429, Mar. 2017
• 地方政治コーパス構築における従来の成果と現在の課題 - 政治・経済分野の応用研究に向けたパネルデータの構築-
  木村泰知; 小林暁雄; 坂地泰紀; 内田ゆず; 高丸圭一; 乙武北斗; 吉田光男; 川浦昭彦, 言語処理学会第23回年次大会, 54, 57, Mar. 2017
  Japanese, Summary national conference
• Extraction of Laboratory Web Pages from University Web Sites
  宮崎 敦也; 酒井 浩之; 坂地 泰紀, 成蹊大学理工学研究報告, 53, 2, 25, 28, Dec. 2016
  In this paper, we propose a method that extracts laboratory front pages from university web sites. Our method extracts the laboratory front pages by using SVM and applying some rules. Moreover, we developed the laboratory search system which is able to retrieve laboratory front pages extracted by our method. We evaluated our method and it attained 85.0% precision and 65.5% recall, respectively, 成蹊大学理工学部, Japanese
• 都道府県議会会議録を用いた地方政治コーパスの構築の試み
  田中琢真; 小林暁雄; 坂地泰紀; 内田ゆず; 乙武北斗; 高丸圭一; 木村泰知, 情報処理北海道シンポジウム 2016, 131, 132, Oct. 2016
  Japanese, Summary national conference
• Extraction of New Relevant Companies by Using Extracted Common Elements from Relevant Companies Sets
  田中瑞竜; 酒井浩之; 坂地泰紀, 電子情報通信学会技術研究報告, 116, 213(NLC2016 13-28), 19‐24, 01 Sep. 2016
  Japanese
• Extraction of Characteristic Comments from Nico Nico Douga
  吉岡晋作; 坂地泰紀; 酒井浩之, 電子情報通信学会技術研究報告, 115, 445(NLC2015 44-53), 1‐5, 28 Jan. 2016
  Japanese
• 製品名と特許における名称の言い換え知識の抽出
  坂地泰紀; 所佳祐; 酒井浩之, 人工知能学会全国大会論文集(CD-ROM), 30th, ROMBUNNO.1H2‐4, 2016
  Japanese
• 決算短信PDFからの業績予測文の抽出
  北森詩織; 酒井浩之; 坂地泰紀, 言語処理学会年次大会発表論文集(Web), 22nd, C3‐5 (WEB ONLY), 2016
  Japanese
• ニコニコ動画に投稿された動画の新しいタグの獲得
  坂地泰紀; 小花聖輝; 小林暁雄; 酒井浩之, ファジィシステムシンポジウム講演論文集(CD-ROM), 32nd, ROMBUNNO.WD3‐3, 2016
  Japanese
• Extraction Method of Parliament Members' Statements for Construct a Local Political Corpus from Japanese Prefecture Governments Conference Minutes
  Tanaka Takuma; Kobayashi Akio; Sakaji Hiroki; Uchida Yuzu; Ototake Hokuto; Takamaru Keiichi; Kimura Yasutomo, Proceedings of the Fuzzy System Symposium, 32, 0, 251, 254, 2016
  

  , Japan Society for Fuzzy Theory and Intelligent Informatics, Japanese
• Extraction of Laboratory Front Pages from University Web Sites
  宮崎敦也; 酒井浩之; 坂地泰紀, 電子情報通信学会技術研究報告, 115, 222(NLC2015 17-33), 37, 41, 03 Sep. 2015
  Japanese
• Extraction of Positive and Negative Comments from Nico Nico Douga
  坂地泰紀; 酒井浩之, 電子情報通信学会技術研究報告, 115, 222(NLC2015 17-33), 31, 35, 03 Sep. 2015
  Japanese
• Classification of Comments on Nico Nico Douga for Annotation Based on Referred Contents
  Akihito Ikeda; Akio Kobayashi; Hiroki Sakaji; Shigeru Masuyama, PROCEEDINGS 2015 18TH INTERNATIONAL CONFERENCE ON NETWORK-BASED INFORMATION SYSTEMS (NBIS 2015), 673, 678, 2015, [Peer-reviewed]
  English
• An Extraction Method of Causal Knowledge from Newspaper Corpus
  坂地 泰紀; 酒井 浩之; 増山 繁, 成蹊大学理工学研究報告, 51, 2, 23, 27, Dec. 2014
  This paper proposes a method that extracts causal knowledge from news paper articles via clue expressions. Our method decides whether a sentence includes causal knowledge or not when the method extracts it. Therefore, our method can extract causal knowledge accurately. Furthermore, the advantage of our decision method is to extract causal knowledge manually without dictionaries and patterns., 成蹊大学理工学部, Japanese
• Extraction of Important Articles that Influence the Stock Price of Companies from Financial Articles
  中山 大; 坂地 泰紀; 勝田 研一郎, 成蹊大学理工学研究報告, 51, 2, 53, 60, Dec. 2014
  This paper proposes a method of extracting important articles that influence the stock price of companies from financial articles. In the first step, our method obtains dates that have large difference from previous day at stock price of a selected company. In the second step, our method acquires articles that are published around the obtained dates and concern the selected company from financial articles. In the third step, our method extracts articles that influence the stock price of the selected company by using SVM as a machine learning method from the acquired articles. Finally, we evaluated our method. As a result, our method achieved 69.8% precision and 53.1% recall., 成蹊大学理工学部, Japanese
• Extraction of Positive Comments from Nico Nico Douga
  ISHIBUCHI Junya; SAKAJI Hiroki; SAKAI Hiroyuki, IEICE technical report. Natural language understanding and models of communication, 114, 211, 17, 21, 11 Sep. 2014
  In this paper, we propose a method that extracts positive comments (e.g. "もっと評価されるべき"(motto hyouka sarerubeki: should be valued) from Nico Nico Douga automatically. For example, positive comments extracted by our method are beneficial for analysis of campaign broadcasts., The Institute of Electronics, Information and Communication Engineers, Japanese
• Estimation of Tags Relevant to a Search Query for the Company Search System
  SAKAI Hiroyuki; SAKAJI Hiroki, IEICE technical report. Natural language understanding and models of communication, 114, 211, 41, 45, 11 Sep. 2014
  In this paper, we propose a method that annotates tags corresponding to search queries in our company search system. These tags are estimated by our method automatically. For example, if a search query is "textmining", our method estimates "development", "offer" and "introduction" as tags. Then, our company search system is able to retrieve companies that develops "textminig" technology by using the estimated tags., The Institute of Electronics, Information and Communication Engineers, Japanese
• Classification of comments on Nico Nico Douga for annotation based on referred contents
  IKEDA Akihito; KOBAYASHI Akio; SAKAJI Hiroki; MASUYAMA Shigeru, IEICE technical report. Natural language understanding and models of communication, 114, 211, 47, 52, 11 Sep. 2014
  For Nico Nico Douga, a video sharing website in Japan, one of the largest feature is posted comments. The comments are one of the feature of posted videos. The comment data may apply to online advertisement, and video retrieval methods that are different from conventional ones etc. However, if we perform processing of all comments, there are difficulties in operation of the huge processing. For this reason, in order to apply to them, pre-annotation to the comment data is needed. In this study, we perform annotation of referring contents in Nico Nico Douga's comments., The Institute of Electronics, Information and Communication Engineers, Japanese
• Automatic Associating the Nikkei Press-Releases with Nico Nico Douga
  SHIBAHARA Keita; SAKAJI Hiroki; SAKAI Hiroyuki, IEICE technical report. Natural language understanding and models of communication, 114, 211, 75, 79, 11 Sep. 2014
  In recent years, animation sites, such as animation share services and animation distribution services, have spread by development of the Internet. Our focus is to associate an animation and the Nikkei Press-Releases via animation's tags and press-release's words., The Institute of Electronics, Information and Communication Engineers, Japanese
• Extraction of Character's Impression Expressions from Twitter
  Tahara Yukina; Sakaji Horiki; Sakai Hiroyuki, IEICE technical report. Natural language understanding and models of communication, 113, 429, 5, 10, 06 Feb. 2014
  In this paper, we introduce a method of extracting character's impressions from Twitter. We focus Yuru-chara that is one of the mascot characters and extract impression expressions (e. g., "Pretty". "Heal") corresponding to Yuru-chara's names. We make an impression expressions dictionary which consists of 363 impression expressions classified into "delighted", "angry", "lament", "frightened", "embarassed", "like", "annoyed", "excited", "amazed", "reliable""entertain""other". Our method acquires Tweets containing the impression expressions and the Yuru-chara's names and extracts impression expressions each Yuru-chara from the Tweets. Finally, we evaluated our method., The Institute of Electronics, Information and Communication Engineers, Japanese
• Extraction of Character's Impression Expressions from Twitter
  Tahara Yukina; Sakaji Horiki; Sakai Hiroyuki, IEICE technical report. Natural language understanding and models of communication, 113, 429, 5, 10, 30 Jan. 2014
  In this paper, we introduce a method of extracting character's impressions from Twitter. We focus Yuru-chara that is one of the mascot characters and extract impression expressions (e. g., "Pretty". "Heal") corresponding to Yuru-chara's names. We make an impression expressions dictionary which consists of 363 impression expressions classified into "delighted", "angry", "lament", "frightened", "embarassed", "like", "annoyed", "excited", "amazed", "reliable""entertain""other". Our method acquires Tweets containing the impression expressions and the Yuru-chara's names and extracts impression expressions each Yuru-chara from the Tweets. Finally, we evaluated our method., The Institute of Electronics, Information and Communication Engineers, Japanese
• A Causal Expressions Search System for PDF Files of Summary of Financial Statements
  坂地 泰紀; 酒井 浩之; 増山 繁, 人工知能学会全国大会論文集, 28, 1, 4, 2014
  人工知能学会, Japanese
• An Extraction Method of Causal Knowledge from Newspaper Corpus
  SAKAJI Hiroki; MASUYAMA Shigeru, IEICE technical report. Natural language understanding and models of communication, 111, 119, 7, 10, 30 Jun. 2011
  This paper proposes a method that extracts causal knowledge from news paper articles via clue expressions. Our method decides whether a sentence includes causal knowledge or not when the method extracts it. Therefore, our method can extract causal knowledge accurately. Furthermore, the advantage of our decision method is to extract causal knowledge manually without dictionaries and patterns., The Institute of Electronics, Information and Communication Engineers, Japanese
• Determining Causal Relation at Sentences in Newspaper Articles
  坂地 泰紀; 増山 繁; 酒井 浩之, 情報処理学会研究報告, 2010, 2, 4p, Aug. 2010
  情報処理学会, Japanese
• Determining Causal Relation at Sentences in Newspaper Articles
  SAKAJI Hiroki; MASUYAMA Shigeru; SAKAI Hiroyuki, IEICE technical report. Natural language understanding and models of communication, 110, 142, 47, 50, 15 Jul. 2010
  We propose a method for determining causal relation in newspaper sentences with clue expressions. Clue expressions indicating causal relations sometimes have mean other than causal relation. For example, clue expression "から"(kara:from) means causal relation or other mean (e.g. start points). Therefore, we need to exclude clue expressins that mean other than causal relation for extracting causal knowledge correctly. Our method uses expanded linguistic ontology as semantic feature and extracts additional learning data automatically. As a result, performance of our method is improved., The Institute of Electronics, Information and Communication Engineers, Japanese
• 商品ページからの属性・属性値抽出と同一商品クラスタリング手法
  坂地泰紀, 言語処理学会第16回年次大会, 2010, 371, 374, 2010
• Extracting Effect-Technology Terms and Semantic Processing for Generation of an Effect-Technology Type Patent Map
  野中 尋史; 小林 暁雄; 坂地 泰紀, 人工知能学会全国大会論文集, 24, 1, 4, 2010
  人工知能学会, Japanese
• Extracting solution-effect expressions from patent documents via a bootstrapping method
  坂地 泰紀; 野中 尋史; 酒井 浩之, IEICE technical report, 109, 142, 85, 92, 22 Jul. 2009
  電子情報通信学会, Japanese
• Extracting Solution-Effect Expressions from Patent Documents via a Bootstrapping Method
  Hiroki Sakaji; Hirofumi Nonaka; Hiroyuki Sakai; Shigeru Masuyama, IPSJ SIG Notes, 2009, 14, 1, 8, 15 Jul. 2009
  We propose a method Cross-Bootstrapping for extracting solution-effect expressions from patent documents automatically. Solution-effect expressions are useful for generating patent maps. Our method extracts expressions using two clues and statistical information via a bootstrapping method. Furthermore, the advantage of our method is to extract expressions without dictionaries and patterns given manually. Finally, we experimented our method. As a result, our method achieved sufficient performance for generating patent maps., Information Processing Society of Japan (IPSJ), Japanese
• 2-D-4 景気動向を示す根拠表現の抽出と分類(統計)
  坂地 泰紀; 酒井 浩之; 増山 繁, 日本オペレーションズ・リサーチ学会秋季研究発表会アブストラクト集, 2007, 194, 195, 27 Sep. 2007
  社団法人日本オペレーションズ・リサーチ学会, Japanese
• Extraction and Analysis of Basis Expressions that Indicate Economic Trends
  SAKAJI Hiroki; SAKAI Hiroyuki; MASUYAMA Shigeru, IPSJ SIG Notes, 2007, 76, 151, 156, 24 Jul. 2007
  In this research, we propose a method to automatically extract basis expressions that indicate economic trends from newspaper articles by using a statistical method. We also propose a method to classify them into positive expressions that indicate upbeat, and negative expressions that indicate downturn in economy. It is important to foresee the economic trends for companies, governments and investors to forecast stock places and sales of goods. Therefore, we considered that the companies, governments and investors are able to forecast economic trends by using basis expressions extracted from newspaper articles concerning economic trends. In this time, we extracted basis expressions, and classified them into positive expressions or negative expressions as information to forecast economic trends. We evaluated our methods using NIKKEI newspaper articles from 1990 to 2005. The results showed that the method to extract basis expressions, attained presicion 71.43% and recall 33.33%. The classification method attained F-measure with positive expressions 0.695, and F-measure with negative expressions 0.849., Information Processing Society of Japan (IPSJ), Japanese
• Extraction and Analysis of Basis Expressions that Indicate Economic Trends
  SAKAJI Hiroki; SAKAI Hiroyuki; MASUYAMA Shigeru, IEICE technical report. Natural language understanding and models of communication, 107, 158, 151, 156, 17 Jul. 2007
  In this research, we propose a method to automatically extract basis expressions that indicate economic trends from newspaper articles by using a statistical method. We also propose a method to classify them into positive expressions that indicate upbeat, and negative expressions that indicate downturn in economy. It is important to foresee the economic trends for companies, governments and investors to forecast stock places and sales of goods. Therefore, we considered that the companies, governments and investors are able to forecast economic trends by using basis expressions extracted from newspaper articles concerning economic trends. In this time, we extracted basis expressions, and classified them into positive expressions or negative expressions as information to forecast economic trends. We evaluated our methods using NIKKEI newspaper articles from 1990 to 2005. The results showed that the method to extract basis expressions, attained presicion 71.43% and recall 33.33%. The classification method attained F-measure with positive expressions 0.695, and F-measure with negative expressions 0.849., The Institute of Electronics, Information and Communication Engineers, Japanese

• Pythonによる金融テキストマイニング
  和泉, 潔; 坂地, 泰紀; 松島, 裕康
  朝倉書店, Jul. 2022, 9784254275889, vi, 174p, Japanese
• Pythonではじめるテキストアナリティクス入門
  榊, 剛史; 石野, 亜耶; 小早川, 健; 坂地, 泰紀; 嶋田, 和孝; 吉田, 光男
  講談社, Mar. 2022, 9784065274101, 253p, Japanese
• 金融・経済分析のためのテキストマイニング
  和泉, 潔; 坂地, 泰紀; 松島, 裕康
  岩波書店, Jan. 2021, 9784000299015, viii, 163p, Japanese

• 生成AIへの言語モデルの進化
  金融庁金融経済学勉強会, Oct. 2024, Public discourse
  [Invited]
• 言語モデルを用いた経済ナラティブインデックスの生成
  坂地 泰紀
  進化経済学会関西大会オータムカンファレンス, Sep. 2024, Public discourse
  [Invited]
• テキストマイニングを用いた地方景況感の分析
  坂地泰紀
  第54回サイバーワールド(CW)研究会, Aug. 2023
  Aug. 2023, [Invited]
• 生成AIがもたらす金融業界での利用の可能性
  坂地 泰紀
  アイフィスオンラインセミナー, Aug. 2023, Public discourse
  [Invited]
• 経済因果チェーンの構築とその応用
  坂地泰紀
  Workshop on Microeconomic Analysis of Social Systems and Institutions: Theory, Experiment, and Empirical Studies, Mar. 2023
  Mar. 2023, [Invited]
• BERTと因果抽出を用いた気候変動ナラティブの可視化／指数化
  坂地泰紀; 金田規靖
  日本銀行金融研究所ファイナンス・ワークショップ, Nov. 2022
  Nov. 2022, [Invited]
• Economic Causal-Chain Search using Text Mining Technology
  Hiroki Sakaji
  Knowledge Graphs in Finance and Economics, AKBC Conference 2022 Workshop, Nov. 2022
  Nov. 2022, [Invited]
• 決算短信定性的情報の活用方法・分析ポイントについて～決算短信への期待～
  坂地泰紀
  決算短信からのデータ抽出に関するハンズオン「JPX総研(日本取引所グループ)/AWS共催」, Oct. 2022
  Oct. 2022, [Invited]
• 地域経済の未来をデータで読み解く！
  坂地泰紀; 砂川恵太; 上島邦彦
  日本データ取引所主催シンポジウム「データ流通市場の歩き方」, Jan. 2022, Public discourse
  [Invited]
• 金融テキストマイニングの最新技術とその応用
  坂地泰紀
  日本経済研究センター主催「AI・ビッグデータ経済モデル研究会」, Jan. 2021, Public discourse
  [Invited]
• テキストを用いた経済・経営分析のための人工知能技術
  坂地泰紀
  長岡市主催「ビジネスのためのAI・IT・IoT活用セミナー」, Dec. 2020, Public discourse
  [Invited]
• 因果チェーンとネットワーク学習によるソースごとのCOVID19に関する捉え方の違いの抽出
  坂地泰紀; 久野遼平
  東京大学金融教育研究センター・日本銀行調査統計局共催 ビッグデータフォーラム, Nov. 2020, Public discourse
  [Invited]
• 多様なテキストデータを対象にした金融テキストマイニング
  坂地泰紀
  第293回MPTフォーラム, May 2019, Public discourse
  [Invited]
• 言語処理技術を用いた金融テキストマイニング
  坂地泰紀
  北海道大学講演会(電子情報通信学会・日本知能情報ファジィ学会), Jul. 2018, Public discourse
  [Invited]

• 知能ソフトウェア特論, 2024年, 修士課程, 情報科学院
• 情報理工学演習Ⅳ, 2024年, 学士課程, 工学部
• 情報理工学実験Ⅰ, 2024年, 学士課程, 工学部
• ソフトウェア工学, 2024年, 学士課程, 工学部
• データサイエンス, 2024年, 学士課程, 工学部
• 計算機プログラミングⅠ, 2024年, 学士課程, 工学部
• 計算機プログラミング演習, 2024年, 学士課程, 工学部

• IEEE
• THE JAPANESE SOCIETY FOR ARTIFICIAL INTELLIGENCE
• THE ASSOCIATION FOR NATURAL LANGUAGE PROCESSING
• THE INSTITUTE OF ELECTRONICS, INFORMATION AND COMMUNICATION ENGINEERS
• INFORMATION PROCESSING SOCIETY OF JAPAN
• ACL

• Causal Extraction
  Hiroki Sakaji, [Software]

• AI農業社会実装プロジェクト
  農林水産省における研究開発とSociety5.0 との橋渡しプログラム（BRIDGE）
  Sep. 2023 - Mar. 2026
  農林水産省, Coinvestigator, 23836805
• Economic Narrative Simulation Using Causal Information
  PRESTO
  Oct. 2022 - Mar. 2026
  坂地 泰紀
  JST, Principal investigator, 22717675
• 経済シナリオ分析のための因果関係インスタンス認識技術の確立
  科学研究費助成事業 基盤研究(C)
  01 Apr. 2021 - 31 Mar. 2024
  坂地 泰紀
  本年度は、まず、因果関係インスタンスを取得するために、決算短信に対してタグ付与を行った。その後、タグ付けを行った決算短信データ、タグ付与済みの英語ロイターニュース記事、FinCausalのデータセットを用いて実験を行い、日英の文書から因果関係インスタンスを抽出できる手法の開発に取り組んだ。結果的に、BERTとグラフニューラルネットワークを組み合わせることで既存の手法よりも高い精度で因果関係インスタンスを抽出できる手法の開発に成功した。具体的には、全てのデータセットにおいて、F値が0.75以上で因果関係インスタンスを抽出することができた。この結果を論文としてまとめて国際会議に投稿したが、残念ながら不採録となった。そのため、2022年度は論文のブラッシュアップを行い、再度、投稿を行う。
  因果関係インスタンスを抽出するためにドメイン特化のBERTモデルの構築も行った。モデル
  構築にあたり、グラフィックカードであるNvidiaのV100を購入予定であったが、V100よりも価格が安いうえに性能が高いNvidiaのA6000が発売されていたことから、こちらを2個購入し、モデル構築や実験に利用した。
  作成した事前学習モデルは、Web上で公開しており、誰でも無料で利用可能となっている。全ての公開したモデルのダウンロード数を合わせると、現時点で6,700件以上あり、多くの方に利用して頂いている。こちらの研究に関しては、SIGFINなどの国内研究会で発表済みである。こちらの研究に関しては、SIGFINなどの国内研究会で発表済みである。
  日本学術振興会, 基盤研究(C), 東京大学, 21K12010
• System Design and Dynamics of Data Exchange Market
  Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B)
  01 Apr. 2020 - 31 Mar. 2024
  早矢仕 晃章; 清水 たくみ; 松島 裕康; 深見 嘉明; 坂地 泰紀
  本研究の目的は、データ流通市場の健全な発展を目指し、マーケットにおける異種のデータ・法規制・ヒトの動的な相互作用の解明とそれに基づくデータ流通の制度設計である。
  2年目となる本年度は、昨年度整理したデータ流通市場の構成要素とその関係性を検討し、市場シミュレータ開発のためのモデル構築に取り組んだ。また、IEEE International Conference on Big Data 2021にて国際ワークショップCross-disciplinary Data Exchange and Collaborationを企画・運営し、国際的なコミュニティ作りにも注力した。
  はじめに、(1)行政情報システム研究所で公開しているメタデータサイト及び、(2)データ分析コンペティションサービスKaggleの2つのデータ交換プラットフォームを分析した。(1)では組み合わせによって価値を持つデータのモデル化の手掛かりを得た。(2)では、COVID-19関連データを対象とし、データ公開初期のユーザ行動がコミュニティ形成とデータ利活用を促進することが分かった。
  続いて、Kaggleデータセットを昨年度開発したデータ類似性計算手法に適用した。その結果、プラットフォームに依らない構造的特徴の一部が明らかとなった。
  これらの解析から得られたデータとユーザの特徴から複数のモデルを作成し、データ生成器とデータ市場シミュレータを試作した。これにより、観測可能な相互作用が限定されるデータ流通市場の理解に向けた次年度の進展が期待できる。
  また、近年注目されるIDアーキテクチャの分散化がデータ流通にもたらす影響と効率性を評価するために、分散型と集中型IDアーキテクチャを分析した。今後は、データ流通のインセンティブや導入コストの観点から、マルチエージェントシミュレーションに向けたパラメータを整理と検証を進めていく。
  Japan Society for the Promotion of Science, Grant-in-Aid for Scientific Research (B), The University of Tokyo, 20H02384
• データ取引の市場設計：被験者実験による研究
  Apr. 2023 - Mar. 2024
  関西大学ソシオネットワーク戦略研究機構, Coinvestigator
• データ流通市場の設計：新しい情報技術の人工市場における実験研究
  Apr. 2021 - Mar. 2023
  渡邊 直樹; 難波 敏彦; 早矢仕 晃章; 坂地 泰紀; 松島 裕康; 清水 たくみ; 深見 嘉明
  関西大学ソシオネットワーク戦略研究機構, Coinvestigator
• Construction of A Local Political Corpus that Links the Background, Process, and Results of Discussions and Its Interdisciplinary Application
  Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B)
  01 Apr. 2016 - 31 Mar. 2020
  Kimura Yasutomo
  In this research, a corpus that correlates "background of discussion", "process of discussion" and "result of discussion" is constructed with the aim of activating and interdisciplinary research on local politics. Specifically, we aim to link three language resources, newspaper articles (background of discussions), minutes of local assembly (process of discussions), and ordinances (results of discussions) from the viewpoint of region, time, and issues. The purpose of this research is the following three points. A) Collection and organization of three language resources (local assembly minutes, ordinances, newspaper articles) B) Build a corpus that links three language resources from the perspective of region, time, and issues C) Applied research in politics, economics, linguistics, and information engineering using the above results
  Japan Society for the Promotion of Science, Grant-in-Aid for Scientific Research (B), Otaru University of Commerce, 16H02912
• Company search system for employment support by using financial statements and web information
  Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (C)
  01 Apr. 2015 - 31 Mar. 2019
  Sakai Hiroyuki
  In this research, we aimed to automatically extract the information related to the business of companies from financial statements and web sites of the companies, and to develop the company search systems by using the extracted information. As research results, we developed a method to extract causal information, a method to extract causal correlation (cause and result), a method to extract sentences of forecasted business performance from financial statements, and a method to estimate related companies by using the causal information, respectively. We also developed company search systems by using causal information extracted by our methods. The developed company search systems are published on the website (http://www.ci.seikei.ac.jp/sakai/).
  Japan Society for the Promotion of Science, Grant-in-Aid for Scientific Research (C), Seikei University, 15K00315

• 情報処理装置、情報処理方法、およびプログラム
  Patent right, 田中良典; 佐久間洋明; 佐藤史仁; 小寺俊哉; 和泉潔; 坂地泰紀
  特願2019-1952
  特開2020-112931
  特許6615392
• 抽出システムおよびプログラム
  Patent right, 平野正徳; 坂地泰紀; 松島裕康; 和泉潔; 加藤惇雄; 森岡嗣人; 長尾慎太郎; 木村笙子
  特願2018-244861
  特開2020-107051
  特許6596565