Evaluating Event Similarity Based on Causal Relationship

2021 Fiscal Year Final Research Report

• Project/Area Number: 19K20631
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 90020:Library and information science, humanistic and social informatics-related
• Research Institution: Takushoku University (2021); Tokyo Metropolitan University (2019-2020)
• Principal Investigator: Sumikawa Yasunobu 拓殖大学, 工学部, 助教 (70756303)
• Project Period (FY): 2019-04-01 – 2022-03-31
• Keywords: 歴史情報学 / 計算論的歴史学 / 情報検索 / 図書館情報学 / 因果関係 / テキストマイニング / 歴史学習

Outline of Final Research Achievements

We designed and implemented an algorithm to evaluate the similarity between events representing causal relationships, and presented a paper summarizing the results at the international conference WI-IAT. In order to use the results, we developed a dataset to create a learning environment in which past causal relationships can be used analogously to learn modern causal relationships, implemented clustering and classification algorithms, and published the results in the international journal IJDL and Applied Sciences. Finally, we implemented a chatbot on Twitter that uses the results of this research so that anyone can use the results. In addition, a paper discussing the academic value of this chatbot was published in the Transactions of the Institute of Electronics, Information and Communication Engineers (IEICE) as a recommended paper by the Technical Committee on Data Engineering Research, and received the DEIM2020 runner-up for the best paper award.

Free Research Field

計算機科学

Academic Significance and Societal Importance of the Research Achievements

歴史を学習することの意義は世界中で認められている。実際、多くの先進国では小学校から歴史を学ぶ授業が開講されている。しかし、「歴史は繰り返す」という言葉があるように、過去の知見を現代で活用できる場面が多く存在している。
本研究成果は、現代と過去の類似する因果関係を検索するアルゴリズムと、因果関係を検索した学習環境の実現に向けた基盤となるデータセットや環境を実現した。