Research on unified formal semantics based on the analysis of textual entailment corpora

2014 Fiscal Year Final Research Report

• Project/Area Number: 24652089
• Research Category: Grant-in-Aid for Challenging Exploratory Research
• Allocation Type: Multi-year Fund
• Research Field: Linguistics
• Research Institution: National Institute of Informatics
• Principal Investigator: MIYAO Yusuke 国立情報学研究所, コンテンツ科学研究系, 准教授 (00343096)
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: 形式意味論 / 含意関係認識

Outline of Final Research Achievements

This research aimed to study a unified framework for a formal representation of natural language semantics. Natural language semantics involves logical relations, semantic relations among words and phrases, tense and aspect, and discourse relations, while they have complicated interactions with each other. However, linguistic studies mainly focused on a specific issue in each research field, and no framework for representing these features in a unified way. Therefore, we developed a unified formal representation of natural language semantics based on the analysis of textual entailment corpora. In addition, we developed a system for textual entailment recognition, and demonstrated that our framework allows for fast and high accuracy semantic inference.

Free Research Field

自然言語処理