Textual Entailment Recognition of Cross Lingual Passages

2017 Fiscal Year Final Research Report

• Project/Area Number: 15K12097
• Research Category: Grant-in-Aid for Challenging Exploratory Research
• Allocation Type: Multi-year Fund
• Research Field: Intelligent informatics
• Research Institution: Toyohashi University of Technology
• Principal Investigator: TSUCHIYA Masatoshi 豊橋技術科学大学, 情報メディア基盤センター, 准教授 (70378256)
• Project Period (FY): 2015-04-01 – 2018-03-31
• Keywords: 含意関係認識

Outline of Final Research Achievements

This research project aimed to resolve the cross lingual textual entailment recognition. First of all, this research proposed the method to build the large cross lingual textual entailment corpus using a parallel corpus which was designed for machine translation. The experimental construction of the corpus of English-Japanese sentence pairs using wikipedia English-Japanese parallel corpora showed that the proposed method to build the textual entailment corpus was effective to collect a large number of positive examples, but it was weak to collect negative examples. Especially, it was difficult to collect negative examples which contains logical contradictions. And more, the experimental result showed that the simple crows saucing method to collect negative examples cannot avoid the hidden bias.

Free Research Field

自然言語処理