Building Paraphrase Typology through Comparative Linguistic Analysis

2015 Fiscal Year Final Research Report

• Project/Area Number: 25730139
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Intelligent informatics
• Research Institution: National Institute of Information and Communications Technology (2014-2015); Future University-Hakodate (2013)
• Principal Investigator: Fujita Atsushi 国立研究開発法人情報通信研究機構, ユニバーサルコミュニケーション研究所多言語翻訳研究室, 主任研究員 (10402801)
• Project Period (FY): 2013-04-01 – 2016-03-31
• Keywords: 言い換え / パラフレーズ / 類型化 / 自然言語処理

Outline of Final Research Achievements

We aimed to build a language-independent typology of paraphrases through collecting and analyzing a wide variety of paraphrase examples. We collected the examples in three ways: (i) extracting those illustrated in the books and articles in the related research fields, including linguistics, natural language processing, and language pedagogy, (ii) manually analyzing existing dataset created for benchmarking the systems for recognizing textual entailment, and (iii) incrementally pre-editing sentences so that they can be translated at a higher quality by a machine translation system. We also made an attempt to find a set of descriptive features toward explicit definition of each type of paraphrases. In addition to the features proposed in previous work, i.e., the surface-level scope of paraphrases and linguistic features associated with the paraphrases, the connotational differences that paraphrases incur and contextual elements that could be affected by paraphrasing, were examined.

Free Research Field

自然言語処理