Acquisition of Phrase Correspondences Using Natural Deduction Proofs

Proceedings of NAACL-HLT 2018, pages 756–766 New Orleans, Louisiana, June 1 - 6, 2018. c⃝2018 Association for Computational Linguistics Acquisition of Phrase Correspondences using Natural Deduction Proofs Hitomi Yanaka1 hitomiyanaka@g.ecc.u-tokyo.ac.jp Koji Mineshima2 mineshima.koji@ocha.ac.jp Pascual Mart´ınez-G´omez3 pascual.mg@aist.go.jp Daisuke Bekki2 bekki@is.ocha.ac.jp 1The University of Tokyo 2Ochanomizu University 3Artificial Intelligence Research Center, AIST Tokyo, Japan Abstract How to identify, extract, and use phrasal knowledge is a crucial problem for the task of Recognizing Textual Entailment (RTE). To solve this problem, we pro- pose a method for detecting paraphrases via natural deduction proofs of semantic relations between sentence pairs. Our so- lution relies on a graph reformulation of partial variable unifications and an algo- rithm that induces subgraph alignments between meaning representations. Experi- ments show that our method can automat- ically detect various paraphrases that are absent from existing paraphrase databases. In addition, the detection of paraphrases using proof information improves the ac- curacy of RTE tasks. 1 Introduction Recognizing Textual Entailment (RTE) is a chal- lenging natural language processing task that aims to judge whether one text fragment logically fol- lows from another text fragment (Dagan et al., 2013). Logic-based approaches have been suc- cessful in representing the meanings of complex sentences, ultimately having a positive impact on RTE (Bjerva et al., 2014; Beltagy et al., 2014; Mineshima et al., 2015, 2016; Abzianidze, 2015, 2016). Although logic-based approaches succeed in capturing the meanings of functional or logi- cal words, it is difficult to capture the meanings of content words or phrases using genuine logical inference alone. This remains a crucial problem in accounting for lexical relations between content words or phrases via logical inference. To solve this problem, previous logic-based approaches use knowledge databases such as WordNet (Miller, 1995) to identify lexical relations within a sen- tence pair. While this solution has been success- ful in handling word-level paraphrases, its exten- sion to phrase-level semantic relations is still an unsolved problem. There are three main difficul- ties that prevent an effective identification and use of phrasal linguistic knowledge. The first difficulty is the presence of out-of- context phrase relations in popular databases such as the Paraphrase Database (PPDB) (Ganitkevitch et al., 2013). PPDB may suggest paraphrases that do not adhere to the context of our relevant text segments nor to their semantic structure, which might be problematic. The second difficulty is finding semantic phrase correspondences between the relevant text seg- ments. Typical approaches only rely on sur- face (Beltagy et al., 2013) or syntactic correspon- dences (Arase and Tsujii, 2017), often producing inaccurate alignments that significantly impact our inference capabilities. Instead, a mechanism to compute semantic phrase correspondences could potentially produce, if available, more coherent phrase pairs and solve the recurring issue of dis- continuity. The third difficulty is the intrinsic lack of cov- erage of databases for logical inference despite their large size. Whereas there is a relatively small number of possible word-to-word corre- spondences and thus their semantic relations can be enumerated, the same is not true for all phrase pairs …

Content truncated to 500 words