中文文字蘊涵系統之特徵分析

Feature Analysis of Chinese Textual Entailment System

黃文奇(Wen-Chi Huang)

指導教授 ： 吳世弘

朝陽科技大學/資訊學院/資訊工程系/碩士(2011年)

摘要

文字蘊涵(Textual Entailment)的定義是判斷兩個句子能否互相推論。推論可分為五種類型：正向、反向、雙向、矛盾、獨立。這五種類型分別代表著不同的蘊涵關係。文字蘊涵辨識(Textual Entailment Recognition)是相當困難的自然語言處理問題。由於中文文字蘊涵的文獻較缺乏，本篇論文將中文文字蘊涵辨識提出了一個流程，提供給之後想要做這個題目的人的作為一個參考。中文的文字處理相較於英文的文字處理有許多不同的難處，在本篇論文中，我們將介紹處理中文的文字處理遇到的難處以及處理的流程。我們的系統使用支援向量機(Support vector machine, SVM)作為區分類型的演算法。使用的特徵分為兩個方向：1.文字特徵2.語意特徵。

關鍵字

支持向量機 ； 文字蘊涵 ； tree kernel

並列摘要

The goal of textural entailment recognition is to decide whether one sentence can deduce the other sentence. There are five types of entailment relation: forward, backward, bi-directional, contradict, and independent. Textual entailment recognition is a hard task in the field of natural language processing. Though there are many works dealing with the textual entailment recognition in English, there are very few references on textual entailment recognition in Chinese. In this paper, we proposed a process on Chinese textual entailment recognition. Our system is based on a well known classifier, support vector machine (SVM). We collect various features for this task, including surface features, syntactic features, and semantic features.

並列關鍵字

support vector machine ； textural entailment ； tree kernel

參考文獻

[8] Marta Tatu, Dan Moldovan, “A semantic approach to recognizing textual entailment”, In Proceedings of HLT/EMNLP 2005, pages 371–378, Vancouver, Canada, 2005

[9] Christiane Fellbaum, “WordNet: An Electronic Lexical Database”, The MIT Press, 1998.

[10] Dan I. Moldovan and Vasile Rus, “Logic form transformation of WordNet and its applicability to question answering”, In Proceedings of the 39th Annual Meeting of ACL, pages 402–409, Toulouse, France, 2001.

[12] Sebastian Padó and Mirella Lapata, “Dependency-based construction of semantic space models”, Computational Linguistics, Volume 33, No. 2, pages 161–199, 2007.

[13] Jeff Mitchell and Mirella Lapata, “Vector-based models of semantic composition”, In Proceedings of the 46th Annual Meeting of ACL: HLT, pages 236–244, Columbus, OH., 2008.