基於特徵為本及使用SVM的文本對蘊涵關係的自動推論方法

Textual Entailment Recognition Using Textual Features and SVM

這篇論文的目的是提出一個能判斷文本對蘊涵關係的系統。本系統主要使用7項由觀察資料所得之特徵作為輸入項，並以SVM作為預測模型。以NTCIR-10中RITE-2提供的資料評估本文所提方法，整體的Macro F1-measure為46.35%，較先前研究所提出的方法為佳。而該方法使用的特徵數量少、模型運算與實作簡單，可以在實際應用上有更好的效果。

The aim of this paper is to propose a system, which can automatically infer entailment relations of textual pairs. SVM is utilized as a prediction model of the system and seven features of textual pairs are employed to be input of the prediction model. The performance of this system is evaluated by dataset in CT-MC task held by RITE-2 of NTCIR. Macro-F1 of the proposed method is 46.35%.