Recognizing entailment in Chinese texts with feature combination

In recent years, the natural language processing community has been manifesting increasing interest in textual entailment recognition among English texts. Yet, so far, not much attention has been paid to textual entailment recognition in Chinese texts. Recognizing entailment can be cast as a classification problem, and in this paper, a classification model based on support vector machine is constructed to detect semantic relations in Chinese text pair, including forward entailment, reverse entailment, bidirectional entailment, contradiction and independence for the multi-class task. We introduce different feature combinations based on four kinds of features, containing Chinese surface textual, Chinese lexical semantic, Chinese syntactic and Chinese linguistic phenomena features, to our classification model. The experimental results on NTCIR RITE-3 data collection show that the accuracy of our classification model using the feature combination with all the four kinds of Chinese textual features achieves a much better performance than all other systems on multi-class task.