A Semantic Feature for Relation Recognition Using a Web-based Corpus

Selecting appropriate features to represent an entity pair plays a key role in the task of relation recognition. However, existing syntactic features or lexical features cannot capture the interaction between two entities because of the dearth of annotated relational corpus specialized for relation recognition. In this paper, we propose a semantic feature, called the latent topic feature, which is topic-based and represents an entity pair at the semantic level instead of the word level. Moreover, to address the problem of insufficiently annotated corpora, we propose an algorithm for compiling a training corpus from the Web. Experiment results demonstrate that latent topic features are as effective as syntactic or lexical features. Moreover, the Web-based corpus can resolve the problems caused by insufficiently annotated relational corpora.