Study on Chinese Open Domain Question Answering based on Support Vector Machine

For reason of providing precise answer to users which is better than traditional search engine, question answering (QA) has become a hot research spot. Currently, most researches focus on Knowledge base and community based question answering systems, studies on extracting answer from plain text are rarely mentioned. This paper studied open domain question answering based on support vector machine. A large scale synonym set was borrowed to improve system performance at every stage. Two kinds of new semantic features were introduced to extract answers from text. In order to verify effectiveness our methods, we implemented our experiment in the domain of Chinese and designed a prototype system named XiaoQ. Experimental data shows that our system works well at a promising level which proved that this paper has some reference value.