Research On Chinese Sentimental Question Answering In The Restricted Domain

With the explosion of information, the vast available information makes it an attractive resource for answering a variety of questions that users may have. Traditional search engines return a list of relevant documents but not the exact answers, and users often engage in perusing relevant information that they need manually. To overcome these shortcomings, more and more research organizations and companies do efforts to explore new generation of information retrieval systems. One of the most important directions is Question Answering (QA) system.Presently, many scientific research institutes have been investigated factoid questions frequently-asked questions (FAQ), some mature question-answering systems have been widely recognized. However, relatively little study has been done on opinion or sentimental question answering. It cannot be neglected that people would also like to know about other people's emotion, thought, opinions towards particular topics or celebrities. We find that people like to know focal figures' preferences and other people's opinions about them. In this paper, the research of sentimental QA was concentrated on entertainment focal figures' preferences and the public's feelings or opinions about them.Sentimental question answering has to deal with the opinion holders, polarity and other factors related to sentiment analysis compared to factual question answering. A great number of sentimental questions have been collected from internet, and the model of Chunk-CRF and heuristic rules are applied to classify them into five types based on the emotional features of questions. The classification method is different from the factual question classification which is mainly based on interrogative words, and it has to take opinions and responses from users into account. Combined with polarity and recognition results of Chunk-CRF, different answers are extracted according to different question types. The experiment shows that the sentimental QA system is effective and efficient.