Interactive Question Answering Based on FAQ

Abstract

A question answering system receives the user’s question in nature language, and answers it in a concise and accurate way. An interactive question answering (IQA) provides a natural way for users to express their information requirement. There are two key points for IQA. The first is how to answer a user’s question in a continuous question answering process. The second is the way that the question answering system interacts with the user. In this work the answers are from FAQ knowledge base which is extracted from community question answering web portals. The syntactic, semantic and pragmatic features between question and candidate answers and context information are used to construct models by ranking learning method to extract the answers. And the question answering system requests user to feedback of the answer. It is a naive and effective interactive method. The results of experiments show that our method is effective for interactive question answering.KeywordscontextFAQInteractive question answeringranking learning