Parsing Chinese Questions Based on Conceptual Network

We present a new approach to analyze Chinese questions based on a semantic knowledge base called conceptual network (ConNet). ConNet is composed of concepts, each of which describes the meaning of a Chinese word. Since there are no word boundaries in Chinese, we firstly segment each question into words, and then transform all the words into an ordered concept list. After that, an algorithm is proposed to combine the concepts into one semantic structure. Because the semantic structure carries plentiful semantic information, which is organized in a structured way, the parsing result can be well used in question answering systems