Research on the Question Answering System in Chinese Based on Knowledge Represent of Conceptual Graphs

Abstract

The core of question answering is a searching based on natural language understanding. Because of the flexibility and complexity of Chinese, the study of the Chinese question answering has become very difficulty. The current Chinese question answering relies on keywords to query and retrieve the answer. The nature of these systems is the keyword matching. It is a kind of simple natural language understanding. Its answering accuracy is low. In this paper, the research methods of the question answering in Chinese syntax and semantic analysis, which based on the knowledge representation of the conceptual graphs, is proposed. According to the methods, the question answering is formal defined, and the system structure is designed. The conceptual graphs of the interrogative sentences in Chinese are also constructed by common statement in Chinese interrogative sentences in this paper. After test, the conceptual graphs of the interrogative sentences reflect the basic situation of existing questions in Chinese, and the solution is feasible. The research methods are valuable for the question answering and natural language understanding.KeywordsConceptual graphsKnowledge representationQuestion answeringNatural language understanding