Research And Design Of Constructing Knowledge Base In The General Knowledge Question Answering System

Knowledge base as a storehouse for storing knowledge in the question answering system plays an important role in the whole system. A knowledge base with self-contained functions and knowledge can consumedly improve efficiency and accuracy of a question answering system. So in this paper, the idea and function of constructing knowledge base is mainly studied. In this paper, knowledge base studied is general knowledge base. Because general knowledge contains a large number of fields and has poor stability, it is important for a knowledge base to collect, classify and maintain a plenty of knowledge. In order to overcome the weakness that a large-scale knowledge need be constructed in a conventional question answering system based knowledge, which would consume a flood of manpower and material resources, and answers of the question answering system based conventional information retrieval and information extraction have low accuracy and well develop the advantages that answers are exact and reasoning and computing can be done, in this paper the two methods above are combined to construct an efficient knowledge base. The reseach of knowledge base in this paper is as follows: 1. Question-based Object-oriented Knowledge Representation method is the main method used to construct knowledge base in this paper. It is based on the Object-oriented representation method. Every object is described by a text. The types of the questions often asked by users to the class or object are used as their attributes, and the content of the text that can answer the questions are used as the values of the attributes. In this paper, the questions asked by users are divided into ten types, and every type corresponds to an attribute. The names of these attributes would be marked in texts as markers. The question-based object-oriented knowledge representation method being used can avoid the complicated process that knowledge is formalized into knowledge points by manpower. 2. The texts stored into knowledge base are automatically classified by using the vector space model. 3. The original texts are automatically marked by making use of the feature dictionary to extract features of sentences. In this paper, knowledge is collected and classified and texts are marked mainly by computers automatically working , posteriorly by manpower. The method can reduce manpower and greatly improve accuracy. 4. The method that can obtain answers by making use of the knowledge base in the paper...