Domain Knowledge For Knowledge Retrieval Model

The thesis is focusing on domain knowledge handling based on conceptual network, concentrating on conceptual network construction and calculation. On the base of Concept Theory, domain knowledge representation is analyzed and a model is proposed based on concept network. The knowledge represented conceptual network model should be classified into three levels, which are high, medium, and low levels. Domain knowledge is depicted in medium level, otherwise low level model is a supplement to the medium level model. Each level is emphasizing on different aspects in knowledge representation. Here the medium and low level concept networks are demonstrated in detail.Applying ontology as a guidance in medium conceptual network construction, a basic concept is regarded as an object which has complicated characteristics relating to attribute, behavior, description, wordset, etc. Then derived concept and compound concept are come from basic concept. In the concept model, a conceptual network is predigested into a concept tree. Concept node distance can be calculated according to level relations of nodes in concept tree. Thus calculation of distance between concept nodes is substituted by mathematic method. This method provides some theorical directions and reasonable measurements for medium level conceptual network construction.The low level conceptual network is formed by relations between low level concepts. Research has been done for text information set, which is the source of low level concept network construction. Centroid vector classification is used to acquire eigenset of concept and update the concept knowledge base. Thus the concept knowledge base is completed with the updating of the knowledge. The model of low level conceptual network is also provided as well.