Computing and Explaining Knowledge Relevance in Exploratory Search Environment

The main goal of the exploratory search is to find the knowledge that can meet user's information needs and solve a problem. The searching is also a knowledge acquiring and learning process and user's requirements often evolve with the process. It will be helpful to explain why the information found is relevant to users' requirements during searching. The relevance between a query and a resource not only depend on representation of resource but also is affected by user's cognitive state, base on this assumption, a knowledge relevance model and knowledge space are built to represent resources semantics and user's cognitive state by combining description logics and conceptual graph in this work. By analyzing the structure of knowledge space, this work defines the sequential relation on the concepts space that generated by refinement operator and concepts neighbors according to topology structure of conceptual graph, which provide key information for data relevance computing. Probability based on model theoretic semantics of description logics is adopted to show the quantity relationships between concepts. The knowledge relevance is determined from both structure perspective and statistical perspective of knowledge space. Based on the knowledge space structure. Interactive knowledge visualization techniques are adopted to explain the relevance between key knowledge concepts. A dataset collected from WebMD is collected and a knowledge visualization prototype is developed, which prove the rationality and interpretability of the knowledge relevance computing method proposed.