The Research On Concept Retrieval Model Based On Rhombus-Thinking Process

With the rapid explosion of information, retrieval engine is desired to have higher Precision and Recall for better user experience. Intelligence becomes the key of retrieval systems. Concept retrieval which expands key words based on Ontology is a research focus. The existing concept retrieval systems expand key words through Synonymy and Hyponymy, ignore other semantic relations in Ontology, and lack the function of semantic disambiguation. It is meaningful for retrieval system to resolve synonym and polyseme without semantic annotation at present. At the same time, users'developing demands can't be neglected. It is necessary to guide users'demand availably.In order to provide better information service and achieve the function of inspiring user demand, the concept retrieval model based on rhombus-thinking process is proposed. On the basis of rhombus-thinking process which reflects the habits of human mind, query expansion is performed in accordance with the expansion way of matter element. Expansion options are chosen as query options according to semantic relatedness and the designed range. The semantic retrieval and keyword retrieval are comprehensive utilized to document ranking. Users can have better understanding about relevant knowledge through returned expansion recommended options.The research suggests expanding ontology element based on matter theory. Inference rules are summarized for acquiring expansion options. The semantic relatedness evaluation models are established according to the difference of objects. A simple experiment is given to verify the validity and feasibility of the whole model.The implement of such a retrieval model can not only enhance the semantic cognitive ability of system, but also can return expansions reflecting users'potential demands and helping deep recognition about domain knowledge. The proposed semantic relatedness evaluation models are meaningful for the optimization of such research.