Based On The Concept Of Semantic Distance, Semantic Similarity

As a development of current Web, Semantic Web aims to enrich the Web data with semantic format and make them machine-readable. Thus it will be convenient for human to operate. But the implementation of Semantic Web faces some theoretic and technical difficulties, such as semantic interoperating, specification organizing and intelligent aggregating. To solve the problems, we designed a semantic native XML database system, and did research on some essential theories and technologies in detail. The structural levels of SNAX meet the need of the definition of Berners-Lee's Semantic Web, therefore many theories, technologies and methods applied in SNAX are also helpful to the implementing of Semantic Web.Firstly, the thesis introduced the architecture of Semantic Web proposed by Berners-Lee. After summarizing the theoretic and technical difficulties the Semantic Web faces, we proposed our research direction of constructing a miniature Semantic Web system SNAX. The system included Web data gathering, data storing, accessing, ontology mapping and multi-functional querying.Secondly, through analyzing the process of Web information gathering and ontology mapping, the paper discussed the important role of semantic similarity computing in SNAX. Following this, we proposed the methods for computing the semantic similarity based on SUMO. Through assigning different weights to different edges in hierarchy tree of SUMO, we put forward two different measures for semantic similarity according to the shortest path distance approach.Finally, the thesis evaluated our algorithm. We chose fifty pairs of concepts for comparing randomly. And through comparing the results gained from the machine with those collected from human's judgments, we established the appropriate values of the parameters. Therefore the method we proposed has been proved to be valid.