(1149)A SEMANTIC WEB SEARCH METHOD FOR E-LEARNING AND ESCIENCE IN THE REALM OF ELECTRONIC AND COMPUTER TECHNOLOGIES

摘要

In this paper we present a method for semantic search of textual resources in the area of electronics and computer science, specialized for scientific research or education. Our method is knowledge and web statistics based. It relies on specialized domain or pedagogical knowledge, represented in computer - processable format as a set of interrelated ontologies. Our method use Word Sense Disambiguation (WSD)and Semantic Query Refinement (SQR)techniques to propose personalized and context-sensitive help to the user in search query formulation. We are working on it implementation as a basic method in semantic metasearch engine for improving the search quality (precision and recall)of scientific papers, tutorials and other educational or scientific textual resources. Our web search engine is implemented as a web search tool in the virtual laboratory for research and education in the realm of electronic and computer technologies. It includes semantic query refinement assistant, helping students and scientists in formulation of precise queries (or query disambiguation), web search engines selection tool and a tool for postprocessing and reordering of returned results. Reordering is based on snippet analysis and semantic annotation of returned documents, using domain ontologies. In this paper we also discuss usability of several recent web sources of knowledge (wikipedia, dictionaries, and other sources of definitions)for search query disambiguation tack and show it usefulness in case of technical and computer science terminology disambiguation. We also present, discuss and estimate preliminary results of an web statistics based approach for recognizing the words, that are monosemous in computer science domain (and therefore doesn’t need disambiguation when sending as a queries to domain-specific DLs), but are used in other domains, and should be disambiguated when sending as a query to multidomain libraries. We will estimate and compare the results of various strategies for query disambiguation or refinement strategies as well as changes in the quality of search results as a result of reordering.