الفهرس 
ContentsOnly 14 pages are availabe for public view
 |  | from | 16 |  |  |
| | | from | 16 | | |
Abstract
Most of traditional Web content is suitable for human consumption not for machines and due to the huge amount of information on the web; it is necessary for the web to be machine understandable. Keyword-based search engines, such as AltaVista, Yahoo, and Google, are the main tools for searching for information on the Web. The main problemes of these search engines is the loss of semantics. Semantic Web is a system that allows the machines to understand and respond to requests for complex human-based meaning and give richer data processing which overcomes the main problems of Keyword-based search engines. This thesis describes both architecture and prototype of ArabSearch, a semantic module that helps user to get more relevant results when searching for information using a keyword-based search engine. The implementation of ArabSearch is divided into two stages: ontology development stage and application development stage. Implementation of ArabSearch focused on the modification of the query string itself rather than modifying the search engine. Concept mapping and semantic matching is used to extract keyword semantics, keyword’s synonyms and Arabic synonyms to be inserted in the query string. Finally, the modified query is forwarded to search engines. ArabSearch is implemented using Jena (a java frame work) with the help of ontology that developed for the education domain. ArabSearch uses the Arabic semantics of concepts to get documents that contain the required meaning but in Arabic. A comparative study is performed to compare the precision and recall of results of a keyword-based search engine such as Google with precision of results of a semantic search engine such as DuckDuckGo. Results showed improvement in precision and recall when using ArabSearch in both search engines. Results showed significant increase in precision when using the modified query compared with the original query.