![]() | Only 14 pages are availabe for public view |
Abstract The need for sharing knowledge between peers in virtual enterprises is rapidly growing. This knowledge may exist in different formats like web pages or libraries of shared documents over private or public networks. To search this data the early techniques suggested were to search the whole documents space which impacts the performance badly and also retrieves lots of unrelated results. Information retrieval techniques suggested indexing the documents before searching to save them in easier fomlat for searching algorithms. The indexing process uses the vector space model to represent the data as a vector of temlS and weights of every temT. These vectors can be used to compute the similarity between documents to support the recommendation process. Also categorizing data and searching only relevant categories can help in enllancing performance and relevance of results. Unfortunately, every peer has his/her own point of view to categorize his/her own data. The problem arises when a user tries to search for some information in his/her peers’ exposed data. The seeker categories must be matched with its neighbors categories. There are many proposed solutions to support the matching process and provide the end users with the best recommendations suitable for their requests. Every technique makes the matching process on its own way and presents the results to the user. Categories are usually classified in a tree hierarchy with different content which makes the matching process very difficult. In this work, we purpose a way to enl1ance the recommendation process based on using implicit ontology relations. This helps in recognizing better matched categories in the exposed data, which can recommend to the user where he can find his target. We show that this approach improves the quality of the results with an acceptable increase in computation cost (namely, the execution time of indexing and querying). |