Computer Science علم الحاسوب

Permanent URI for this collection

http://dspace.alquds.edu:4000/handle/20.500.12213/1091

Browse

Browsing Computer Science علم الحاسوب by Author "Alaa J. S. Eghreib"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    New IR & Ranking Algorithm for Top-K Keyword Search on Relational Databases ‘Smart Search
    (AL-Quds University, 2014-11-12) علاء جمعه سليمان اغريب; Alaa J. S. Eghreib; بديع سرطاوي; د. رشيد جيوسي; د. محمود الصاحب
    Database management systems are as old as computers, and the continuous research and development in databases is huge and an interest of many database venders and researchers, as many researchers work in solving and developing new modules and frameworks for more efficient and effective information retrieval based on free form search by users with no knowledge of the structure of the database. Our work as an extension to previous works, introduces new algorithms and components to existing databases to enable the user to search for keywords with high performance and effective top-k results. Work intervention aims at introducing new table structure for indexing of keywords, which would help algorithms to understand the semantics of keywords and generate only the correct CN‟s (Candidate Networks) for fast retrieval of information with ranking of results according to user‟s history, semantics of keywords, distance between keywords and match of keywords. In which a three modules where developed for this purpose. We implemented our three proposed modules and created the necessary tables, with the development of a web search interface called „Smart Search‟ to test our work with different users. The interface records all user interaction with our „Smart Search‟ for analyses, as the analyses of results shows improvements in performance and effective results returned to the user. We conducted hundreds of randomly generated search terms with different sizes and multiple users; all results recorded and analyzed by the system were based on different factors and parameters. We also compared our results with previous work done by other researchers on the DBLP database which we used in our research. Our final result analysis shows the importance of introducing new components to the database for top-k keywords search and the performance of our proposed system with high effective results.