Meta-search engines are created to reduce the burden on the user by dispatching queries to multiple search engines in parallel. Decisions on how to rank the returned results are made based on the query's keywords. Although keyword-based search model produces good results, better results can be obtained by integrating semantic and statistical based relatedness measures into this model. Such integration allows the meta-search engine to search by meanings rather than only by literal strings. In this article, we present Multi-Search+, the next generation of Multi-Search general-purpose meta-search engine. The extended version of the system employs additional knowledge represented by multiple domain-specific ontologies to enhance both the query processing and the returned results merging. In addition, new general-purpose search engines are plugged-in to its architecture. Experimental results demonstrate that our integrated search model obtained significant improvement in the quality of the produced search results.
Authors
Mohammed Maree
Saadat M Alhashmi
Mohammed Belkhatir
Pages From
379
Pages To
391
ISSN
0
Journal Name
Journal of Information & Knowledge Management
Volume
10
Issue
4
Keywords
Meta-search; ontology; natural language query understanding; semantic and statistical-based relatedness measures; collection fusion; experimental validation
Abstract