earticle

영어

Information retrieval systems and Web search engines apply highly optimized techniques for compressing inverted indices. These techniques reduce index sizes and improve the performance of query processing that uses compressed indices to find relevant documents for the users' queries. In this paper, we proposed a novel technique for querying compressed Arabic inverted indices in search engines. The technique depends on encoding Arabic terms stored in the inverted indices of Web search engines and information retrieval systems. This minimizes the storage space required for those terms in the index and decrease the number of comparisons needed for query matching in the query processing stage. The number of comparisons is decreased by minimizing the number of bytes that represent the terms in the index and applying the same encoding technique to the keywords of the query. The results showed a 38% reduction in total size of the index. The average number of comparisons to find a word is also decreased in the new index. Both sequential and binary searches were decreased by: 13.58%, and 38.63% respectively relative to the total number of comparisons of each keyword in the query.