Metasearch engine result optimization using reformed genetic algorithm

Abstract

Metasearch engine is a system that applies several different search engines, merges the returned results from the search engines and presents the best results. Principal component of the metasearch engine is the method applied for merging the given results. The most of existing merging algorithms are relied on the information achieved by ranking scores which is integrated with the results of different search engines. In this paper, a reformed genetic algorithm (RGA) is proposed for aggregating results of different search engines. In the RGA, a chaotic sequence is applied to select the parents to mate, preventing the RGA to get stuck in local optima. The combination of pitch adjustment rule and uniform crossover (CPARU) is also proposed to further mutate of chromosomes. In the problem of optimizing search engine results, the proposed method tries to find weights of documents’ place to allocate each document to the best place. Therefore, the only required information is to know the number of the search engines that finds each document in the corresponding place. Accordingly, this technique works independently of the different search engines’ ranking scores. The experimental results have depicted that the RGA outperforms the genetic algorithm (GA), Borda method, Kendall-tau genetic algorithm (GKTu) and Spearmen's footrule genetic algorithm (GSFD) methods.