Achieving more accurate results searching in Persian semantic web pages applying Markov’s model and analysis of user’s prefetch behavior

The enormous amount of information on the World Wide Web has led search enginesto make use of advanced techniques to improve the efficiency of data recovery anddeliver results that are better for the user. Extending queries is one of these techniques.The methods that have been developed so far to expand the queries, have mainly beendeveloped based on a dictionary and only use linguistic features. In this article, first,by using the Markov’s model, the possible pages and topics of the user request onPersian web pages were prefetched and integrated with emotional analysis and thepattern of user’s survey behavior. Finally filtering on fetched results to display moreaccurate information, led to a method providing a semantic web search engine. Toevaluate the proposed method, RapidMiner was used to simulate and C# implementthe method. In the proposed method, the pre-fetched search time of Web pages wasabout 2.92 times more than the Trie-Traverse method, compared to the Listing method2.25 times, and improved 1.53 times than the BedTree method. Memory usage wasimproved by about 3.35 times compared with the Listing method and 2.01 times higherthan the BedTree method, which provides the performance of the proposed method.

Keywords:

Search Engine , Semantic Web , Semantic Search , Markov Forecast , User Behavior Analysis , Persian Semantic Search