Zohreh Anari - Department of Computer Engineering and Information Technology, Payame Noor University, I.R. of Iran
Babak Anari - Department of computer engineering, Shabestar Branch, Islamic Azad University Shabestar Branch, Shabestar, Iran

As the number of web pages increases, search for useful information by users on web sites will become more significant. By determining the similarity of web pages, search quality can beimproved; hence, users can easily find their relevant information. In this paper, distributed learning automata and probabilisticgrammar were used to propose a new hybrid algorithm in order to specify the similarity of web pages by means of web usagedata. In the proposed algorithm, a Learning Automata (LA) for each web page is assigned which its function is to evaluate association rules extracted by hypertext system. This learningprocess continues until the similarity of web pages are determined. Experimental results demonstrate the efficiency of the proposed algorithm over other existing techniques

Web Mining, Association Rules, Learning Automata,Distributed Learning Automata, Hypertext Probabilistic Grammar