Performance Analysis of Web Page Recommendation Algorithm Based on Weighted Sequential Patterns and Markov Model

Web usage mining techniques helps the users to predict the required Web page recommendations. In recent times, there has been a considerable significance given to sequential mining approaches to construct web page recommendation systems. This paper focuses on developing a web page recommendation approach for accessing related web pages more efficiently and effectively using weighted sequential pattern mining and markov model. Here we have developed an algorithm called, W-PrefixSpan, that is the modification of traditional Prefixspan algorithm including the constraints of spending time and recent visiting to extract weighted sequential patterns. Then by utilizing weighted sequential patterns recommendation model is constructed based on Patricia-trie data structure. Later the web page recommendation of the current users is done with the help of markov model. Experimentation is done with the help of synthetic dataset and we present the performance report of web page recommendation algorithm in terms of precision, applicability and hit ratio. The results have shown that, the precision of our algorithm is improved by 5% than the previous algorithm. Also we have achieved high applicability in the support of 50 % and in terms of hit ratio, the proposed algorithm ensured that the performance is considerably improved for various support values.