International Journal of Electronics Communication and Computer Technology

Predicting the next page to be accessed by the Web users has attracted a large amount of research. In this paper, a new web usage mining approach is proposed to predict next page access. It is proposed to identify similar access patterns from web log using K-mean clustering and then Markov model is used for prediction for next page accesses. The tightness of clusters is improved by setting similarity threshold while forming clusters. In traditional recommendation models, clustering by nonsequential data decreases recommendation accuracy. In this paper involve incorporating clustering with low order markov model which can improve the prediction accuracy. The main area of research in this paper is pre processing and identification of useful patterns from web data using mining techniques with the help of open source software. KeywordsSimilarity measure, K-mean, Markov model