A novel framework for an efficient online recommendation system using constraint based web usage mining techniques

With the fleetly development of the internet, discovering useful knowledge from the World Wide Web became a censorious issue. With the huge volume of information present in the internet, user needs a help via recommendation system. From the user’s log data lot of recommender systems developed to predict the user’s next request when they view the web pages. However, each recommender system has its own advantages and drawbacks. A novel hybrid recommender system is proposed using a modified DBSCAN, modified prefix span algorithm and Genetic algorithm to mine the user’s sequential navigational patterns, and then a hybrid recommendation model is proposed. The proposed Hybrid recommender system produces better prediction accuracy than the existing single recommender systems. Testing results recommend that the hybrid system is better in predicting the next request page of the web user.