A Survey on User Behaviour Prediction Using Web Server Log

Abstract

Web prediction is a classification problem in which we attempt to predict the next set of request that a user may make based on the knowledge of the previously visited pages. User behaviour prediction deals with collecting the web server logs, prepossess the data and analyze the pattern. The main goal of this process is to extract the useful pattern from raw data collection. Prepossessing contains cleaning the data, user identification and session identification which saves 80% of processing data. Web server log contains much useful information like server ip, time, date, error code, browser etc. there are various prediction models like n-gram session identification, association rule mining, markov model and support vector machine are available. Here comparison is done on the basis of accuracy, processing time and scalability.