Mining the web data using data mining techniques for identifying and classifying the user access behavioral patterns.

Abstract

The one of the largest and most widely used document repository is worldwide web. It has been used for mining data since many decades. It has been proved as one of the most helping platform to assimilate, disseminate and retrieve information. But unfortunately its success has only become its enemy. It seems like an ocean of information in which users are drowning not sailing. The information is so huge, diverse, dynamic and unstructured natured that users face the problems of information overloaded while interacting with the web. Here the issue of QoS cops up. It’s needed for web developer to know what the user really wants to do, predict which pages the user is interested in and provide the user the WebPages by knowledge of users navigational patterns to improve QoS. This project mainly focuses on cleaning the data i.e. sever web log file, processing the data according to some specific strategy, identifying the users using maximal forward reference algorithms and classifying them into predefined classes. Here supervised learning is used to train the classifier. We have carried out this project using an educational institute’s log file as input data. Our project work has been used to implement the model for providing the desired information to the user if the data at the backend can be maintained appropriately and grouped into different classes. We have done the implementation using the corresponding data can be given to the user, ie instead of giving hundreds of links related topic to the user, more appropriate links pertaining to the user can be given. Hence the prevision rate can be increased.