“FP BASED HYBRID APPROACH FOR PREDICTING WEB USER’S FUTURE REQUEST”

Authors: Rahul Kaul, Shraddha Kumar

Published in: IJRRETAS, Volume 2, Issue 5, pp. 1-7 (01 May 2016)

Impact Factor: 2.3 (2016)

Certificate: View Certificate

Abstract

The use of Web technology has increased by an excellent extent within the recent times. Millions of users spend time on internet to get data or for recreational activities. Along with satiating their own purpose, the users leave behind a detailed path of all the online pages accessed and also the frequency with which they’re accessed. This information is of dominant use to several industrial domains like ecommerce websites, social networking sites, entrepreneur franchises, etc. This web log contains lot of information so it is preprocessed before modeling. The web log file is preprocessed and converted into the sequence of user web navigation sessions. The web navigation session is the sequence of web page navigated by a user during time window. The user navigation session is finally modeled through a model. Once the user navigation model is ready, the mining task can be performed for finding the interesting pattern. Modeling of web log is the essential task in web usage mining. The prediction accuracy can be achieved through a modeling the web log with an accurate model to improve the performance of the servers, caching is used where the frequently accessed pages are stored in proxy server caches. Pre-fetching of web pages is the new research area which when used with caching greatly increases the performance. In this paper, a better algorithm for predicting the web pages is proposed. Clustering of web users according to their location using clustering is done and then each cluster is mined using FPGrowth algorithm to find the association rules and predict the pages to be pre- fetched for storing in cache. HMM is used to focus on the hidden stages of web log files. Keywords: Web Usage Mining, Semantic Web, Domain, FP Algorithm, Sequential Pattern Mining and Hidden Markov Model, Prediction, web log.

Introduction

Web Usage Mining: In recent times, Web Usage Mining has emerged as a popular approach in providing Web personalization [1]. Web usage mining is concerned with finding user navigational patterns on the World Wide Web by extracting knowledge from web usage logs (we will refer to them as web logs). The assumption is that a web user can physically access only one web page at any given point in time that represents one item.

Conclusion

Web usage mining model is kind of mining to server logs. Web usage mining used for the improvement of improving the requirement of the system performance, the customers relation and realizing enhancing the usability of the website design. The main goal of the proposed system is to identify usage pattern from web log files. FP Growth Algorithm is used for this purpose. Apriori is a classic algorithm for association rule mining. The main drawback of Apriori algorithm is that the candidate set generation is costly, especially if a large number of patterns and/or long patterns exist. The main drawback of Markov model is that it can’t focus on the hidden stages of web log files. The FPgrowth algorithm is one of the fastest approaches for frequent item set mining. The FP-growth algorithm uses the FP-tree data structure to achieve a condensed representation of the database transaction and employees a divide-and conquer approach to decompose the mining problem. Our experimental result shows that the FP-growth & HMM method is efficient and scalable for mining both long and short frequent patterns. In future the algorithm can be extended to web content mining, web structure mining.