Research on Personalized Recommendation Based on Web Usage Mining Using Collaborative Filtering Technique

Abstract:

 Collaborative filtering is the most successful technology for building personalized recommendation system and is extensively used in many fields. This paper presents a system architecture of personalized recommendation using collaborative filtering based on web usage mining and describes determinedly data preparation process. To improve recommending quantity, a new personalized recommendation model is proposed in which takes the good consideration of URL related analysis and combines the K-means algorithm. Experimental results show that our proposed model is effective and can enhance the performance of recommendation.


Key-Words:

Collaborative filtering, Personalized recommendation, Web usage mining, Data preparation, Cluster algorithm, Similarity

Introduction:

The World Wide Web (WWW) provides a vast source of information of almost all types and this information is often distributed among many web servers and hosts. If these chunks of information could be extracted from the WWW and integrated into a structured form, they would form an unprecedented source of information. The resulting growth in on-line information combined with the almost unstructured web data necessitates the development of powerful yet computationally efficient web data mining tools. Web data mining can be defined as the discovery and analysis of useful information from the WWW data. The web mainly involves three types of data: data on the WWW, the web log data regarding the users who browsed the web pages and the web structure data. Thus, the WWW data mining should focus on three issues: Web Structure mining, Web Content mining and Web Usage mining[1~4]. Web usage mining includes the data from server access logs, user registration or profiles, user sessions or transactions etc. A survey of some of the emerging tools and techniques for web usage mining have been presented[3]. Current research issues in web data mining in the context of the web warehousing project called WHOWEDA (Warehouse of Web data) were discussed.

Web personalization recommendation is an important task from the user point of view as well as application point of view. Personalization recommendation helps the organizations to develop customer-centric Web sites. For example, web sites that display products and take orders are becoming common for many types of businesses. Organizations can thus present custom Web pages created in real time, on-the-fly, for a variety of users such as suppliers, retailers and employees. The web log data obtained from various sources such as proxy server, web server, etc. helps for web personalization recommendation according to interest and tastes of users community. Personalized recommendation content enables organizations to form lasting and loyal relationships with customers by providing individualized information, offering and services. For example, if an end user customer visits the site, he should see pricing and information that is appropriate to him, while a re seller will see a totally different price and shipping instructions. This kind of personalized approach can be effectively achieved by using web mining tools.

Personalization recommendation is any action that makes the web experience of a user personalized to the user's taste. The experience can be something as casual as browsing the web or as significant as trading stocks or purchasing a car. Existing approaches used by many web-based companies, as well as approaches based on collaborative filtering (CF) [6,7,8] rely heavily on getting human input, e.g. user profile, for determining the personalization actions. Currently, more and more AI technology has been applied to improve the capability of recommendation system. To gain exact and real-time recommendation, some recommending methods have been constructed based an different theory[9~12] , such as collaborative filtering algorithm, bayesian network, association rule mining, clustering, hurting graph, knowledge-based recommendation, etc. In these recommending methods, collaborative filtering algorithm is a successful method, widely applied in many e-commerce systems, such as recommending movies or news for user. CF algorithm evaluates the current customer’s near neighbours according to the rating data.

Through neighbours’ rating data, the current customer’s evaluation for a new product can be forecasted, then, the recommendation for current customer can be obtained. So, the similarity measurement among customers and customers’ classifying are the foundation of personalized recommendation system. There are many classifying methods and algorithms have been applied in many applications, but in e-commerce system, the customers’ classifying has its unique features. Recently, a considerable amount of work has been carried out on web usage mining. Mobahser . presented automatic personalization of a web site based on web usage mining. Techniques have been developed to predict HTTP requests using path profiles of users. Extractions of usage patterns from web logs using data mining techniques have been presented The rest of this paper is organized as follows: In section 2, we discuss web usage mining technique, basic collaborative filtering model and personalization recommendation system. A system architecture of personalization recommendation using CF technique based on web usage mining is proposed at last. The process of data preparation is detailedly described in section 3. In sction 4, we give a method of clustering user transactions combined similarity of URLs. A recommendation online algorithm is given in section 5. Experimental results and the discussion of the results are presented in section 6. Finally, conclusion and future work are given in section