Efficient and Anonymous Web-Usage Mining for Web Personalization

• DOI: 10.1287/ijoc.15.2.123.14444
• Publication Date: 2003-12-29T15:49:49Z
• Authors: Cyrus Shahabi; Farnoush Banaei-K...
• Keywords: Web analytics; Web engineering

The World Wide Web (WWW) is the largest distributed information space and has grown to encompass diverse resources. Although web growing exponentially, individual's capacity read digest content essentially fixed. full economic potential of will not be realized unless enabling technologies are provided facilitate access Currently personalization most promising approach remedy this problem, mining, particularly web-usage considered a crucial component any efficacious web-personalization system. In paper, we describe complete framework for mining satisfy challenging requirements applications. For online anonymous effective, usage must accomplished in real time as accurately possible. On other hand, should allow compromise between scalability accuracy applicable real-life websites with numerous visitors. Within our web-usage-mining framework, introduce user-tracking accurate, scalable, implicit collection data. We also propose new model, feature-matrices (FM) discover interpret users' patterns. With FM, various spatial temporal features data can captured flexible precision so that trade off based on specific application requirements. Moreover, tunable complexity FM model allows real-time adaptive pattern discovery from define novel similarity measure specifically designed accurate classification partial navigation patterns time. Our extensive experiments both synthetic verify correctness efficacy efficient personalization.