AdaptiveWeb Recommendation for New Navigation Trends

Abstract

All website managers want to attract more users. Web Usage Mining is a pattern-identifying device widely adopted for this purpose, but it has proved to be inadequate as the demand for on-line work increases. Though Web Usage Mining helps identify the users’ surfing habits and does it effectively, the process of pattern discovery is time-consuming as it requires processing all access records occurred in the past days or hours. New trends of web browsing, however, are forming all the time. Web content needs to be updated accordingly. If the new content is hot or noticeable, users’ behavior will change instantly. E-commerce or news websites are concrete examples. On-line users are practical and goal-oriented. They will not use a recommendation engine again if the information it provides no longer meets their needs. It is therefore essential that a recommendation engine have a high level of accuracy. In this paper we propose a recommendation engine that can quickly respond to new navigation trends and provide users with the best suggestions on hyperlinks. This engine is especially effective when there is a change in web content but this information has not been assimilated into regular patterns yet. Ultimately, the research aims to equip websites with facilities that can customize users’ needs automatically and efficiently.