The Research And Implementation Of Personalized Recommendation Based On Web Mining

Internet contents variety news range of types and sources. And exiting news aggregation website are very irrational for the reason that aggregation website presents the same news to each user ignoring the user’s interests. In order to reduce users’useless clicks and improve user experience, this thesis design and implement a news aggregator platform and personalized news recommendation system based on the user’s interest.In this thesis, with the analysis of logs we summed up user interest into long-term interest and short-term interest. Different strategy was deployed according to the type of user interest. Considering the characteristic of long-term interest, we use Tag-Entity Table as the mainly approach of modeling, which consist of news categories and the words belong to this categories. For short-term interest, we build a independent module for hot news extracting. Finally, we mixed the two types of news described above and then resort them for presentation. On the other hand, we have done a lot of work on mining user access patterns through user web logs. And we got some interesting rules as a supplement to user interest model. In the end of this thesis, our personalized news recommended system was evaluated from two aspects.