社交网站用户兴趣模型研究

鉴于社交网站广告投放系统的需要,将用户的兴趣爱好分为长期兴趣和短期兴趣两种,利用用户的注册信息获取长期兴趣,利用用户的浏览行为获取短期兴趣。为这两种兴趣分别建立用户的稳定兴趣模型和即时兴趣模型,并将二者结合建立实时的用户兴趣模型,用以完整实时的表现用户的兴趣爱好。同时,采用基于自然遗忘的模型更新方法,将人类的自然遗忘结合到模型的更新方法中。最后,通过实验验证了该模型的有效性和准确性。

Research on User Interest Model of Social Network Sites

In view of the needs of ADS-Advertisers which are used in social network sites, user interests are divided into two categories which are long-term interest and short-term interest. By using the user registers information to get long-term interest and short-term interest could be gathered by catching users' browser behavior, the stable interest model and the immediate interest model could be built separately. Using both models together, a real time user interest model will be created which could form complete interest of users in real-time. Meanwhile, the update method is based on natural forgetting that means to combine human forgetting with the model update method. In the end, the validity and accuracy of this model will be verified by the experiment.