Personalized Model Combination for News Recommendation in Microblogs.

Facing large amount of accessible data everyday on the Web, it is difficult for people to find relevant news articles, hence the importance of news recommendation. Focused on the information to be used and the way to model it, each of the existing models proposes its own algorithm to recommend different news to different users. For these models, personalization is only done at the recommendation level. But if the user chooses a model that is not appropriate for him, the recommendation may fail to work accurately. Therefore, personalization should also be done at the model level. In our proposed model, the first level is defining four atomic recommendation models that make fully use of the social and content information of users and the second level is adapting to each user that atomic models effectively used. Experiments conducted on two real datasets built from Twitter and Tencent Weibo give evidence that this double level of personalization boosts the recommendation.