Wikipedia enriched advertisement recommendation for microblogs by using sentiment enhanced user profiles

Advertisement recommendation on the Web is a popular research problem. For microblog platforms, different requirements arise due to the differences in the context of social media and social network. In this work, we propose an advertisement recommendation technique for microblogs. The proposed solution uses all contents of the messages (texts, captions, web links, hashtags), and enhances them with sentiment data and followee/follower interactions expressed as microblog posts to generate a new user model. As another novel feature, Wikipedia Good Pages are used as general background knowledge for matching user profiles and advertisement contents. On the basis of the similarity between advertisement vectors and user profile vectors, the most related advertisement for the selected user is determined. Evaluation results show that the proposed solution performs better for advertisement recommendation on microblog platform and works faster in comparison to other techniques.