Research On A Recommendation Algorithm Based On User Topic And Emotional Orientation

At present, although traditional information recommendation technologies can meet some requirement of users, due to their all-purpose characteristics, they still can not satisfy the requirement of different users' personalized feature from the different background, with the different intention and at the different time. How to provide users with information which corresponds with their personalized feature, this topic has become a significant research project. Currently, in the information domain, the research of emotion computing has attracted increasing attention, it plays a significant role in the realization of information processing based on intelligence. To personalized information recommendation systems, the degree of user's emotional preference can express well the user's emotional requirements. With the widening of applied area of the systems and the increasing new requirements continually, the research of requirement model based on user emotion may become a new direction of personalized services. A personalized information recommendation algorithm based on user topic and emotional orientation model is presented in this paper. Some theories and perspectives of personalized information recommendation and artificial emotion is researched in this paper, research of the important points and technological innovation as following.1. First, it filters the original texts for the first time based on user's topic model and recommends the information to user.2.Second, user express own emotional preference to the recommended information by the explicit feedback method that choice emotional words which such as very like, like, unconcern, dislike and so on . It set up user emotional model by user's emotional requirement, meanwhile, taking emotional orientation of the emotional words into account. A method of expanding user personalized requirements is designated by combined with emotional orientation and user emotional state.3. Finally, the information texts is classified based on the method of combining with support vector machines algorithm and the weight of emotional orientation, it recommend the appropriate information to user.Compared with recommendation algorithm of the single user topic model, the algorithm is more effective on personalized recommendation service. The algorithm has higher precision and generates better recommendation results.