A Personalized Recommendation System Based On Emotional Weight

With the continuous development of the Internet; although the traditional recommendation technology can meet certain needs of people, the fuzzy on the project description of user rating data, it has been unable to meet different purposes, different period and different preference of user query recommendation. Therefore, how to recommend personalized information needs of users to provide a better has become an important research topic. In recent years, comment has been heavily involved in the research on the Internet and accepted by many scholars. And the comment expresses the views, attitudes and emotional needs of users. Therefore, as users continue to be excavated Information, in the personalized recommendation system, the introduction of user information to the emotion comment recommendation system will become a new research direction recommendation system.A personalized recommendation system based on user reviews, which is by analyzing the reviews of user to find the topic comment preference item when the user selects。Thereby expand the user’s personalized information recommendation. However, in the choice of projects of users, independent reviews of projects with small and free-form features, which makes rich user emotional expression and lack of emotional expression users, differences in their comments information. Rich emotional expression of users, it’s personalized information is more abundant and perfect. And lack of emotional expression on the project and the user is not fully express their views and needs. So, strictly extended by comment, improve the user’s personalized information is likely to cause the user information is not accurate, resulting in inaccurate recommendation.To solve this problem, we propose a personalized recommendation system based on emotional weight, which is the effective use of information by user emotional weights recommended. The main contents of this paper are the following: first, to quantify user emotional comment. Evaluation unit extracts and emotional bias discrimination based on the information when the user select items expressed, thereby obtaining quantitative data user topics emotion; second, the user emotional weight calculation. Get the user data after the emotional topic, you can get emotional factor user themes, user reviews of topics, themes commented times, and effective way to measure the user’s emotional weight. So rational and effective use the user emotion comment; third, find similar users based on the degree of emotional expression value data and comprehensive user rating comment data; fourth, build a prototype system. Based on the emotion of this article personalized recommendation system test and evaluation information extraction algorithms were compared with other experiments, our algorithm is more accurate.