Personalized microblog recommendation using sentimental features

• Main Authors: Wenjuan Cui, Yi Du, Zhihong Shen, Yuanchun Zhou, Jianhui Li
• Format: Conference Proceeding
• Language: English
• Subjects: Collaboration; Computational modeling; Feature extraction; History; Microblogging; personalized recommendation; Real-time systems; Sentiment analysis; Twitter
• ISSN: 2375-9356
• DOI: 10.1109/BIGCOMP.2017.7881756

Microblogging services have been popular in recent years. There are a large number of real-time microblog messages generated in each day which results in the information overload problem especially for the users with many followees. Personalized microblog recommendation can help the users out of the trouble of information overload. It is an interesting and important research topic with wide applications. Many kinds of features are used in the microblog recommendations in the existing algorithms. In this paper, except for the features such as the user's microblog posting history, reposting history, the relations with other users, and the contextual knowledge, we also utilize the sentimental information to help with the microblog recommendation. We first build a sentiment classifier based on the contextual information of the microblogs and get the sentimental feature set. Then a latent factor model incorporating the sentimental features and other information in microblogs is designed. Finally, we develop an experiment plan to evaluate the performance of the method. We believe that the utilization of the sentimental features will improve the performance of microblog recommendation.