Research On News Recommendation In Social Media Based On User Comments

Social media is a kind of new online media giving users a great space to interact with each other, which enables each user to create and disseminate information. The continuous development of Internet technology had made the social media demonstrate a strong vitality; Web has become an important means of social media to disseminate information. Social Media includes Internet forums, blogs, twitters, wikis, podcasts, instant messaging, social networking and content community, and so on. One form of social media of particular interest here is self-publishing, or customer-generated media. In self-publishing, a user can publish an article or post news to share with other users. Other users can read and comment on the posting and these comments can, in turn, be read and commented on. The news posting makes it presents features of strong time-bound, large quantities, fast incremental and large scale. In this article, we present a social media news recommendation system that based on users’comments. It considers the news posting and relevant comments when recommend news, makes news recommendation consider more users’preference and interest.The Section 2 of this article describes the related knowledge about this work. First, we describe the definition and development of news recommendation. Second, we describe the related knowledge about information retrieval. In this part, we describe the related technology in detail, especially PageRank algorithms and the application in this article. The next part, we describe the related knowledge and the core technology of topic detection and topic tracking, including the topic models construction and the technology of topic detection and topic tracking. And in the next part, we describe the definition of relevance language models and its core technology. By the way, we describe the related work about user interface. In the last part, we describe the unique features of this work, which have not been attempted in previous work.This paper describes the design details of news recommendation system in Section 3. First starts from the description of comments relation, and explains the comments score which calculated from the graph-based model. The next part describes the topic profile construction. And the next, we describes the use of relevant language model for relevant news stories retrieval. We describe the user interface in the last part.This paper describes the experimental design and experimental evaluation in details in Section 4. The experimental data comes from the news site Digg’s news and its comments, the recommended news from Reuters news websites. This paper uses accuracy and novelty to evaluate the performance of News Recommendation System. There are three experiments presented in this paper to verify the combined news reports and commentary outline document for the news recommendation system performance impact. After experimental verification and further information presented in this paper, it obtained the superior performance of the recommendation system.The final Section summarizes the research work of this paper and outlooks the future research. The summary includes the news recommendation system’s overall design and the system specific implementation proposed by this paper, and noted that the system deficiencies and shortcomings; The outlook is mainly about the future research of news recommendation system presented in this paper, but also prospects the research framework in this paper applied to other research areas.