The Research On The Interest Analysis And Its Application Of Micro-blog Users

With the continuous improvement of the Internet, as the product of era, the social network has been widely used and development. The micro-blog, which is a complex network with the help of social relationship, based on its high-efficiency information propagation ability and the function characteristic, has become the most valuable social networking media platform with the explosive growth of users. More and more users want to express their emotion and communicate with others, no matter that, they also want to find the information they are interested in. Therefore, according to the needs of users, how to excavate the user preference from the massive social relations, pushing the information they are interested in, not only reduce the cost of the user experience, but also optimize the networks, all of which has become the central study point of the scholars and experts.In this paper, micro-blog user’s interests analysis method and the personalized friends recommendation are proposed, including the following two aspects:(1) A method to analysis micro-blog user’s interests based on the domain knowledge has been proposed. According to the requirement at different levels, the user’s attribute characteristics have been analyzed with the technology of data mining methods, we have discovered that the tag vocabulary and the blog content can’t on behalf of the user’s interests because of its irregular and the widely covered. On this basis, we have built and organized the domain knowledge base according to the user’s behavior characteristics, an idea is innovative proposed that user’s interest could be analyzed by investigating the key node users to whom he/she pays attention to. The interviews and statistical validation show that the method is effective.(2) On the basis of the user interest analysis method, the personalized friend recommendation has been regarded as the actual application scenario. We selected the recommended candidates from the second relationship with the target user, Combined with the relationship type and the interaction strength, respectively, we calculate the intimacy and similarity between the users at different level, and then work out the user recommendation value eventually. We output the final results in the shape of TOP N. Finally, the experiment has been set up to test the proposed method, through the relevant evaluation index, a comparison has been made between the method and the current recommendation strategy of micro-blog, the precision of this method has been verified. More than that, the sensitivity of the result has also been test when the user interest changing with time goes by.