Research on Analysis Method of Network User Preference

At present, researches on user preferences are becoming increasingly widespread. Traditional researches use simple cosine similarity algorithm to calculate the similarity between users and just focus on one factor. So this paper study a more complete preferences research method and consider more factors of network to analyze data. This paper gives content-based recommendation algorithm based on the user preferences. It uses cosine similarity algorithm to calculate the similarity between users and uses LDA to get improvement, and screens according to user impact factor. Finally, this paper analyses micro-blog user data in detail. As a result, we get more accurate recommendation results and verified the feasibility of the algorithm by experimental analysis.