Clustering Based Social Relationship Recommendation Algorithm

Nowadays,with the rapid development of information technology,a large amount of fragmented and irregular information begins to fill our daily life,as the providers and users of such information,people not only hope to profit from it and get accurate information data,but also hope to protect their privacy,not to be used by people with intentions,recommendation algorithm is the primary choice of information filtering at present,and there are various methods to deal with the above problems,this paper proposes a recommendation algorithm based on social relationship,by using the similarity factor inherent in social relationship,data analysis is carried out on hidden social relationship,and potential similar relations between users are mined,it not only meets the needs of users to obtain accurate information and data,but also avoids the problem of personal privacy disclosure,he main research contents are as follows:(1)Clustering based social relationship recommendation algorithm(CBSR)is using rating data to calculate a user’s social activity,at the same time,the Canopy algorithm is used to optimize the K-means algorithm clustering center,the final set of similar users can be calculated by adding potential social factors into the revised similarity,score prediction is made based on similar user sets and the final recommendation results are obtained.(2)Faced with the problem of single data source,a combined recommendation algorithm combining movie attributes and social relations is proposed(MASR),firstly,the movie item portrait is constructed according to the movie attribute label information,and then the extracted feature words are put into the Doc2 Vec model to obtain movie label document vector data for similarity calculation,then,the hybrid similarity is obtained by weighted aggregation with the similarity calculated based on social relationships to form a similar movie set and make recommendations.