Research On Fast Building Of Social Group And Group Recommendation

Social network itself includes hardware, software, services and applications, it will connect people together through a computer network, and to approach the really social network by an unlimited trend. However, user exposure to the data, but they can not appear effective access to useful information. The emergence and development of search technology and recommender technology, to some extent, greatly alleviated this problem.Search engine technology requires the user to give a clear message needs so it provides data to user. However recommender techniques without such restrictions. Personalized recommender technology used mostly in e-commerce, online travel and social networks. With the development of social networks, the target for recommender is often more than one person. Group recommender refers to the target for recommender is more than one person.The study for group recommender relate not only traditional personalized recommendation system, but also refer to the assessment for the recommender for the group, the interpretation of the recommender results, social choice theory and rank aggregation mechanism.Group recommender system, which requires to recommend a list to group and try to meet all individual’s preferences. Previous studies focused on aggregating of individual user preferences and impacting of individual in the group. Without taking into account real-world applications, such as the group should be generated quickly due to individual’s preference. Aiming at this application scenario, this paper modeling the group recommender system using the following techniques:locality sensitive hashing, collaborative filtering and aggregation algorithm based on voting. So that the system can generate the group rapidly and the recommender result well. Finally, we implement the system in real social network data sets. The experiment results show that the system has the ability to generate required groups quickly. Compared to baseline, our model performs better, where mean nDCG value increases 0.7%,0.6%,2.6%,4%,9%,16% when group scale is 2,4,8,16,32,6...