Flickr group recommendation using rich social media information

Today online social media communities have spanned the globe, browsing news from social networks almost becomes an essential part in our daily life. Groups organized by users always share something interesting. Joining groups which fit the users׳ tastes will help them to obtain information. However, traditional group recommendation methods usually focus on how to recommend an item to a group of users. In this paper, we study how to recommend groups to an individual user and reveal the factors which push a user to join groups. In social networks, a commonly adopted recommendation method takes advantage of the tastes of a user׳s trust neighbors and recommends groups which his/her neighbors have joined. It will performs poorly for the inactive users who have few trust neighbors. To overcome this problem, we try to find users׳ similar neighbors using tag information, which is not only from users׳ photos but also from their favorite photos and the common friend information. Hence we propose a group recommendation scheme utilizing users׳ trust neighbors and similar neighbors׳ tastes. We do the experiments on a real-world Flickr dataset and obtain a promising result especially for inactive users.