Recommender systems on social networks and websites have been developed to reduce the production and processing of queries. The purpose of these systems is to recommend users various items such as books, music, and friends. Users' recommendation on social networks and i More
Recommender systems on social networks and websites have been developed to reduce the production and processing of queries. The purpose of these systems is to recommend users various items such as books, music, and friends. Users' recommendation on social networks and instant messengers is useful for users to find friends and for marketers to find new customers. On social networks such as Facebook, finding target users for marketing is an integrated feature, but in instant messengers such as Telegram and WhatsApp, it is not possible to find the target community. In this paper, by using graph and modeling the intergroup behavior of users and also defining features related to groups, a method for recommending Telegram users has been presented. The proposed method consists of 8 steps and each step can be considered a separate method for user recommendation. The data used in this paper is a real data set including more than 900,000 supergroups and 120 million Telegram users crawled by the Idekav system. Evaluation of the proposed method on high-quality groups showed an average reduction in error by 0.0812 in RMSE and 0.128 in MAE.
Manuscript profile
