New Framework for Discovering Important Posts in Social Networks

Due to the growing popularity of social networks, user’s access to various information in these platforms is increased. Users communicate, click, view and share different contents that reflect their interests, making the themes on such social networks just as diverse as the users themselves. The behavior of members of social networks allows their interests to be narrowed down to a particular category. Detecting important posts defined as the most visited posts can provide some interesting insights from cyberspace user’s activities. In this paper, a framework for discovering daily important posts (most popular posts by views count) is introduced. This framework includes 3 main parts; the first is data gathering, then a model for identifying advertisement messages based on machine learning methods is provided and the third part is a method for clustering text content of posts using LSH. To the best of our knowledge, this paper is the first attempt to use LSH for clustering Persian texts. The proposed framework employed on Telegram instant messaging service in this article but it is also applicable to other social networks such as Instagram and Twitter