چكيده به لاتين
The problem of the evolution of groups in different time periods is an important issue that is discussed in political, social and commercial fields. For instance, periodic movements, the growth of religious and radical groups can be fundamental examples. In the digital world, online groups (such as telegram groups, Facebook, etc.) have grown significantly, in terms of the number of users involved, the amount of activity and the amount of impact. But because of the costly nature of large-scale computations, perhaps the most basic questions about how the growth or changes in such networks are unanswered; what are the structural features and what are they going to be? Which of the groups will have faster growth? The structure of the information network lies in itself, which can be viewed as a set of features. Using this feature extraction you can achieve this. Feature extraction can be obtained by learning a feature. The most important feature of learning is the generic feature that makes it possible to use on a variety of issues.
You can not ignore the level of user activity over time to predict joining a group. A user may have a large network of friends but has not done anything in the group for a significant period of time. In many of previous works, the number of posts and comments sent has been used to measure the level of users' activity, but such information is not always available in the database. Here, we want to consider features such as membership in different time periods as an activity, for example, if a person is a member of many groups but has not been a member of any groups lately, he has less activity than a person who is member of less groups in general but has recently become a member of many groups. By using such features for user activity level, it is possible to increase the accuracy of prediction of group changes.
