Social network sites have attracted millions of users with the social revolution in Web2.0. A
social network is composed by communities of individuals or organizations that are connected
by a common interest. Online social networking sites like Twitter, Facebook and Orkut are
among the most visited sites in the Internet chew, (2008). In the social network sites, a user can
register other users as friends and enjoy communication. However, the large amount of online
users and their diverse and dynamic interests possess great challenges to support such a novel
feature in online social networks kwon, (2010). In this work, we design a general friend
recommendation framework based on cohesion after analyzing the current method of friend
recommendation. The main idea of the proposed method is consisted of the following stagesmeasuring
the link strength in a network and find out possible link on this network that is yet to
be established; detecting communities among the network using modularity and recommending
friends. Considering the noticeable attraction of users to social networking sites, lots of
research has been carried out to take advantage of the users ‘information available in these
sites. Knowledge mining techniques have been developed in order to extract valuable pieces of
information from the users’ activities. This paper deals with a methodology to generate a social
graph of users’ actions and predict the future social activities of the users based upon the
existing relationships. This graph is updated dynamically based on the changes in the selected
social network. The forecasting performed is based upon some predefined rules applied on the
graph.