Graph based forecasting for Social networking site

Abstract

The Social networking site play an important role in today's world thereby attracting lots of researchers to take advantage of the user's information available in these sites. Mining the database using different algorithms like association rule mining require multiple database scan. In this research forecasting is based on the directed weighted social graph. It deals with visualization of a dataset and prediction of some occurrences based upon this data. The methodology proposed is to generate a social graph of user's actions and predict the future social activities using graph mining. A dataset from the social networking site is considered and converted to a directed, weighted social graph. This graph is updated dynamically based on the changes in the database of social networking site. By creating some mathematical rules applied on the graph, we could project the future activities of users in terms of community memberships, the strength of a relationship between two users without knowing the content of the discussion. We can also find the most popular community. To find the efficiency of this method, the result interpreted by this experiment will be compared to other methods used for prediction like Apriori etc.