Mining the Volatile Social Networks

(project, diploma, bachelor's, master's thesis)

Social Networks like Blogs, Twitter, Facebook, Youtube are the new trend of our society. Through these networks a person can publish in the Web large amounts of data regarding his/her life, interests, opinions, work etc. As the person is growing, his/her interests are evolving in time. In the same way, the type of content and the type of data that he/she shares on the internet are changing too.

Since, we expect that a person's published data should reflect the changes in the person preferences, interests and/or life status or society status, the goal of this project is to extract useful patterns from the data that social network's users are publishing and also to monitor how these patterns evolve over time.

We are searching for new ways/algorithms to monitor the "online life" of a person, to understand the way that his/her profile is evolving and to outline any changes in his "social network" life. Except for the content generated by the user, his/her links within each network (describing interactions with other users) could be also used for the analysis.

Except for studying single users, we are also interested in studying content and network evolution within the whole social network, e.g. in Twitter. In particular, we are interested in extracting topics of interest and detecting their evolution in time (i.e. stories) and also, in detecting communities of users and monitoring how these communities evolve over time. Moreover, we are interested in detecting the most influential users for a specific topic and also in recommendation techniques (for example, what users to follow in Twitter).

Example of user content evolution

Example of user network evolution

Tasks

• User profile definition / Profile change detection and monitoring
• Topic detection and monitoring (i.e. story building)
• Find the most influential users for a specific topic/user
• Recommendation techniques

Requirements
Good programming skills in Java
Knowledge of KDD concepts (mainly clustering and classification)
Independent work

Contact
If you are interested in this topic or if you have further questions please contact:
Irene Ntoutsi
Note that the communication and the cooperation would be in English.