My M. Tech. Project was in area of Social Network and my topic was Study of the characteristics of different types of Facebook users by exploiting users timeline activities. Online Social Network(OSN) has become an important part of the Web. Millions of people including celebrities, politicians use OSN for delivering messages, but anyone can use it as a platform for spreading misinformation. Many malicious activities and misinformation spreading starts by infiltrating OSN.

This study focuses on three types of users; (i) Users who send friend requests to arbitrary users, (ii) Users who accept friend requests from strangers and (iii) Users who reject friend requests from strangers.

This study analyses the characteristics of the above three types of the users by exploiting their Timeline activities.

In this study we also tried to detect Sybil user(fake user). Sybil detection algorithm are available, but all are based on Topological and statistical properties. Depending on user's activities there is no machine learning algorithm. In this project we applied machine learning techniques on user's activities to classify user into Sybil or Honest user.

As we dont have labels for dataset, I created fbsybil chrome extension. You can download it from here. For instructions on how to install it and use it click here. For installing fbsybil extension you can see below video.

If following features are present in user's profile, that profile may be fake(Sybil). Priority of feature is decreasing from top to bottom.

1. If single real Profile photo is there or all photos are of famous people or sceneries or No photo.

2. If School/College/Work place is inappropriate.

3. If from long time nothing shared or No status updated.

4. If no Group Joined or no Page liked.

5. If there are comments like "Thanks for adding me" or "Is that you" or "Show Original Photo" and still all these comments are unanswered.

6. If random friends are added.