Doctoral studentMax Planck Institute for Software Systems
Campus E 1 5
Office phone: +49-681-9303-8612
I am a doctoral student at Max Planck Institute for Software systems (MPI-SWS). My advisor is Dr. Krishna Gumamdi. I completed my dual degree curriculum (B. Tech and M. Tech) in Computer Science and Engineering from IIT Kharagpur, India in July 2010. My M Tech advisors were Dr. Debdeep Mukhopadhyay and Dr. Abhijit Das. My Erdös Number is 4.
My research interests broadly lie in the area of networked systems with an emphasis on privacy of users in such systems. More specifically I am interested in measuring and analyzing privacy related user actions in large networked systems (e.g., today's online social networks) which yields answer to many important and exciting questions with practical significance. My current work focuses on understanding how users behave to protect their privacy in online social media and to further help them by designing privacy-preserving tools.
Controlling online privacy via exposure: In this project we argue that although access control is the most widely used method to ensure privacy of their content from other users, but access control can not capture many of the privacy violations actually happening in Online social networks. We argue that the main reason is: online users care less about who can access the content and care more about who will actually access the content. We define exposure set of a piece of content as the set of users who accesses the piece of content eventually. In this work we show that one should actually control exposure to manage privacy of their content and we address the challenges in designing such systems. Further we explore how the users today are trying to control longitudinal exposure (the exposure of their content posted long back in time) and investigate the problems with current exposure control methods.
Characterizing Anonymous Social Media Content: Recently anonymous social media sites like Whisper and Secret have become quite popular. However how the users are using these sites is still not clear. In this project our goals are to better understand the usage of such sites and to investigate how their usage differs from that of non-anonymous social media sites like Twitter. More concretely we aim to look into both the what and why of the content that are shared anonymously in today's anonymous social media. Our investigation also comprises understanding and detecting known cyber aggressions like hate speech which (intuitively) aggravated when users post anonymously in these social media sites.
Simplifying friendlist Management in Online Social Networks (OSNs): Today, OSNs like Facebook and Google+, provide a friendlists abstraction which allows the user to classify their friends into different groups. Friendlists are very useful because it allows the user to easily share content with select friends. However, creating a friendlist is a tedious manual process today. In this project we developed a Facebook application, which automatically proposes friendlists for a user by leveraging the structure of the user's local social network.
Defending against Sybil attacks using Social Networks: We identified that existing Sybil defense schemes using social networks could be classified into two broad categories, namely Sybil detection schemes and Sybil tolerance schemes. We show that all existing Sybil tolerance schemes are designed by mapping the social network to a credit network. We developed a Sybil tolerant system called Genie, to limit large-scale data aggregation in online social networks. Sybil tolerance schemes do not suffer from the limitations of Sybil detection schemes, but they do not scale to large-scale social networks, thus hindering their practical deployment. We developed a novel technique to scale up Sybil tolerance schemes by scaling up operations over the underlying credit network.
We created some online applications as part of our research to help social network users better understand and manage their privacy.
Check Your Secondary Digital Footprint on Twitter: In Twitter, people may converse with you by mentioning your name in their tweets. These conversations constitute your secondary digital footprint. Secondary digital footprints are not created or controlled by you. However, they can still leak your personal information. Our Twitter application aims to help you check what information others leak about you on Twitter (You will need a Twitter account to use it ).
Friendlist Manager: Friendlists in Facebook are a great way to share your content with the people you intend to. But they are a huge pain to create and update. Our Facebook application was designed to facilitate and simplify management of your friendlists. Unfortunately the new version of Facebook API do not allow developers to fetch the data the app needed to use, consequently the app is not live any more. You can check the functions of this (now discontinued) app here.
Privacy IQ: Privacy IQ is a quiz that measures both your understanding of how privacy works on Facebook and your knowledge of your own privacy settings. However due to the change in Facebook API this app too is not live any more. You can check the functions of this (now discontinued) app here.
I enjoy teaching quite a bit and so far contributed in the following courses during my doctoral studies as teaching assistant and tutor. In addition to teaching (tutorials and some lectures) my responsibilities included helping to design the course, designing and evaluating assignments.
My curriculum vitae is available in PDF format (updated on December 2016).