Masters Thesis Research Update - Anurag, Dibjyajyoti and Sumit
Speaker: Anurag Korde
Start: Tuesday, December 06, 2011, 10:30AM
End: Tuesday, December 06, 2011, 11:30AM
Location: ITE 325 - B
Abstract: In this week's lab meeting, Anurag Korde, Dibyajyoti Ghosh and Sumit More will give an update on how their Masters thesis research is progressing.
Anurag will talk about "Entity Linking and Disambiguation for Smartphone platforms". With increasing number of social networks, smartphones and applications there is a need of creating a framework for information integration across social networks and applications and create an unified record for each person/entity. The problem in information integration is that of entity disambiguation. It determines whether two objects in an ontology refer to same real world object. This will enable system to learn a large amount of contextual information. We will try to build a system which integrates information across Facebook, Twitter, Google contacts, phone contacts, calender, etc and create a super profile of a person.
Dibyajyoti will be talking about "Application privacy control in smartphones through semantic context awareness". Existing mobile operating systems primarily offer permission based policies at install time and runtime. Projects like Secure Application INTeraction (Saint) and Mockdroid extend a popular mobile middleware to restrict access to resources based on application settings and by incorporating signature and application configuration based policies. This work focuses on resource obfuscation through semantic policies. He will describe the prototype system he is working on and will briefly touch upon the challenges.
Sumit will talk about the progress on his thesis work: A 'pre-facto' threat/vulnerability detection model.
Data is the currency of today's IT era. With the number of data sources accruing with each passing day; cyber crimes of data theft, espionage etc are increasing as well. A post-facto analysis of such cyber crimes do help in preventing similar future attacks; however the system remains vulnerable to newer types of attacks. We try to analyse different data sources and come up with a pre-facto threat/vulnerability detection model which will monitor the data streams and predict if a vulnerability is being exposed or if there is a potential future attack in making.
Host: Tim Finin