Name: Nitin Agarwal
Homepage: http://www.cs.cmu.edu/~nitina (Never completed and hasn't been updated in a while)
Student: 2nd Year Master Student at LTI
I am crediting the course. I hope to learn in general how interaction in social networks can be understood by mining text data corresponding to the interaction. Other phenomena like evolution of a social network over time etc. which might manifest itself in form of text are also of deep interest to me.
For my research I have been working mainly on probabilistic models of text especially in the genre of scientific articles. I have decent understanding of latent Bayesian models. I have been working on a summarization system for scientific articles. I have been using latent models such as LDA and variants of the model and other clustering techniques to assist in discovery of common themes across scientific articles.
I have some ideas about a potential project which are in the form of a rough draft at this link . I propose a graphical model for modeling topics conditioned on relation between the authors in a networked set of scientific articles (such as Citeseer or ACL anthology). At this point I am really not sure if its viable amount of work for a course project, so I would be discussing that with my project team mates to come up with more attainable objectives.