This project modelled the effect of contact duration when broadcasting large messages in a dynamic network. Take the example of a video that has to be broadcasted in a bluetooth network, where the connection between devices keeps changing. This project acknowledged that the entire file may not be sent in one contact. We modelled the contact duration using Gaussian and Poisson distributions and addressed the question 'Which broadcasting algorithm works best for this contact duration model?'
As a group of two people, we implemented two citation recommendation systems based on PageRank and Linear Regression and compared the results of the two.
As a group of ten people, we built an automatic highlights generator for cricket matches. The system uses commentary from the website espncricinfo.com as the data and runs a Naive Bayes classifier to determine if an event is a highlight or not. Some of the features used were the presence of a exclamation mark and the length of commentary. I built the parser and the system for computing precision and recall.