List of some ongoing and past projects

Currently, I am involved with two different research groups: AlgoViz and Ensemble.

AlgoViz: This is a portal that provides a community oriented virtual space for educators who use algorithm visualizations. AlgoViz portal was officially launched at the beginning of Fall 2009. Currently, we are looking at how we can create an active online community. To increase user participation, we are offering various services as a way to lower the participation barrier for a new user. We are also collecting traffic data and analyzing it to find various trends that might be helpful in assisting existing users. AlgoViz provides metadata of its' catalog entries following the OAI-PMH protocol.

Ensemble: A distributed portal for CS education. This is the other project I am involved in. At Ensemble, we collect computing education material, provide a place for communities to collaborate and share education material, and host various education tools.

Previous projects

PIM: Organizing files could be a daunting task for many. Organization often requires going through the contents of the files and making sure that a directory contains files with similar theme. When categories overlap or are closely related to each other, it becomes even more cumbersome to decide where to finally place the file. Remembering the placement of such confusing files are also tricky. To ease the process of organizing and re-finding of a material we developed a prototype called Content based Intelligent Organizer. This tool looks at the content of a file and uses this information to suggest some relevant labels that already exist in the tool or are new. The tool learns from the user behavior and uses the knowledge as a guidance for the next set of label suggestions.

InfoViz: This is a course project for analyzing the cellular signaling pathways (STKE). Signaling pathways are relations between proteins that transform cellular signals to appropriate biological responses. Our observation is that a relation between two components of a signal can appear in more than one pathway that might aid the biologists to identify a new phenomenon. We developed a tool that provides a visualization of the performance of clustering the pathways as well as linking two different pathways that might not have anything in common. More about this project.

Graph based document mining: This was my Masters thesis (PDF) which involved using an Association Rule mining algorithm for text document. We created document graphs using WordNet. Then used FP-Growth algorithm to find frequent subgraphs and clustered the documents using those graphs. Because we used WordNet, the frequent subgraphs actually represented frequent senses appearing in documents. Mining the FP tree for normal transaction database, for which FP-Growth as created; is easier compared to large document-graphs as the items of a traditional transaction are individual and have no direct connection among them. In contrary, as we look for subgraphs within graphs they become related to each other in the context of subgraph similarity. The computation cost makes the original FP tree mining approach somewhat inefficient for text documents. We have modified FP-growth thus making it possible to generate subgraphs from the FP tree for text documents.