I have been involved with and worked on a number of research projects in Web IR and bioinformatics areas for the past several years. For example, I worked on a project to develop an information retrieval system for the gene information. Text mining technologies were used to extract a number of essential keywords for some important genes. Based on these keywords, we built an indexing and searching system for a collection of gene information.
Currently I am working on a peer-to-peer social information retrieval application called Sixearch.org (6S). The application was designed to address the scalability limitations of centralized search engines by providing a collaborative platform for searching and sharing Web documents among participating peers. In this project, an adaptive query routing algorithm was designed and integrated with a topical crawler and a search engine to provide an alternative search application. A new evaluation framework, instead of the traditional precision and recall approach, was also developed to assess the performance of peer networks.
Another project I am working on is to employ a version of the model-based pattern matching approaches previously developed for blob identification in time-lapse microscopy experiments. We hypothesize that the criteria for determining threshold values for correct object identification will be inherently limited by the local, and less so by global, signal to noise properties of the image. Therein, we have created methods for automatically determining the threshold values for correct object identification at a non-arbitrary confidence level based on both the local goodness of fit score and a local measure of signal and noise.