My research interests include Semantic web, web services, databases and AI. Specifically, I would like to focus on applying Artificial Intelligence techniques and methods to all aspects of the web service composition problem. I currently work with in the LSDIS Lab at the University of Georgia. Here is my CV and Research Statement.
For the past year, I have been working with Dr. Prashant Doshi in the area of Web service compositions (WSCs), specifically adaptive WSCs. WSCs are multiple Web services strung together to form a new service with different functional properties. This creates a degree of desireable service automation. These are business enterprise applications that are currently being utilized by many companies, including IBM and Oracle. Our projects and research thus far has attempted to optimize a WSCs by minimizing total overall cost in a volatile or dynamic environment. Volatility refers to changing of Quality of Service parameters (ie cost, time, reliability) in the life cycle of the WSC. In order to function at an optimal level, the changes in the parameters must be monitored and used to create a more robust web process.
We have incoroporated various concepts from the AI community to do this. First, we have modeled the WSC as an Markov Decision Process, which identifies the stochastic nature of Web services. Then we have added a mechanism called the Value of Changed Information (VOC), borrowed from Value of Information theory, which will help the WSC adapt to chages that occur by deciding if newly revised information will actually impact the process. The work of VOC has been demonstrated in several conferences (see the Publications page) and has shown promising results. We believe we can improved this methodology in the near future.
Our new research will get a little more interesting in the next few months or so, focusing instead on the process designer (user client) itself. We intend to enhance our dynamic system by including risk-sensitive agents, that is, WSCs that take "more realistic" preferences into account. For example, when it is possible to incur huge losses in a process, a decision-maker may decide to change how his/her process will be setup. Stay tuned for updates in this branch of our research.
I have had a few opportunites to get involved in professional service, where I have been able to actively engage in not only demonstrating my ideas but learning about what is going on in the general services community. I have served as a paper reviewer in multiple venues such as SCC2007 and SCC2008. Also, I have been heavily involved with the highly successful Workshop on Web Service Composition and Adaptation (WSCA-2008 and WSCA-2009) in conjunction with the IEEE SCC and ICWS conferences. We had a real good turnout for both workshops and hope that we can progress even further next year.
If I get an academic position (knock on wood), I would like to organize a research group focused on the game of poker. Besides being a fun, interactive (and occasionally profitable) game, it is the ultimate AI problem. In those terms, it is a complex, multi-agent, non-deterministic system in which decision making requires thorough knowledge of the environment as well as an agent preference setting sensitive to riskful behavior.