Research Areas
1 Distributed Artificial Intelligence
2 Computational Intelligence
3 Internet Computing
Current Projects
1 Intelligent Agents on the Web
Searching for authoritative Web resources on certain topics is a common task for users of the Web. Information on networks these days grows drastically (exponentially), making an effective use of information more difficult. Accessing information on the Web requires an efficient tool for search and communication. We are not satisfied with matches from available search engines because either text-based or link-based search engines suffer from different limitations.
Artificially intelligent web agents may help overcome information explosion in ways of helping a user to choose right web tools and conduct parallel multiple queries searches in a user friendly way. Research is being undertaken in the Faculty in developing intelligent search engines for information retrieval on the Web, and I am the chief researcher. I am responsible for identifying the limitations of current search engines and designing an Evolutionary Approach to overcome them. So far, I have implemented a search agent called Evagent. Web applications based on Evagent are being constructed. Analyses to the results of the applications with Evagent have illustrated that given a topic, the agent can identify the most authoritative Web resources.
What makes this successful is that I can identify the limitations of current search engines, and I think of the Web search as an optimisation problem. With evolutionary paradigm I proposed, the agent is not just for optimising; it is also creating conditions in which optimisation occurs. To my knowledge, this is the first time the evolutionary approach is introduced into the area of Web search.
Future research needs to focus on a deeper level of understanding the effectiveness and performance of our evolutionary algorithm and exploiting this information for the methodology improvement. Another area of future research can focus on contributing Evagent to more Web applications, such as Search-by-Example, Mirrored Hosts Finding, and Web page Categorization. Future work will also focus on moving Evagent from client side as an agent to server side as a search engine. More search agents like Evagent are distributed, and they can search different parts of the Web in parallel. By this parallel computation, we can cope with the dynamic nature of the Web very well.
2 Agent-based Volunteer Computing Grid.
This research work has initialised the idea of agent-based volunteering computing on the Web, which aims at high-performance parallel computing networks to be formed easily, quickly and inexpensively by enabling ordinary Web users to harness their computers’ idle processing power without needing professional skills for setting up.
A good motivation for volunteer computing could be the conventional belief that the Web is connecting millions of mostly idle machines. Harnessing these idle machines could prove to be a powerful global computing infrastructure. Volunteer computing makes it easy for people on the Web to pool their existing computer resources, and work together to solve large computational problems that no one can solve alone. Volunteer computing makes it possible to build parallel computing networks much larger and much more quickly than possible before. Potentially, such networks can involve thousands, even millions of computers distributed around the world, and can achieve performance levels of current supercomputers.
Previous Projects
1 Distributed Computation Based on Intelligent Agents
Supported by the Ministry of Education of China
June 2000 ~ October 2001
2 Distributed Swarm and Its Application
Supported by the National Natural Science Foundation of China
August 2001~ October 2001
3 Synthesis of Solution in Distributed Expert Systems
Supported by Edith Cowan University, Australia
January 1999-May 1999
4 Agent-Oriented Software Development Environment
Supported by the National High Technology Research and Development Program of China
June 1996-June 1998