Our research addresses feasibility issues on applying state-of-the-art technologies on multi-agent systems and semantic Web technology to various advanced application systems.
Approximation Approaches for Very Large-scale Combinatorial Auctions
Based on our ultra-fast approximation approaches for very large scale combinatorial auctions, we solve numerous feasibility issues on efficient allocation of resources (e.g. electricity, radio frequency and bandwidth, etc.) among self-interested attendees, as well as providing efficient measurement and simulation of performances of those methods.
Fast and Approximate Query Processing Techniques for Feasible Reasoning and Ontology Mapping-based access to Linked Open Data
We are developing techniques and architectures to realize feasible access to heterogeneous open data with ontology-mappings and approximation of inference on them, by applying AI-based techniques.
Highly Scalable Multi-agent Traffic Simulations Allowing Complex Behaviors of Agents on Muti-core Devices.
Cooperation and negotiation among agents is also crucial to make software systems smarter and more efficient. We are developing foundations of implementing such mechanisms as well as testing and using those mechanisms efficiently on various scenarios such as large-scale traffic simulations with modern multi-core hardware systems.
Own web site
We encourage students to actively develop their skills and present their works at high-level conferences and journals.
If necessary, we also develop base technologies for an implementation. (Figure shows the MiLog agent programming platform)