Pilot Projects

We selected a few social information systems as pilot projects. Those projects should cover the three components of collective intelligence: "humans," "agents," and "knowledge." We explain these three pilot projects below. The following table shows the characteristics of each pilot project, and the figure illustrates the role of pilot projects in our research activities.

This project is directed toward sharing language resources (dictionaries, parallel texts, machine translators, morphological analyzers and so on) in the world as atomic services, and to create field-oriented composite services. For example, the board of education in each prefecture of Japan prepares multilingual parallel texts and guidance documents for pupils and students, whose first language is not Japanese. By integrating those language resources, high quality language support can be provided for various education fields.

This project will develop atomic services that search/analyze/organize information about human resources and construct human teams that can solve tasks as a composite service. Although an ad-hoc labor market can be formalized by using matching theory, the existing theory has difficulty dealing with human factors such as human relations. To overcome this difficulty, we will conduct subject experiments and participatory simulations and find the combination of expertise that offers the highest added value.

This project is to share traffic information such as traffic regulations and traffic jams as atomic services and model human drivers who behave under social norms including traffic regulations as composite services. We will collect driving data from human subjects on a 3D driving simulator to model human drivers. For example, traffic in the forthcoming aging society can be reproduced by massive multiagent simulations with high ratios of aged driver agents. In addition to solving the problems of traffic jams, this project contributes to environmental issues such as predicting the amount of CO2 emission.

Table : Salient Features of Pilot Projects

Pilot Project	Language Resource Sharing	Ad-hoc Labor Market	Large Scale Traffic Simulation
Salient feature	Realize "knowledge" sharing thru' cooperation of intellectual property holders with various incentives.	"Humans" having a variety of skills and intentions get together. Create a new labor market.	Traffic is predicted by using "agents," which model a variety of human drivers.
Research issues on collective knowledge structuring	Create new services by combining various language resources (dictionaries, translation software) distributed in the world.	To develop a system able to accumulate information about humans having a variety of technologies and skills, and search/analyze/ organize the accumulated information.	Accumulate origin/destination data of drivers with diverse goals, and combine them with road information/ sensing data in real time.
Research issues on collective decision making	Match incentives among language resource providers and users. If multiple providers contribute to a single service, reasonable profit sharing should be realized.	Forming a virtual team able to attain a goal by ensuring workers' incentives are compatible with each other. Matching should satisfy economic and psychological constraints.	Human driver's behavior is affected by information of other drivers. Adequate information service is required.
Likelihood of actual implementation	High. Collaboration with various research institutes already started to share language resources.	Medium. A huge amount of resumes are available.	High. A large scale experiment is possible with an already developed multiagent simulator.

Figure : Role of Pilot Projects