CERN Domain

A real world scenario from the CERN institute, where conflicting beliefs regarding the operation of the particle accelerator were detected, was studied and used to test the applicability of research work carried out (see Beliefs and Conflicts in a Real World Multiagent System).
CERN is an European research institute and its particle accelerator compound is one of the world's most sophisticated high energy research centres. The accelerator operation and the maintenance of the underlying computing control system are complex tasks, difficult to survey, and where the necessary knowledge and the control system are naturally distributed.
The specific cooperation scenario studied is centred around an expert system developed at CERN using database technology referred as BCD Checker. This scenario is particularly well suited for the investigation of conflict resolution because conflicts and conflict resolution are inherent and necessary for the everyday operation of the accelerators. If the agent that creates the beam schedule for the accelerators had complete knowledge about the physical constraints of the accelerators and their control system, then it would be too complex. As a result, the BCD Editor agent has incomplete knowledge - when it creates plans to achieve its goals, it creates beliefs that may be inconsistent with the beliefs of the BCD Checker agent. The BCD Checker agent incorporates the knowledge about the accelerators and control system constraints. So, in order to create an executable beam schedule for the accelerators, the agents may have to revise their beliefs and create new plans. The application of the developed distributed belief revision model to solve these problems and the relationship between belief revision and conflict resolution were investigated.

Project Location Domain

The research work carried out was also applied to the project location domain and resulted in the development of an intelligent distributed decision support tool called Dynamic Interactive Project Location Multi-Agent Test bed (DIPLoMAT).
Project development is a complex task comprising three main stages: (i) the project design phase - where the effort is focussed on the search of the technical solutions; (ii) the project location phase - where the effort is concentrated on finding locations which satisfy the project requirements as well as comply with the applicable legal restrictions; and, finally, (iii) the implementation phase.
Typically, the project team after establishing the technical solutions, identifying the project land requirements (area, natural resources, transports, price, etc.), and finding an adequate location, has to submit the whole package to the required evaluation agencies. The evaluation agencies role is to verify, within their domain of competence, the compliance of the applicable legislation. The outcome of the evaluation agencies is crucial since it decides the future of the project: an approval leads directly into the implementation phase; a conditional approval leads to requirements re-assessments and/or project re-location, and re-evaluation cycle; and a rejection dictates the unsuccessful end of the project. To avoid unwanted rejections or endless location searches, resubmission and re-evaluation cycles the design stage should take into account, not only the project requirements (technical and economical solutions worked out by the project team) but also, the external applicable restrictions (social, legal, and environmental constraints verified by the evaluation agencies). In this scenario adequate project location is an inherently distributed activity where conflicting beliefs regarding the location of the projects occur frequently.
The aim of the intelligent decision support system developed (see Environmental Decision Support: a Multi-Agent Approach for an early description of the application) is to assist the project team in the finding adequate project locations which comply with the applicable legislation and satisfy the project requirements. The expected inputs are the project requirements (natural resources, area, transport requirements, etc.) identified by the project team and the physical/social/economical characteristics of the candidate region (stored in a Geographical Information System). The output is the set candidate areas that satisfy the applicable restrictions and the project requirements (it can be an empty set if there no adequate areas were found).
To ilustrate the reasoning performed enclosed you find the result of the search for an adequate location for an industrial project, according to the specified project equirements. The areas found are the small red patches in the lower right display (the Soil for Industry display).  
A fundamental feature of the DIPLoMAT system is its ability to perform true WHAT/IF analysis. This ability is fully supported by the implemented argumentation based distributed belief revision model called Distributed Belief Accomodation and Revision (DeBATER).