The main aim of the project is the development of a logic programming system that will offer the necessary facilities for building distributed constraint logic programming applications as well as agent based systems. The secondary aim of the project is the implementation of a distributed application using the proposed platform, in order to evaluate the product of the project. This application concerns the distributed construction of an inter-departmental schedule for courses and exams. The selected problem presents a number of interesting features like scheduling shared resources, co-ordination issues between the distributed solvers, etc. The pilot application will serve for evaluating the strengths and weaknesses of the system and of course for debugging purposes.
A logic programming platform for such applications should support basically two features:
- Communication facilities for co-ordinating the teams of agents participating in the application environment.
- Constraint Solving Capabilities.
The proposed system will be based on the existing CS Prolog II system developed by ML. CS-Prolog II is a system that offers:
- parallel program execution through sequential communicating Prolog processes
- real time features
- inter-process and network communication (TCP/IP)
- WEB based GUI
- an interface for accessing external data bases (PostgreSQL).
Possible application areas of the system include the usual CLP applications such as planning, scheduling, time-tabling, resource allocation, configuration, operational research, etc. However, the target application area of the system is focused on parallel/distributed systems, such as distributed planning and scheduling, distributed workflow management and resource allocation, etc. The area of distributed constraint systems has gained significant interest in the last few years by the research community and is considered to be one of the most promising ones.
The proposed pilot application concerns the distributed construction of inter-departmental schedule for courses and exams. The problem is common in most universities and especially in departments that belong to the same faculty, since the former usually share lecture rooms for courses and exams and in many cases human resources (lecturers) as well. The pilot system will consist of a number of independent solvers (agents) each one corresponding to a distinct department that will
- form a schedule according to its own needs
- co-operate to produce the final schedule of the faculty.
A number of interesting research issues arise from such an architecture that concern basically the negotiation between the departmental agents, communication issues, the implementation of an appropriate scheduler for each faculty etc. It should be noted that the algorithms/methods developed during this implementation could potentially be employed to a variety of scheduling problems that occur in other areas.
Partners
- Intelligent Systems group, Dept. of Informatics, Aristotle University of Thessaloniki, Greece
- I. Vlahavas
- Ilias Sakellariou
- Nick Bassiliades
- ML Consultanting & Computing Ltd., Budapest, Hungary