Objective
The MADS project aimed to build three levels of tools for the development of expert systems:
-message-passing languages
-description languages for knowledge representation
-reasoning and strategy programming.
MADS was to develop techniques for implementing message-passing languages that exploit new highly parallel architectures. The techniques were to cover issues of run-time support such as allocation, migration, garbage-collection and persistency of actors. A description system was developed to support the basic mechanisms of knowledge representation: conceptual hierarchies, inheritance, and attributions. To perform reasoning on the description system, rather than providing a fixed strategy, primitives and constructs were developed that allowed the programming of deductive strategies tailored to specific applications. Strategy execution was to be performed concurrently by a large number of actors each exploring a small portion of the knowledge-base.
The project developed techniques for implementing message passing languages that exploit new highly parallel architectures. The techniques covered issues of run time supportsuch as allocation, migration, garbage collection and persistency of actors. A description system was developed to support the basic mechanisms of knowledge representation: conceptual hierarchies, inheritance, and attributions. To perform reasoning on the description system, rather than providing a fixed strategy, primitives and constructs were developed that allowed the programming of deductive strategies tailored to specific applications. Strategy execution was performed concurrently by a large number of actors each exploring a small portion of the knowledge base. Prototypes of message passing architectures and two description systems, OMEGA and KRS, that will exploit highly parallel machines, were developed. OMEGA is a description oriented knowledge representation which allows reasoning and includes a viewpoints mechanism. Graphical tools provide an efficient environment to support the knowledge engineering activity. KRS is a concept system which allows reasoning and includes inheritance mechanisms and a consistency maintenance system. A graphical user interface was developed. Parallelism in knowledge representation was investigated. The use of the language LISP in parallel processing and appropriate knowledge representation led to the implementation DELPHI Common LISP, including a multithread facility for concurrency, and the integration of the language PROLOG.
Prototypes of message-passing architectures and two description systems, OMEGA and KRS, that will exploit highly parallel machines, were developed:
-OMEGA is a description-oriented knowledge representation which allows reasoning and includes a viewpoints mechanism. Graphical tools provide an efficient environment to support the knowledge engineering activity.
-KRS is a concept system which allows reasoning and includes inheritance mechanisms and a consistency maintenance system. A graphical user interface was developed. Parallelism in knowledge representation was investigated.
The use of LISP in parallel processing and appropriate knowledge representation led to the implementation DELPHI Common LISP, including a multi-thread facility for concurrency, and the integration of PROLOG.
Exploitation
The prototypes that have been implemented constitute the first steps to future industrialisation. Both OMEGA and KRS are being used in pilot applications outside the project.
Fields of science
Programme(s)
Topic(s)
Data not availableCall for proposal
Data not availableFunding Scheme
Data not availableCoordinator
55049 Viareggio Lucca
Italy