Work Area: Logics and Logic Programming
Keywords logic programming, types and objects, non-monotonic reasoning, meta-reasoning, knowledge-base systems
Start Date: 1 August 92 / Duration: 36 months / Status: running
[ participants / contact ]
Abstract This project is concerned with the extension of logic programming to improve its knowledge representation and problem-solving capabilities. Three main areas of knowledge representation will be investigated: types and objects, non-monotonic reasoning, and meta-reasoning. The work on problem solving will be concerned with program development and will contribute to knowledge engineering. The project builds on the results of COMPULOG (3012).
COMPULOG II has the long term goal of developing the foundations for an integrated, logic-based software environment, including associated software development tools, that deals with all aspects of computing. The language and tools will extend the logic programming paradigm by incorporating enhancements from related paradigms in the areas of databases, artificial intelligence and mathematical logic. COMPULOG II will develop and use the Gödel language developed in COMPULOG to integrate many of the other developments.
The work is organised into four themes. The first two are concerned with extending the expressive power of logic programming and developing methodologies for building applications. Special emphasis will be given to integrating objects and logic in the context of data and knowledge based systems. Work on non-monotonic reasoning will be concerned both with extensions and restrictions of negation as failure, and with its use to formalise common sense reasoning. Special attention will be given to the computational behaviour of such formalisation. Work on meta-reasoning will concentrate on improving the technology for meta-programming and on treating theories (programs, program specifications, databases, and knowledge representations) as objects. The third theme, program development, is concerned with the dynamics of constructing, maintaining and analysing appropriately extended logic programs, program specifications, databases, and knowledge representations. The fourth will implement and integrate appropriate results from COMPULOG, using the Gödel language as a platform for integrating work on types, meta-reasoning and program development, and integrating the same techniques with techniques of non-monotonic reasoning for knowledge base systems. This work will focus on the development of structured knowledge representations and on query evaluation and integrity checking methods. Extensions needed to make logic programming closer to the natural language used in legal documents (types, non-monotonic reasoning, and meta-reasoning) will also be investigated.
Types and Objects area
Two approaches towards an integration of object-oriented features with deductive systems have been considered. On the one hand, the deductive OODB system ConceptBase, developed in Aachen, has been enriched by concept language characteristics. On the other hand, more features have been added to concept language in order to make them applicable to practical systems, where ConceptBase has been chosen as a testbed when possible. Both approaches concentrated on procedures for querying knowledge bases and for reasoning about queries.
Non-monotonic and Metareasoning area
Emphasis was placed on the following main research topics:
Program Development area
The work has mostly focused on control issues and specification issues. More specifically the following topics were studied for logic programming languages: specification, synthesis and transformation of programs, program analysis, termination, occur-check problem, partial deduction.
Knowledge Base Systems
The ConceptBase deductive object-oriented database was further developed with the aim of using it as a platform for testing the integration of solutions from deductive databases, object-oriented databases, and concept languages. Also the subject of belief revision, dealing with time in deductive databases, and the problem of diagnosis and repair were studied.
Both theoretical and practical work was pursued. In particular, ways of efficient implementation of Goedel meta-programming facilities were studied. Also, the problem of emitting code which is as clean as possible (eg does not use the var predicate) was investigated.
The work focused on the interpretation of legal knowledge and its formalisation in declarative systems. A number of extensions of logic programming needed to make it closer to the natural language used in legislation have been identified.
The knowledge gained from the research carried out in this project will have an impact on the design of logic-based software environment. The programming language Godel will be a useful tool for declarative programming based on logic programming in which non-monotonic and meta- reasoning can be properly formalised. The deductive object-oriented database system ConceptBase will be a sophisticated tool for database applications.
In the selection below preference was given to publications which appeared or were accepted for publication. Additionally, publications in different areas were selected.
Further information about COMPULOG 2 is available from the COMPULOG 2 home page <URL:http://www.comlab.ox.ac.uk/archive/compulog.html>.
SMC/CWI - NL
NL - 1098 SJ AMSTERDAM
Katholieke Universiteit Leuven - B
ECRC Munich - D
RWTH Aachen - D
Universität des Saarlands-DFKI - D
Universitá di Pisa - I
Universitá di Roma La Sapienza - I
Universitá di Roma Tor Vergata - I
UNINOVA Lisbon - P
University of Uppsala - S
Imperial College of Science, Technology and Medicine - UK
University of Bristol - UK
University of Edinburgh - UK
Université d'Aix Marseille II - F
Prof. Dr. K.R. Apt
tel +31/20 5924135
fax +31/20 5924199
COMPULOG 2 - 6810, August 1994
please address enquiries to the ESPRIT Information Desk
html version of synopsis by Nick Cook