The action aims to provide a more comprehensive account of the application of formal, object-oriented techniques to the whole development cycle of large-scale information systems, with emphasis on cooperative requirements analysis, conceptual modelling, specification, reification (implementation), certification, and reconfiguration. To this end, the action addresses the following research issues: semantic domains for object-oriented systems, languages and logics, coping with concurrency, transactional reification, interfacing, reconfiguration and distribution; modularisation and configuration techniques for specification in-the-large of object-based systems; joint use of several specification logics; default-based, modal specification techniques; compositional reasoning; abduction techniques in verification; object-oriented conceptual modelling; formal description of software development processes.
Several alternative semantic domains (eg hidden sorts, D-oids, sheaves, event automata, Kripke structures) have been investigated, compared and used in establishing the semantics of several languages (eg FOOPS, OBLOG, TROLL); techniques have also been proposed for generating an object domain from any given temporal logic.
Nonmonotonic specification: the usefulness of defaults in system specification was demonstrated; different semantics of defaults within propositional OSL have been proposed by imposing alternative preference relations on object life cycles; a metalogic for reasoning about default theories and an abstraction of instantiation mechanisms for defaults are being investigated.
Conceptual modelling: a comparative study of object oriented methods and conceptual modelling languages was carried out; the formal semantics of OMT is being developed with the context of an integration with TROLL; a method was developed for writing and validating CMSL specifications; certification techniques for OBLOG using OSL have been developed and tested.
Software development process: a formal, object oriented metamodel for an application context centred paradigm was proposed and compared to the traditional project centred paradigm.
Most of the actual work is done within the following subgroups: SG1: semantic domains; SG2: concurrency; SG3: configuration; SG4: defaults and reasoning; SG5: conceptual modelling; and SG6: software development process. Work within each subgroup is carried out via e-m, in individual visits to other sites, and in working meetings held on average twice a year. The work is reflected in joint papers and monographs. Most participants have an active role in at least two subgroups. A general meeting (GM) is held once a year for overall coordination and strategic exchange of ideas, back to back with a workshop (WS) open to people from outside IS-CORE (both academia and industry). During the first year the following working meetings have taken place: SG1 meetings in Lisbon (October 5-6, 1992) and Oxford (March 29-31, 1993); SG3 meeting in Lisbon (November 2-4, 1992); SG4 meetings in London (December 4-5, 1992) and Hannover (June 1-3, 1993); SG5 meeting in Braunschweig (October 18-23, 1992). The GM/WS'93 are scheduled for September 1993 in Hannover.
In the short term, exploitation of the results produced by IS-CORE will be possible within several R&D projects on languages and methods for information system design where the participants in IS-CORE are engaged, together with industrial partners. In the medium and long term, a wider impact on software engineering can be expected through more effective use of formal, object-oriented methods and specification languages. Work on deductive capabilities should also have an impact on knowledge engineering.
53754 Sankt Augustin
SW7 2BZ London
5000 LE Tilburg
1007 MC Amsterdam
OX1 3QD Oxford