Jump Over Left Menu
Introduction by Dr J M Taylor
Background to the RARM
In May 1982 a Proposal for a Specially Promoted Programme (SPP) in Intelligent Knowledge Based Systems (IKBS) was presented to the Computers and Communications Subcommittee of the SERC. This had been prepared by an ad hoc Panel chaired by Dr J .M. Taylor, Chairman of the Subcommittee. The Panel members were Dr J. Fox (Imperial Cancer Research Fund), Dr J. Howe (Edinburgh University), Dr P. Kent (Rutherford Appleton Laboratory), Dr R. Kowalski (Imperial College), Mr D. Price (Marconi Avionics), Dr K. Sparck Jones (Cambridge University), Dr V. Stenning (Systems Designers Ltd) and Mr J. Monniot (SERC).
The Proposal was for a ten-year programme on IKBS, which were seen as a major development in computing in the nearer future. IKBS were defined by the Proposal document as systems of limited intelligence carrying out only a simple task, but one complex enough to involve ramified knowledge and to require inference to manipulate this. IKBS were seen as a step forward from the unintelligent systems of the present to the more fully intelligent systems of the distant future, by being systems able to work on an incompletely-specified task, if only on one such task. The objectives of the Programme were to promote the research and advanced development required to design and construct IKBS, to transfer IKBS technology to industry, and to increase the supply of qualified and experienced manpower needed for this very difficult work.
The main elements of the proposed Programme were therefore
- the support of academic research on the basic capabilities essential to IKBS, on the generic tasks they carry out, and on the technologies providing the virtual machines on which they are implemented;
- the promotion of joint academia/industry demonstrator projects building complete, well-founded IKBS for particular domain applications;
- the encouragement of education and training through courses, studentships and fellowships, etc;
- the provision of appropriate common base and infrastructure facilities.
The Proposal emphasised the need, in the context of growing international activity in the area, for UK action on IKBS. It further stressed the urgent need, as a prerequisite for UK action, for a build-up of the UK IKBS community. The community is small and fragmented, and has suffered from past attacks on artificial intelligence and from more recent cuts in university funds. Effective research in IKBS requires a large effort over a long period, and a successful Programme thus depends on early identification and consolidation of research groups and project teams.
The Proposal document therefore recommended that a RARM be held as soon as possible for the twin purposes of bringing the community together and of obtaining its detailed, collective response to the SPP suggestions for IKBS R&D.
Form of the RARM
Since the Meeting was intended to stimulate community action on IKBS, it combined a broad spread of invitations with a tightly-structured programme based on working groups focusing on the technical areas of IKBS research, and on their joint implications for the proposed IKBS Programme.
The two-day Meeting was attended by some 120 people. The invited participants represented a large number of actual and incipient IKBS R&D enterprises . There were about 75 people from universities and polytechnics, and about 35 from industry, government and other organisations.
The major part of the Meeting was devoted to work in 9 parallel technical Syndicates, dealing with Knowledge Representation, Inference, Natural Language, Vision and Object Manipulation, the Man-Machine Interface, Expert Systems (2), Machines and Programming - Research, and Machines and Programming - Common Base. Though it was recognised that research in IKBS cannot be compartmentalised in this way, it was nevertheless thought that the Syndicate topics constituted useful foci for working purposes, and that this concentration, plus the small number of people in each Syndicate, would allow effective formulation of recommendations for the IKBS Programme. Further, each Syndicate session began with presentations from each member on their work and interests, promoting the exchange of information between industry and academia and providing a base for the Syndicate's development of its contribution to the definition of the SPP. To structure these contributions, each Syndicate was invited to provide a specific output in the form of lists of
- research programmes and projects in its area, with indications of timescales, relative priorities, and interested parties;
- requirements from other areas;
- necessary support facilities;
- demonstrator projects, with indications of interested parties;
- training opportunities and requirements.
To complement the technical divisions of the IKBS field represented by the Syndicate sessions, these were followed by 4 parallel Group sessions devoted to major topics cutting across all the technical areas:
- Group 1, Research Programmes and Projects, was designed to consolidate, within the overall IKBS Programme, the more detailed Syndicate programme and project proposals, and to draw together possible collaborative groups of workers.
- Group 2, Infrastructure Requirements, was to consolidate Syndicate recommendations under this head.
- Group 3, Training and Education, was similarly to consolidate the training requirements and opportunities identified by the individual Syndicates.
- Group 4, Interfaces with Other Areas, was intended to define the interfaces, and hence interaction between, the proposed IKBS activity and other key areas of computing and information technology, notably VLSI, system architectures, software technology, and man-machine interaction.
The Meeting programme was thus designed to exploit the technical knowledge and interests of the members of the community, to refine and make more concrete the details of the proposed Programme and, as a necessary support for this, to begin the formation of groups of workers for particular programmes and projects.
Output from the RARM
The Meeting was agreed by the participants to have been extremely useful, both as a general morale raiser, and as a device for stimulating the communal effort involved in an SPP. Though funding for the Programme is not yet available, the need to be quick off the mark when it does come was widely recognised. Thus it was agreed that further useful preparation for the Programme can be made by following up the RARM with specific proposals for special-purpose workshops, individual grants, etc., for which regular SERC support may be sought.
The specific outcomes of the Meeting are the reports of the Syndicates and Groups which form the body of this Report. These analyses and recommendations cover the whole area of the proposed Programme in considerable detail and, most notably, provide many individual research programme and demonstrator project summaries, supported by notes of those interested in participating in them. The possible academia/industry collaborations noticed here are of particular importance. More generally, the reports demonstrate that there is already, though on too small a scale, the dedicated and wide-ranging research on IKBS which is essential as a base for the expansion of activity sought by an SPP.
Thus taken together, the working group reports show
- that IKBS is an area not merely of concern to existing research workers, but of rapidly growing interest to the larger UK computing community.
- that there is a wide spread of worthwhile research ready to go;
- that the many individual efforts would benefit from the coherent sponsorship of research that an SPP would allow;
- that the outcome of such a Programme would be a significant development of IKBS ideas and implementations in the UK.
I would like to thank the Chairmen and Rapporteurs of the Syndicates and Groups for their contributions to the success of the RARM and to the weight of this Report.