Keywords
Citation
Harwood, C.J. (1999), "Lecture Notes in Artificial Intelligence 1415. Subseries of Lecture Notes in Computer Science. Methodology and Tools in Knowledge‐Based Systems", Kybernetes, Vol. 28 No. 9, pp. 1084-1094. https://doi.org/10.1108/k.1999.28.9.1084.2
Publisher
:Emerald Group Publishing Limited
The 11th International Conference on Industrial and Engineering Applications of Artificial Intelligence and Expert Systems – IEA‐98‐AIE, held at Benicàssim, Castellón, Spain, June 1‐4, 1998 has been given extensive coverage by the publishers Springer, and the proceedings have been published in two volumes of some 2,000 pages. Volume I has the title Methodology and Tools in Knowledge‐based Systems and Volume II: Tasks and Methods in Applied Artificial Intelligence.
The temptation is to review the two together but the division into the two has to be respected even if the editors were content to reprint the same preface in each volume. Volume I is considered here but both are contributions to the conference’s “theme for 1998” which was “New Methodologies, Knowledge Modeling and Hybrid Techniques”.
In their preface the three editors of Volume I reiterate what is the accepted aim of Applied Artificial Intelligence, that is to “render computational a large portion of non‐analytical human knowledge”. This they say is done by first building knowledge level models of analysis and synthesis tasks in scientific and technical domains, such as those performed daily by human experts in the field. There are numerous examples of these fields and more are appearing in rapid succession. They range from medical diagnosis, telecommunications engineering, architecture to education. What is then required is that these models of such widely differing applications are transformed so that their entities and relations can be linked to the primitives of a programming language. Finally, we are told by the authors, a computer program can be produced which is then subject to the usual software engineering “checks and balances”. They write about validation, evaluation and maintenance at a time when the worries of the arrival of the year 2000 demand that the word verification should be a dominant feature of any software engineering programme.
The organisers of this conference have recognised that there has been concern about the sound foundations and methodology, as well as the necessity of developing efficient procedures to make models operational. Most researchers and developers in this area would agree that since its origins in the 1950s there has been a lack of methodology and foundations and would welcome any attempt to make knowledge engineering into a more robust discipline and study. The papers published in this volume attempt to do that and we are told they have been conscientiously reviewed by two referees per paper. The papers published are said to be the final version of the accepted papers that incorporate the reviewer’s comments. Readers, however, are entitled to ask whether they contain the views of other participating delegates. The current formula of organising conferences seems to preclude any opportunities for authors to revise their papers as a result of their presentations and the resulting discussions. Unfortunately, nowadays we are on arrival at a conference given a published version of all accepted papers. In this volume the published contributions have been limited to some ten pages per paper. It is therefore not clear whether these were the versions published before the meeting and consequently were perhaps abridged versions, or very short presentations. Readers may well agree with the publishers and the conference organisers that two volumes of proceedings are enough and being succinct has advantages not only for conference goers but also for referees and in particular book reviewers.
As to the standard and variety of the papers included in this collection of contributions the editors tell us that of the 291 contributed and invited papers submitted from 41 countries 187 were selected. The range of topics must have been encouraging to workers in this AI research area. To help readers the volume is divided into four main parts:
- 1.
(1) Methodological aspects (seven papers).
- 2.
(2) Knowledge modeling (seven papers).
- 3.
(3) Formal tools – with the sub‐topics:
- •
Fuzzy knowledge representation and inference (eight papers)
- •
Uncertainty, causal reasoning and Bayesian networks (four papers)
- •
Qualitative reasoning (13 papers)
- •
Neural networks (one paper)
- •
Evolutionary computing (eight papers)
- •
Object‐oriented formulation (one paper)
- •
Hybridization techniques (four papers)
- •
Context‐sensitive reasoning (three papers)
- •
Multi‐agent systems (five papers)
- •
Divisible problems (four papers).
- 1.
(4) Generic tasks of analysis – with sub‐topics:
- •
Perception (six papers)
- •
Natural language understanding (four papers)
- •
System identification (three papers)
- •
Monitoring (three papers)
- •
Fault diagnosis (six papers)
- •
Predictive models (four papers).
The divisions of topics into these sections follows the Call for Papers, but some additional ones have been included as a result of them being the subject of invited sessions. The list included here gives a guide, therefore, to both the standard of the contributions and also to the popularity of certain lines of research and development. It would, of course, be interesting to know how many papers in each section were refused by the referees. A statistic that not many editors of books or journals are happy to divulge.
The editors are absolutely right when they say that one of the most frequent deficiencies in the majority of methodological developments lies in ignoring the conclusive step about how to render the models operational with the final result of an implemented system. This does lead to both scepticism about some contributions and also, in general, accounts for the evident lack of credibility seen towards many of the AI researches. The real‐world often appears to be a distant place and applications relevant to it not considered seriously. The editors in their preface have summed up the situation in an honest way; they say: AI researchers are sometimes seen as just blowing smoke. Potential readers of this volume should realise that when they peruse it. Questions arise such as: what support is there for this claim? Why isn’t a rigorous methodology applied?
That said there is much of great interest in the text and it was often a pleasure to see the mathematical approach to so many aspects of this research has met with success. Most contributors realise the need for precision in their research and where possible, the presentation of rigorous and validated results of real world applications and problems are the yardstick by which their work should be measured. Discussion papers are, of course, highly desirable and a paper such as Fra n˘ová and Kooli of the CNRS and the Université Paris Sud, on “Theory of Constructible Domains for Robotics: Why?” is one which is both stimulating and useful. Other contributions deserve similar praise. It was surprising that only one paper on Neural Networks was included. Brasil et al., whose authors mainly did come from the Federal University of Santa Catarine in Brazil, examined both complexity and cognitive computing. They described a hybrid expert system to minimise some of the complexity problems presented in the AI field. Most readers will agree with the assessment of these authors that the complex systems associated with human activity are often poorly defined. They believe that cognitive computing provides an effective and efficient way to analyse technological processes and human activities.
An author index is provided which is in itself a useful Who’s Who of scientists working in the various fields.
The Lecture Notes series by Springer has provided some valuable texts and those in the fields of Artificial Intelligence and Computer Science have enhanced this reputation. Publishing the proceedings of a conference presents innumerable problems and readers need to be aware of what they are buying. Short contributions can only whet the appetite or indeed discourage one from buying such a volume. This volume has been well presented and the papers properly refereed and edited. It would therefore be a worthwhile volume to peruse at one’s leisure and a useful reference for, at least, the immediate future. It also has a historical significance and that in itself makes it a text worth consulting in the department or institutional reference library.