Frontiers of Combining Systems 2

Citation

Andrew, A.M. (2001), "Frontiers of Combining Systems 2", Kybernetes, Vol. 30 No. 1, pp. 103-115. https://doi.org/10.1108/k.2001.30.1.103.5

This is a selection of contributions to the Second International Workshop on Frontiers of Combining Systems, held in 1998, abbreviated as FroCoS’98. In all, 20 papers are included, along with a “preface”. The motivation for the series of workshops is of course the observation that interesting areas of investigation or innovation often fall between established “disciplines”, and each of the papers in this collection describes a study that links two such established fields.

It is, of course, true that the whole field of cybernetics arose from this kind of linking, with the posing of such questions as: “What does the reflex of the neurophysiologist have in common with the feedback loop of the control engineer?”. The topics that are linked by the papers in this volume do not range so widely as those linked by the advent of cybernetics, but the general principle, that good material is to be found in the in‐between areas, is demonstrated. The fact that it is now unnecessary to range so widely to find these areas probably reflects the increase in totality of knowledge, with consequently greater compartmentalisation, in the half century since cybernetics emerged.

In the preface, the range of topics covered is indicated by the comment: “This general theme of combining systems has attracted attention from across many disciplines, including applied and theoretical computing, logic, artificial intelligence, and natural language processing”.

The sessions of the meeting had the respective headings: “Theorem proving and rewriting” (six papers), “Logics” (eight papers, one not included in the book), “Systems” (three papers) and “Constraints” (four papers), but the book contents are not subdivided.

All of the papers either show how to combine apparently distinct approaches to their particular area of concern, or else present a general scheme allowing such combinations. Several of the papers refer to formalised approaches to automated deduction, under the heading of constraint problem solving, and others refer to systems of logic (temporal, modal and others). In most of these papers a considerable amount of background understanding is required in order to appreciate the features that had to be reconciled between different systems. Nevertheless, it is apparent that important contributions are made.

One paper in the area of automated deduction refers to the combination of systematic algorithms depending on tree‐searching with stochastic methods that can be represented as hill climbing.

The first paper in the book, mentioned as having been presented in the session on “Systems”, describes a programming language combining features of imperative (or procedural) and declarative programming. Programs appear at first sight to be procedural, until it is realised that the conditions for loops require traversal for all possible values satisfying a particular logical condition. It would be easy to write a purely‐procedural program that would be functionally equivalent, depending on generating all possible values or combinations of values and then testing each for conformity to the logical condition, but the run time would be prohibitive. The incorporation of declarative (or logic) programming principles allows this to be overcome (though details of the inner working of the compiler are not given) and an application to an intriguing numerical problem is shown.

Several papers refer to the combining of different representations of logic or of mathematics. One refers to the sharing of libraries of mathematics between theorem provers. It appears that digital libraries of mathematics have been set up with differing sets of conventions that are not easily reconciled. Means of reconciling two such libraries are given, depending on use of theorem‐proving methods in the transfers between them. The difficulties are such that it is not recommended that all possible pairings of such libraries be similarly treated. A more feasible solution is to reduce each of them, as required, to a basic logical form in which they are all compatible.

Two papers refer explicitly to natural‐language processing. One treats features of English grammar under the heading of “aspectual and temporal phenomena”, with introduction of a formal language for their analysis. Although the point is not specifically made, this illustrates the fact that English grammar is extremely complex when analysed in detail, even though non‐specialists usually only come across verb aspects as a feature of Russian. The other paper is on the important matter of combining syntactic and semantic theories and information in reasoning.

The topics range widely within their essentially computational context, and there is much valuable material here.