============================== VOLUME 17 NUMBER 2 AUGUST 1993 ============================== 1. ON A QUANTUM--STATISTICAL THEORY OF PAIR INTERACTION BETWEEN MEMORY TRACES IN THE BRAIN Ji\v{r}\'{\i} \v{S}lechta, Member of the New York Academy of Sciences, 18 Lidgett Hill, Leeds 8, LS8 1PE, U.K. pp. 109-115 keywords: brain, information, informational contents (coupling, difference, quantum), memory traces, pair interaction, quantum statistical theory abstract: A quantum--statistical theory of pair interaction between memory traces (MTs) in the brain is presented and the basic formulas for its strength are derived. It is shown that the interaction between two memory traces is proportional to the size of the contents of the (free) pieces of information (FPIs) exchanged between them and the number of such exchanges during the given period. The Green function of the propagation of both MTs and the quanta of FPIs exchanged between two MTs, and their properties, are introduced and studied by means of an elementary Feynman technique. It is shown for example, that the `blown up' brain cells found in the brains of people suffering from schizophrenia may be caused by a resonance interaction with them within a ring of MTs (a laser of them) storing a piece of too-simple (crystal-like) information. ----------------------- 2. INTEGRATIVE DOMAIN ANALYSIS VIA MULTIPLE PERCEPTIONS Wilhelm Rossak and Tamar Zemel, Systems Integration Laboratory, Department of Computer and Information Science, New Jersey Institute of Technology, University Heights, Newark, NJ 07102, USA rossak@pluto.njit.edu, Phone: (201) 596-659 pp. 117-137 keywords: domain analysis, system integration, software engineering abstract: Domain analysis is proposed as an essential activity for the integrated development of large, complex systems within an application domain. As an extension of traditional system-analysis methods, it is used as a means to assure global, inter-project coordination. Domain analysis provides a universal, comprehensive, and non-constructive domain model. This domain model is used as a common basis for understanding by all developers in the domain and as essential input for the requirements specification phase in each project. Since an application domain is perceived differently by the many entities who have different relations to that domain, we propose building the domain model as an integration of these perceptions. Each perception represents the phenomena of the domain from the viewpoint of a specific group of users, managers, customers, or authorities. To facilitate this type of domain modeling, we propose using a domain modeling schema (domain schema) that consists of pre-specified element-types (modeling primitives) for the domain. This domain schema can be specialized and adapted to support capturing different perceptions and (re)integration of all perceptions into one comprehensive domain model. The proposed approach generalizes and extends existing system analysis methods and is compatible with object-oriented concepts. ----------------------- 3. A PROLOG-BASED REPRESENTATION FOR INTEGRATING KNOWLEDGE AND DATA Xindong Wu, Department of Artificial Intelligence, University of Edinburgh, 80 South Bridge, Edinburgh EH1 1HN, U.K., Address after 14 July 1993: (xindong$@$coral.cs.jcu.edu.au) Department of Computer Science, James Cook University, Townsville, QLD 4811, Australia. pp. 137-144 keywords: knowledge representation, Prolog, deductive databases, semantic information abstract: Although the history of data base systems research is one of exceptional productivity and startling economic impact, many advanced applications have revealed deficiencies of the conventional data base management systems (DBMS's) in representing and processing complex objects and knowledge. Object-oriented approaches are currently very popular in processing structurally complex objects, while deductive data bases or logic data bases have been proposed as a solution to those applications where both knowledge and data models are needed. However, it has been characteristic of the current deductive data bases that only actual data is represented explicitly in logic, while the data schema is implicitly described in form of predicates. In this paper, we present a Prolog-based representation. It binds the actual data and data schema together in a natural and flexible way. In addition to expressing all the information which can be represented in the entity-relationship (E-R) model, the representation can represent other kinds of semantic information as well. ----------------------- 4. WALKING VIABILITY AND GAIT SYNTHESIS FOR A NOVEL CLASS OF DYNAMICALLY-SIMPLE BIPEDS Jon Kieffer and Ramesh Bale, Interdisciplinary Engineering Program, Australian National University pp. 145-155 keywords: bipeds, equations, simple dynamics abstract: This paper introduces a class of three-link, two-motor, planar bipeds that have mass centers invariantly-fixed at the hip axis and bodies that serve as reaction wheels. The principle advantage of these bipeds is that they are governed by exceptionally simple dynamic equations. This paper derives the governing equations for single-leg support and support leg transfer as well as step-to-step boundary conditions for periodic walking. Closed-form periodic gait trajectories are synthesized which ensure that the body's spin does not build up in the course of periodic walking. Examples show that a realistic model can walk on both flat and inclined surfaces. ----------------------- 5. MODELLING BIODEGRADATION BY AN EXAMPLE-BASED LEARNING SYSTEM Dragan Gamberger, Sanja Seku\v{s}ak, Aleksandar Sablji\'{c}, Rudjer Bo\v{s}kovi\'{c} Insitute, P.O.B.1016, 41001 Zagreb, Croatia pp. 157-166 keywords: inductive learning, biodegradation abstract: In this paper a novel rule-generation system for learning from examples and its application for modelling biodegradation of chemicals are presented. Two rules for biodegradation prediction are generated: the first one for all binary descriptors and a learning set of 48 examples, and the second one with some descriptors extended to integer and floating-point values and a learning set of 160 examples. The results of prediction of test examples by the generated rules are compared with the measured values and the results of two known models: classical fitting model, based on molecular connectivity indices, and a neural network model. Besides good prediction results, the generated rules have the unique characteristic of pointing out some logical dependencies that might influence the better understanding of the biodegradation process. ----------------------- 6. SUCCESSIVE NAIVE BAYESIAN CLASSIFIER Igor Kononenko, University of Ljubljana, Faculty of Electrical and Computer Engineering, Tr\v za\v ska 25, 61001 Ljubljana, Slovenia e-mail: igor.kononenko@ninurta.fer.uni-lj.si pp. 167-174 keywords: naive Bayesian classifier, successive learning, non-linear problems, empirical learning, empirical evaluation abstract: The naive Bayesian classifier is fast and incremental, can deal with discrete and continuous attributes, has excellent performance in real-life problems and can explain its decisions as the sum of information gains. However, its naivety may result in poor performance in domains with strong dependencies among attributes. In this paper, the algorithm of the naive Bayesian classifier is applied successively enabling it to solve also non-linear problems while retaining all the advantages of naive Bayes. The comparison of performance in various domains confirms the advantages of successive learning and suggests its application to other learning algorithms. ----------------------- 7. MORAL HAZARD PROBLEM SOLVING BY MEANS OF PREFERENCE RANKING METHODS Ines Sara\v zin Lovre\v ci\v c, Health Care Institution of Slovenia, Miklo\v si\v ceva 24, 61000 Ljubljana, Slovenia AND Janez Grad, Department of Economics, University of Ljubljana, Kardeljeva pl. 17, 61000 Ljubljana, Slovenia pp. 175-182 keywords: moral hazard, preference ranking, pseudo-model abstract: Moral hazard problems in the field of humanitarian health aid delivery can be difficult to solve, especially in outstanding circumstances caused by human or natural factors. In this paper, we present a solution to this problem by means of preference-ranking methods. The idea of a pseudo-model is also included, where standard input is considered as well as subjective elements. ----------------------- 8. FIFTH GENERATION COMPUTER SYSTEMS (FGCS) PROJECT IN JAPAN (a technical paper) Koichi Furukawa, Faculty of Environmental Information, Keio University, 5322 Endo, Fujisawa-shi, Kanagawa, 252 Japan furukawa@icot.or.jp pp. 183-199 keywords: concurrent and constraint logic programming, fifth generation computer, follow--on project, forecasts, overview, parallel inference system, personal sequential inference machine abstract: In this article, we give a short overview of the FGCS project and describe the research and development of the sequential inference machine PSI. Then, we present our research results on constraint logic programming. Finally, we discuss our research activities in the field of parallel inference from both hardware and software aspects.