Journal:Informatica
Volume 8, Issue 1 (1997), pp. 153–175
Abstract
The aim of program analysis and visualisation (PA&V) is to help the programmer understand a program by means of graphical presentations of different aspects of the program. Program analysis and visualisation systems can be classified according to the specification method of visualisation, e.g., in what way can the user of the system specify his own visualisers. In the article three specification methods (predefinition, annotation and declaration) are discussed and some example systems are presented. Particular attention is paid to the declarative specification method, thus, in addition, knowledge-based program analysers are discussed. Increased understandability and modifiability are argued to be the main advantages of declarative PA&V systems.
The general discussion is continued by a short presentation of a case study, where the declarative and synthesisable visualisation in the NUT system is discussed.
Journal:Informatica
Volume 8, Issue 1 (1997), pp. 139–152
Abstract
ProObj is a Prolog based system for knowledge representation which was strongly influenced by object-oriented and frame-based systems. The paper shortly describes ProObj and then presents a classification mechanism which is based on the ideas of classifiers in KL-ONE like systems.
As a new and very flexible feature we present a user-directed control of classification process. The ProObj classifier gives the user the possibility to guide the classification process by excluding attributes and facets – elements of our representation formalism – from being considered in the classification. By this mechanism we gain a substantial improvement of the efficiency of the classification process. Furthermore, it allows a more flexible and adequate modelling of a knowledge domain. It is possible to build a knowledge base under a particular view where only those attributes of concepts are considered for classification which seem to be relevant for the structure of the domain hierarchy.
Journal:Informatica
Volume 8, Issue 1 (1997), pp. 119–138
Abstract
An overview and comparison of mobile agent systems are presented. The rapidly evolving area of software agents is briefly overviewed. The notion of mobility is analyzed in the context of mobile code languages, and its relation to distributed computing (e.g., client–server model), as well as its possible application areas are studied. Finally the need for a combination of mobility with other features is discussed.
Journal:Informatica
Volume 8, Issue 1 (1997), pp. 83–118
Abstract
The problem is to discover knowledge in the correlation matrix of parameters (variables) about their groups. Results that deal with deterministic approaches of parameter clustering on the basis of their correlation matrix are reviewed and extended. The conclusions on both theoretical and experimental investigations of various deterministic strategies in solving the problem of extremal parameter grouping are presented. The possibility of finding the optimal number of clusters is considered. The transformation of a general clustering problem into the clustering on the sphere and the relation between clustering of parameters on the basis of their correlation matrix and clustering of vectors (objects, cases) of an n-dimensional unit sphere are analysed.
Journal:Informatica
Volume 8, Issue 1 (1997), pp. 57–82
Abstract
Software system engineering has not yet developed an engineering science for its discipline. On the other hand, a lot of fundamental concepts, shared methods, techniques, patterns for structuring software systems, and languages for documenting design decisions has been accumulated over the years. To analyse and systematise the accumulated ideas is the main challenge for computer scientists today. The main objective of this paper is to analyse software system engineering both as a discipline and as an engineering science. A special attention is paid to conceptual modelling formalisms used in software system engineering.
Journal:Informatica
Volume 8, Issue 1 (1997), pp. 43–56
Abstract
The paper present a proposed approach in the context-free language theory. The main new notion is a graph defining a pushdown automaton (PDA). Each vertex of such graph is a pair (state, stack symbol). Each edge corresponds to a “command” and is labelled by input portion being read by the command and by a “charge” describing the stack word transformation. Some paths of the graph represent PDA's computations. The finite automata are a case of the pushdown graphs. The paper contains some of the author's results based on the approach – the notion of a D-language extending the notion of Dyck's language and the theorem on a representation of a context-free language as a morphical image of the intersection of a D-language with a local set.
Journal:Informatica
Volume 8, Issue 1 (1997), pp. 23–42
Abstract
Today's multimedia and hypermedia systems include such a huge amount of data and links, that they should be stored and maintained by a database system. Then a powerful and efficient database schema is needed. The Dexter hypertext reference model offers a widely accepted, powerful modelling technique for nodes and links. We present its stepwise conversion into a relational multimedia database schema. In the obtained hypermedia engine the most important and most time critical operation is the link navigation. We analyze its complexity in detail and optimize it by schema improvements. Finally we present an efficient implementation of the presented ideas: the System MultiMAP, developed at the TU Munich.
Journal:Informatica
Volume 8, Issue 1 (1997), pp. 3–22
Abstract
The paper introduces to plan generation and the attempts and problems encountered with solutions by deductive methods. It presents the Linear Connection Method as a possibility to overcome the traditional shortcomings of logic for this application, which are discussed in great detail. This is followed by a guide to the research work carried out on the Linear Connection Method and related frameworks.
Journal:Informatica
Volume 7, Issue 4 (1996), pp. 525–541
Abstract
In the present paper, the method of structure analysis for multivariate functions was applied to rational approximation in classification problems. Then the pattern recognition and generalisation ability was investigated experimentally in numerical recognition. A comparison with Hopfield Net was carried out. The overall results of using of new approach may be treated as a success.