VOLUME 23 NUMBER 4 1999

Abstracts:

Information Society And The Intelligent Systems Generation

Matjaz Gams
Jozef Stefan Institute, Department of Intelligent Systems, Jamova 39, Ljubljana, Slovenia

In this overview paper we analyze basic laws and properties of the information society in general, and its introduction in Slovenia. It is claimed that information society initiated the emergence of primitive network intelligence demonstrated through intelligent assistants on the Internet. One of the key reasons for emergence of the new software generation is the growth of the Internet, and the other information overload.  The introduction of intelligent systems, and particularly intelligent agents in Slovenia is analyzed. Finally, the EMA employment agent, one of important intelligent agent applications in Central Europe, is described in detail. (pp. 449-454)

Keywords:information age, Internet, intelligent agents, overview paper, viewpoint

A New Perspective in Comparative Analysis of Information Society Indicators

Pavle Sicherl
Law School, University of Ljubljana and SICENTER, Brajnikova 19, 1000 Ljubljana, Slovenia

The analysis of information society indicators can be enriched by supplying a new view of data that can provide new insight from existing data. The slowdown of growth of Internet hosts per 10000 inhabitants in Slovenia after mid-1997 increased the time lag of Slovenia behind leading Finland from 3 years at the end of 1996 to nearly 5 years by August 1999. Time distance methodology is used as a presentation and communication tool to raise awareness of the problem and its consequences in simple understandable terms and to signal the need for an in-depth analysis and action. (pp. 455-460)

Keywords:time distance, S-distance, two-dimensional comparison in time and indicator space

Measuring Information Society: Some Methodological Problems

Vasja Vehovar and  Matej Kovacic
Faculty of Social Sciences, Kardeljeva ploscad 5, 1000 Ljubljana, Slovenia

The paper addresses methodological problem of measuring information society. Both, technical indicators and attitudinal measurements for Slovenia are discussed in this context. In particular, the results related to the interest for information society services are presented. The comparison between Slovenia and European Union -- despite some methodological problems -- shows that the interest for these services is extremely high in Slovenia. Other figures also confirm that Slovenian households and businesses are generally on European average with respect to the penetration of the basic information technologies. However, certain discrepancies with other sources of data call for more efforts in performing these kind of analysis. (pp. 461-465)

Keywords:information society, Internet indicators, survey research, electronic commerce

Modelling of an Information Society in Transition - Slovenia's Position in the CE Countries

Marjan Krisper and Tatjana Zrimec
Faculty of computer and information science, University of Ljubljana, Trzaska 25, 1000 Ljubljana, Slovenia

This paper presents modelling and visualisation of information society development in the six Central European countries associated to the European Union. Modern approach for monitoring and evaluation of the transition processes is presented that enables comparison of the countries' successfulness and analysis of alternative development scenarios. The position of Slovenia, which also experiences gradual transformation to the information society, is shown. (pp. 467-471)

Keywords: modelling of socieconomics data, visualisation, information society

Customer Satisfaction of Information System Integration Business in Japan

Kayo H. Iizuka
Graduate School of Systems Management, University of Tsukuba, 3-29-1, Otsuka, Bunkyo-ku, Tokyo Japan

When we consider about “customer satisfaction” (CS) of system integration (SI) business, “customer” means organization, though many of the cases, when we see the CS in some research papers, “customer” means personal consumer. Understanding the satisfaction structure of organization, being considering organization behavior, must be very important  Moreover SI business provides, not only products itself alone but integrated services and produCTS, In this paper, focusing on these matters, showing the structure of CS in SI business, from statistical analysis of customer survey.   It shows what is the important factor including technology providing, and project management skill, in order to maximizing CS. (pp. 473-476)

Keywords:customer satisfaction, system integration

An Infrastructure For Support Of Digital Signatures

Tomaz Klobucar and Borka Jerman-Blazic
Jozef Stefan Institute, Jamova 39, Ljubljana, Slovenia

In this paper, we present an infrastructure for support of digital signatures in the Information society. Technical aspects are briefly described and a short overview of several existing legal frameworks is given. Certification authorities, certificate policies, signature policies and certification practice statements are identified as important parts of the infrastructure. (pp. 477-482)

Keywords:digital signature, public-key infrastructure, technical framework, legal framework

Using An Electronic Book In Distance Education

Eva Jereb and Branislav Šmitek
Faculty of organisational sciences, University of Maribor, Kidriceva 55a, 4000 Kranj, Slovenia

In scope of project "Phare Multi - Country Programme for Distance Education" in years 1997/98 and 1998/99 we formed more learning material for the need of our new Distance learning centre. The reason for forming this kind of learning material was that we became aware that new forms and methods of distance learning have to assure the bigger independence of students in the whole learning process. For achieving these aims smaller groups of students with high individualisation are appropriate what brings increased level of activities in the whole learning process and enables periodic information and more realistic knowledge control. In this article we present some experience we had by by forming electronic book for subject Office automation and some students opinion about studying with help of electronic book. (pp. 483-486)

Keywords:distance education, electronic book, multimedia in education

Multi-Attribute Decision Modeling: Industrial Applications of DEX

Marko Bohanec
Jozef Stefan Institute, Jamova 39, SI-1000 Ljubljana, Slovenia

University of Maribor, Faculty of Organisational Sciences, Kidriceva 55, Kranj, Slovenia

DEX is an expert system shell for qualitative multi-attribute decision modeling and support. During the last decade, it has been applied over fifty times in complex real-world decision problems. In this article we advocate for the applicability and great potential of this approach for industrial decision-making. The approach is illustrated by a typical industrial application in land use planning, and supplemented by an overview of some other completed industrial applications. The learned lessons indicate the suitability of the qualitative DEX methodology particularly for "soft", i.e., less structured and less formalized, decision problems. Practical experience also indicates the importance of methods that facilitate the analysis, simulation, and explanation of decisions. (pp. 487-491)

Keywords:decision support, multi-attribute decision making, qualitative decision models

Perception-Based Classification

Mihael Ankerst, Christian Elsen, Martin Ester, Hans-Peter Kriegel
Institute for Computer Science, University of Munich, Oettingenstr. 67, D-80538 München, Germany

Classification is an important problem in the emerging field of data mining. Given a training database of records, each tagged with a class label, the goal of classification is to build a concise model that can be used to predict the class label of future, unlabeled records. A very popular class of classifiers are decision trees because they satisfy the basic requirements of accuracy and understandability. Instead of constructing the decision tree by a sophisticated algorithm, we introduce a fully interactive method based on a multidimen-sional visualization technique and appropriate interaction capabilities. Thus, domain knowl-edge of an expert can be profitably included in the tree construction phase. Furthermore, after the interactive construction of a decision tree, the user has a much deeper understanding of the data than just knowing the decision tree generated by an arbitrary algorithm. The inter-active approach also overcomes the limitation of most decision trees which are fixed to binary splits for numeric attributes and which do not allow to backtrack in the tree construction phase. Our performance evaluation with sev-eral well-known datasets demonstrates that even users with no a priori knowledge of the data construct a decision tree with an accuracy similar to the tree generated by state of the art algorithms. Additionally, visual interactive classification significantly reduces the tree size and improves the understandibility of the resulting decision tree. (pp. 493-499)

Keywords:classification, decision tree, data mining, visualization

Generalized Blockmodeling

Faculty of Mathematics and Physics, University of Ljubljana, Jadranska 19, 1000 Ljubljana, Slovenia

Anuska Ferligoj
Faculty of Social Sciences, University of Ljubljana, Kardeljeva pl. 5, 1000 Ljubljana, Slovenia

Patrick Doreian
Department of Sociology, University of Pittsburgh, PA 15260, Pittsburgh,  USA

The goal of blockmodeling is to reduce a large, potentially incoherent network to a smaller comprehensible structure that can be interpreted more readily. In the paper we present an overview of basic ideas and developments in this area. (pp. 501-506)

Keywords:clustering, blockmodeling, social network, pre-specified blockmodel, local optimization

Adapted Methods For Clustering Large Datasets Of Mixed Units

Simona Korenjak-Cerne
IMFM Ljubljana, Dept. of TCS, Jadranska 19, 1000 Ljubljana, Slovenia

Faculty of Mathematics and Physics, University of Ljubljana, Jadranska 19, 1000 Ljubljana, Slovenia

The proposed clustering methods are based on the recoding of the original mixed units   and their clusters into a uniform representation. The description of a cluster consists  for each variable of the frequencies of the variable values over its range partition. The proposed representation can be used also for clustering symbolic data. On the basis of this representation the adapted version of the leaders method and adding clustering method were implemented. We describe both approaches, which were successfully applied on several large datasets. (pp. 507-512)

Keywords:clustering, large datasets, mixed units, hierarchical clustering, cluster description compatible with merging of clusters, leaders method, adding clustering method

Equation Discovery System And Neural Networks For Short-Term Dc  Voltage Prediction

Irena Nancovska, Anton Jeglic and Dusan Fefer
Faculty of Electrical Engineering,Trzaska 25, Ljubljana, Slovenia

Ljupco Todorovski
Jozef Stefan Institute, Jamova 39, Ljubljana, Slovenia

The aim of the paper is to compare the predictive abilities of the novel method for time series prediction that is based on equation discovery with neural networks. Both methods are used for short-term (one-step ahead) prediction and have the ability to learn from examples. With purpose to validate the predictive models, they are applied to several data sets. The successful predictive models could be used for voltage monitoring in a high precision solid-state DC voltage reference source (DCVRS) without presence of a high level standard, and further for voltage correction as a segment in the software controlled voltage reference elements (VRE). (pp. 513-520)

Keywords:neural networks, equation discovery, machine learning

Adaptive On-line ANN Learning Algorithm and Application to Identification of Non-linear Systems

Daohang Sha and Vladimir B. Bajic
Centre for Engineering Research, Technikon Natal, P.O.Box 953, Durban 4000, South Africa

A new on-line adaptive learning rate algorithm for I/O identification based on two ANNs is proposed. The algorithm is derived from the convergence analysis of the conventional gradient descent method. Simulation experiments are given to illustrate the advantages of the proposed algorithm in its application to an identification problem of some non-linear dynamic systems. (pp. 521-530)

Keywords:Soft computing, neural networks, gradient descent method, real-time algorithms, Input/Output modelling

A Spanish Interface To LogiMoo: Towards Multilingual Virtual Worlds

Veronica Dahl, Stephen Rochefort and Marius Scurtescu
School of Computing Science, Simon Fraser University, Burnaby, BC, Canada V5A 1S6

Paul Tarau
Department of Computing Science, University of Moncton, Moncton, NB, Canada E1A 3E9

LogiMOO is a BinProlog-based Virtual World running under Netscape for distributed group-work over the Internet and user-crafted virtual places, virtual objects and agents.  LogiMOO is mplemented on top of a multi-threaded blackboard-based logic programming system featuring Linda-style coordination.Embedding in Netscape allows advanced VRML and HTML frame-based navigation and multi-media support, while LogiMOO handles virtual presence and acts as a very high-level universal object broker. In this talk we shall briefly describe Assumption Grammars (the logic grammar toolused in our Spanish interface to LogiMOO) and how they can help solve some crucial computational linguistic problems such as anaphora and coordination. We shall then discuss our translator from Spanish sentences into LogiMoo commands. Finally, we shall discuss what is needed to parameterize a single language processor into specific natural languages,  with the ultimate objective of  transforming LogiMoo into a multilingual virtual world. In it users from various linguistic backgrounds could communicate using their own language, to be automatically translated into LogiMoo as universal interlingua. (pp. 531-542)

Keywords:virtual worlds, Internet applications, natural language processing, Assumption Grammars, LogiMOO

Efficient Computation Of Frequent Itemsets In A Subcollection Of Multiple Set Families

Hong Shen
School of Computing and Information Technology, Griffith University, Nathan, QLD 4111, Australia

Weifa Liang
Department of Computer Science, Australia National University, Canberra, ACT 2600, Australia

Joseph Ng
Department of Computer Science, Hong Kong Baptist University, Kowloon, Hong Kong

Many applications need to deal with the additive and multiplicative subcollections over a group of set families (databases). This paper presents two efficient algorithms for computing the frequent itemsets in these two types of subcollections respectively. Let T be a given subcollection of set families of total size m whose elements are drawn from a domain of size n. We show that if T is an additive subcollection we can compute all frequent itemsets in T in $O(m2n/(pn)+\log p)$ time on an EREW PRAM with $1\leq p\leq m2^n/n$ processors, at  a cost of maintaining the occurrences of all itemsets in each individual set family. If $T$ is a multiplicative subcollection, we  can compute all itemsets in $T$ in $O(mk/p+\min\{\frac{m'}{p}2^n,n3^n\log m'/p\})$ time on an EREW PRAM with $1\leq p\leq \min\{m,2^n\}$ processors, where $m'=\min\{m,2^n\}$. These present improvements over direct computation of the frequent itemsets on the subcollection concerned. (pp. 543-548)

Keywords:Algorithm, data mining, frequent itemset

SISTER: A Flexible System For Image Retrieval

Monica Mordonini and Agostino Poggi
Dipartimento di Ingegneria dell'Informazione, University of Parma, Parco Area  delle Scienze, 181/A, 43100-Parma, Italy

SISTER is a system for the storing and retrieval of large collections of images on the basis of both textual information and the image content. In particular, SISTER allows the user to formulate queries for different image categories on the basis of color information and specific attributes of the image category; this is possible, because such a system can be easily adapted to support new image categories specializing the acquisition and the retrieval subsystems. SISTER is composed of three parts: i) an image acquisition subsystem automatically extracting the attributes from images, ii) a database management subsystem maintaining the description of images, and iii) a retrieval subsystem allowing the user to retrieve images through a user-friendly graphical interface. The acquisition subsystem extracts image attributes combining image processing and inductive classification modules. Classification modules allow the computation of image attributes that cannot be directly extracted through image processing techniques, but whose value derives from the value of some other image attributes extracted by the acquisition subsystem. (pp. 549-558)

Keywords:image retrieval by content, pictorial information, inductive classifier, pattern matching, image database management system

Tuning of Fuzzy Logic Controller with Genetic Algorithm

Borut Zupancic, Marko Klopcic and Rihard Karba
University of Ljubljana, Faculty of Electrical Engineering, Trzaska 25, 1000 Ljubljana, Slovenia

The paper deals with the fuzzy logic controller (FLC) tuning by the aid of the optimization with genetic algorithm (GA). Because of the complexity of calculations the Sugeno 0th order FLC was used. The GA optimization tuned 25 consequent parameters while the membership functions remained fixed. Because of large number of parameters traditional optimization methods were not successful. Beside appropriately selected GA parameters the choice of appropriate reference signal of the control system was extremely important. Namely it must be selected so that the whole ranges of both FLC inputs are used. It is recommended to plot trajectory of FLC inputs  to see which parts of the truth table is appropriately covered by the inputs and to find out which consequent parameters can not be optimized. Filtering of FLC characteristic is another useful method, which makes output characteristics smoother and so improves responses. The efficiency of the proposed approach were verified and
validated on a hydraulic control system. (pp. 559-564)

Keywords:keywords{control system, fuzzy logic controller, optimization, genetic algorithm

PNNI And The Optimal Design Of High-Speed ATM Networks

Abdella Battou
Center for Computational Science, U.S. Naval Research Laboratory. Washington D.C., USA

Bilal Khan and Sean Mountcastle
ITT Industries Systems Division at the Center for Computational Science, U.S. Naval Research Laboratory. Washington D.C, USA

In addition to being well-dimensioned and cost-effective, a high-speed ATM network must pass some performance and robustness tests. We propose an approach to ATM network topology design that is driven by the performance of its routing protocol, PNNI.  Towards this end, we define performance indicators based on the time and traffic required for the protocol to first enter and subsequently return to the meta-stable state of {\em global synchrony}, in which switch views are in concordance with physical reality.  We argue that the benefits of high call admission rate and low setup latency are guaranteed by our indicators. We use the PNNI Routing and Simulation Toolkit (PRouST), to conduct simulations of PNNI networks, with the aim of discovering how topological characteristics such as the diameter, representation size, and geodesic structure of a network affect its performance. (pp. 565-574)

Keywords:ATM Network Design, PNNI Simulation, PRouST