Abstracts:

Trust is an important ingredient for fostering the development of electronic commerce on the Internet. Recent surveys report on the growth of fraudulent activities on the Internet. We report the results of such an investigation which checked for the amount of fraudulent activities on auction sites. The result shows a significant level of such activities, well above the levels reported by auction site operators. Since fraud can be a major stumbling block for increasing the volume of electronic commerce on the web, it motivated research on trust building activities given the prevalence of fraudulent activities on the web. (pp. 377-383)

Keywords: electronic commerce, fraud, trust

E-commerce sites often recommend products they believe a customer is interested in buying. Many web sites have started to embody recommender systems as a way of personalizing their content for users. This paper presents a recommender system that exploits supervised learning methods to learn user profiles from items previously rated by users. Profiles are used to find, classify, or rank items that are likely to be of interest to the user. A major concern with supervised learning techniques is that they often require a large number of labelled examples to learn accurately. Our proposal to reduce the amount of labelled data required is an algorithm that can learn effectively from a small number of labelled examples augmented with a large number of unlabelled examples. Experiments on a real dataset show that the proposed method is effective. (pp. 385-389)

Keywords: personalization, user profiling, intelligent search and retrieval, Bayesian learning, learning from labelled and unlabelled

Andreas Schmid and Tong-Seng Quah
Information Communication Institute
School of EEE, Nanyang Technological University
50 Nanyang Avenue
SG - 639798 Singapore, Republic of Singapore
E-mail: andreas.schmid@ieee.org

Our project proposes and implements a system for seamless integration of HP E-speak E-services and IBM Aglet mobile agent applications onto a common platform. The developed software enables all applications written in Aglets and E-speak to create joint services across the two platforms. It establishes an environment for collaboration, intercommunication, translation, and bridging. The whole design is thereby highly reusable and portable. This paper also analyses the requirements of E-Commerce and elaborates how effectively a combination of Aglets and E-speak can provide the necessary resources. We suggest a common model for Aglets, E-speak, and traditional Internet components that enhances performance through synergies. The integration combines the advantages of each technology in its respective domains (pp. 391-398)

Cene Bavec
University of Primorska
Faculty of manegment
Cankarjeva 5
PP 345
SI-6104 Koper, Slovenia
E-mail: cene.bavec@guest.arnes.si

Maja Bučar and Metka Stare
Center of International Relations
University of Ljubljana
Faculty of Social Sciences
Kardeljeva ploščad 5
SI-1000 Ljubjana, Slovenia
E-mail: maja.bucar@guest.arnes.si, metka.stare@guest.arnes.si

Until recently, discussion of the ICT impact has been focused on very few issues, predominantly economic growth and productivity. It is becoming increasingly evident that impact is much wider and affects all spheres of economic and social life. The evidence from OECD member countries is reaffirming the importance of ICT as a driver of growth at all levels: at the level of national economy, at the sectoral level and at the level of individual companies. Recognizing different dimensions of ICT and its interlinkages among different sectors is important to fully asses its potential in national economy and individual company. Many argue that current ICT indicators often deal with less important issues and miss to address “softer” views of digital economy like ICT impact on product or service quality, organizational change, quality of human capital, level of globalization, or government policies. In the paper we discuss and asses statistical indicators used at different levels of decision making – from national to companies' levels. We attempt to verify their relevance and usefulness for various purposes. The case of Slovenia is presented to illustrate interpretations and assess presently used indicators. (pp. 399-404)

Martin Žnidarsic
Department of Intelligent Systems
Jožef Stefan Institute
Jamova 39
SI-1000 Ljubljana, Slovenia
E-mail: martin.znidarsic@ijs.si

Marko Bohanec
Department of Intelligent Systems
Jožef Stefan Institute
Jamova 39
SI-1000 Ljubljana, Slovenia
and
Faculty of Administration
University of Ljubljana
Gosarjeva 5
SI-1000 Ljubljana
E-mail: marko.bohanec@ijs.si

Ivan Bratko
Department of Intelligent Systems
Jožef Stefan Institute
Jamova 39
SI-1000 Ljubljana, Slovenia
and
Faculty of Computer and Information Science
University of Ljubljana
Tržaška 25
SI-1000 Ljubljana
E-mail: ivan.bratko@fri.uni-lj.si

DEX is a multi-attribute decision modelling methodology. Its specialty is the use of ordinal data and qualitative utility functions. Numerical attributes must be therefore categorized before use in DEX models. We present the problem of numerical data categorization and propose two methods which simplify and partly automate this task. The methods suggest interval bounds according to the desired number of categories and the preference curve of attribute values. We implemented both methods and made some experiments with typical inputs. The most interesting results are presented and analysed. (pp. 405-409)

Iztok Fister
Mura, European Fashion Design
Plese 2
SI-9000 Murska Sobota, Slovenia
E-mail: iztok.fister@mura.si

Marjan Mernik
Faculty of Electrical Engineering and Computer Science
University of Maribor
Smetanova 17
SI-2000 Maribor, Slovenia
E-mail: marjan.mernik@uni-mb.si

Optimization of markers plays an important role in preparation of order-based industrial production of clothes. Given a matrix of pieces in size numbers and designs, the task is to find a list of combinations of size numbers to accomplish a work order. The outcomeof this step influences the number of cut out pieces, the amount of material used in the production phase, and the speed of the work order processing. As numerous factors affect the production costs and several conflicting criteria can be involved in marker assessment, marker optimization is a demanding task.In this paper, minimum number of markers per work order is used as an optimization criterion. Marker optimization is formally defined as a knapsac problem, and an evolutionary algorithm is proposed to solve the task. It is tested on real problem instances from industrial clothes production and compared with several other algorithms. Its results on complex work orders are shown to be superior to those of other tested algorithms. (pp. 411-415)

Andraz Bezek and Matjaz Gams
Department of Intelligent Systems
Jožef Stefan Institute
Jamova 39
SI-1000 Ljubljana, Slovenia
E-mail: andraz.bezek@ijs.si, matjaz.gams@ijs.si

We analyze fault tolerance in load balancing systems. First, we theoretically analyze some aspects of fault-tolerance of traditional and multi-agent approaches. Second, we investigate efficient fault detection showing that multi-agent systems can increase fault-tolerance by improving fault detection of failed single agents. The presented ideas are applied in a multi-agent system for fault-tolerant network load balancing. A detailed description of fault-tolerant issues in design is also given. Finally, the paper presents evaluation of a multi-agent application for network load balancing. (pp. 417-424)

Keywords: fault-tolerance, multi-agent systems

Aleš Živkovič, Marjan Heričko and Tomaž Kralj
Faculty of Electrical Engineering and Computer Science
University of Maribor
Smetanova 17
SI-2000 Maribor, Slovenia
E-mail: ales.zivkovic@uni-mb.si
WWW: http://lisa.uni-mb.si

In the software industry, many projects fail due to both the misjudgment of a project's size and faulty estimates correlated to this elementary metric. Several methods for software size estimation are present. The Function Points Analysis (FPA) method, however, is most frequently put into practice. After Albrecht introduced the FPA method, several variations evolved. All methods share the same fundamental idea, but differ in procedural steps and metric units. A descriptive approach is usually used for method comparison. To avoid the weaknesses of a descriptive approach, a mathematical model is defined and used for theoretical comparison. The complexity of the mapping functions prevent detailed comparisons -- consequently only general characteristics become evident. Characteristics exposed with a formalization of the rules were further studied in different test scenarios using historical data from past projects. Empirical results showed some limitations of the mapping function and anomalies in the data set used. The possible reasons for deviations in the data set were also analyzed. (pp. 425-432)

Keywords: software metrics, function points, software size estimation, empirical analysis

N. Raja and R.K. Shyamasundar
School of Technology & Computer Science
Tata Institute of Fundamental Research
Homi Bhabha Road
IN-Mumbai 400 005, India
E-mail: raja@tifr.res.in, shyam@tifr.res.in

We explore the role of {\em types} in models of concurrent computation, particularly in the concrete setting of the asynchronous $\pi$-calculus. The major theme of this work may be summarized by the slogan - ``Wherever you see structure, think of types''. We propose type annotations not merely to channels, but also to the highly structured set of processes. The type system guarantees that well typed expressions cannot go wrong. Polymorphic process types formalize extant informal ideas regarding the channel passing and process passing approaches to process mobility. Further, subtyping relation between process types distinguishes between true concurrency and nondeterministic choice. (pp. 433-443)

Keywords: type theory, concurrency, process calculi

Milan Ojsteršek and Aleksander Kvas
Faculty of Electrical Engineering and Computer Science
University of Maribor
Smetanova 17
SI-2000 Maribor, Slovenia
E-mail:ojstersek@uni-mb.si

An efficient scheduling of a parallel program onto the processors is critical for achieving a high performance from a parallel computer system. The scheduling problem is known to be NP-hard and heuristic algorithms have been proposed to obtain optimal and suboptimal solutions. The partitioning algorithm partitions an application into tasks with appropriate grain size and represents them in the form of a directed acyclic graph (DAG). The nodes of the resulting DAG are then scheduled onto the processors of a parallel computer system. We can see that almost all coarse grained program graph nodes don't need all of their input operands at the beginning of their execution. Thereafter they can be scheduled a bit earlier. This type of program graph nodes triggering is called partial strict triggering. The missing operands will be requested later during the execution. Coarse grained program graphnodes send their output operand to all successors, as soon as they produce them. Successors of coarse grained program graph nodes will be scheduled earlier too, because they will receive their input operands sooner. An evaluation of improved CPM, VL and DSH scheduling algorithms is done in this paper. We have improved them with partial strict triggering of coarse grained program grap nodes. (pp. 445-450)

Keywords: parallel computers, program graphs, scheduling

Chuan Zhang
TransACT Communications
470 Northbourne Avenue
AU - Dickson ACT 2602, Australia
E-mail: chuan.zhang@transact.com.au

Jian Yang
Infolab
Tilburg University
PO Box 90153
NL - 5000 LE Tilburg, The Netherlands
E-mail: jian@uvt.nl

Materialized view selection is one of the important areas in data warehouse design. Due to the dynamic nature of data warehouse, materialized views should evolve in response to the application requirement change, which is studied as a dynamic materialized view selection problem. One aspect of dynamic materialized view selection that differs from static materialized view selection is that the existing materialized views should be taken into consideration when selecting new materialized views. In this paper, a framework is presented to illustrate the design methodology for dynamic materialized view selection. A new cost - the materialized view reorganization cost is introduced and incorporated when selecting new views to materialize. Based on our cost model, a set of algorithms for dynamic materialized view selection are developed and some experimental results are provided.(pp. 451-460)

Keywords:materialized views, dynamic view selection, data warehousing

Mäkinen Erkki
Department of Computer and Information Sciences
University of Tampere
Kehruukoulunkatu 1
FI - 33014 Tampere, Finland
E-mail: em@cs.uta.fi

Tarja Systä
Software System Laboratory
Tampere University of Technology
P.O. Box 553
FI - 33101 Tampere, Finland
E-mail: tsysta@cs.tut.fi

Practical applicability of grammatical inference is considered by studying its use for various tasks and modeling problems in software engineering. It is found out that interactive inference methods are potential for ``design-by-example'' tasks. The required user involvement can be used to avoid undesired results. It also helps the user to recognize the key problems in the application domain. Automatic inference algorithms based on identification in the limit, in turn, are suitable for ``comprehension tasks'' like reverse engineering existing software systems, where the constructed model represents an overview of the systems to be studied rather than an explicit design to be aimed at. (pp. 461-467)

Keywords: software engineering, grammatical inference, design

Alpar Fancsali
Dep. Of Telecommunications
BME
Magyar tudósok körútja 2
HU - 1117 Budapest, Hungary
E-mail: alpar@hit.hit.bme.hu

This paper investigates QoS routing in IP networks. The major concern is the real-time selection of paths to fulfil distinct quality of service requirements (such as end-to-end delay or bandwidth requirements). In practice one has to tackle routing with inaccurate information when link measures are subject to random fluctuations described by some given p.d.f.-s. Our goal is to identify a path that is most likely to successfully accommodate the desired QoS parameter. Unfortunately, this task is generally NP-hard. In the paper this problem is attacked by an abstract algebraic extension of the Bellman-Ford shortest path selection algorithm. The paper demonstrates that this extended algorithm is still tractable and gives a very good approximation for the optimal solution of the original NP-hard task.(pp. 469-481)

Keywords: QoS routing, Bellman-Ford algorithm, abstract algebra

Anton P. Železnikar
Volaričeva 8
SI - 1111 Ljubljana, Slovenia
E-mail: s51em@hamradio.si

This paper deals with the constitution and organization of a conscious entity. Informon, as an ordered and named conscious entity, informs in its equally named and informationally disordered (chaotic, irregular) environment, called entropon. Both informon and entropon overlap informationally more or less with other informons and entropons constituting a local or global informational space. A conscious system, as known from the naturalistic point of view, is necessarily a cognitive-emotional system together with other components concerning sensuality, attention, motivation. homeostasis, behavior, motorics, etc., all of them possessing a specific intention (orientation, goal-directedness, decision-making specificity) within the conscious informing. Concrete cognitive, emotional, and otherwise meaningly organized entities come into the conscious foreground. Artificial conscious systems can differ substantially from the natural ones in the functional domain, regarding the future complexity, memory, global interconnection, and informational search possibilities. Artificial entropons can serve as absolute memories for experiences (components, parts of components) and, in this way, can enable informons to use the entire precisely, regularly, and disorderly structured previous experience and their lumps.(pp. 483-494)

Keywords: attention, cognitive-emotional paradigm, complexity, emergentism, entity name, entropon, entroponic, entroponizing, informon, informonic, informonizing, intention, metaphysicalism, phenomenalism, the metatheory of the conscious, the philosophy of the informational

Peter Trkman and Miro Gradišar
Faculty of Economics
Kardeljeva ploščad 17
SI - 1000 Ljubljana, Slovenia
E-mail: peter.trkman@ef.uni-lj.si, miro.gradisar@ef.uni-lj.si

The paper describes exact solution of general one-dimensional cutting stock problem (G1D-CSP) where all stock lengths are different. Branch & Bound (B&B) optimization method is used. The solution is cutting plan with minimized overall trim loss in such a way that order lengths are cut in exactly required number of pieces and only one stock length is in general not cut to the end. If it's long enough then it can be used later and is not treated as a waste. G1D-CSP can also be solved approximately with Sequential Heuristic Procedure (SHP). Comparison between B&B and SHP is presented. It is shown that exact solution is better when the size of the problem does not exceed certain limit. The question is, how to determine this limit, which can be different in different practical situations. An approach, based on decision trees, for the selection of appropriate method for each individual case, is proposed. Numerous examples are calculated(pp. 495-501)

Keywords: cutting, optimisation, branch & bound, decision tree