Site hosted by Angelfire.com: Build your free website today!
The CommonKADS library in perspective
        Auteurs
                Valente A, Breuker J, VandeVelde W
         Source
                INTERNATIONAL JOURNAL OF HUMAN-COMPUTER STUDIES
                Volume 49, no. 4, OCT
          Pages
                391 - 416
    Semaine CC
                9849
           Date
                1998
        Résumé
                The CommonKADS Expertise Modeling Library is the result of work that started as a further
                development of the KADS Library of interpretation models (Breuker et al., 1987). It incorporates new
                views and experiences of the knowledge acquisition community on the reuse of problem-solving
                components, in particular problem-solving methods and their assumptions on domain knowledge. Its
                design allows to store and index an extensible set of modeling components of different levels of granularity,
                as well as generic (partial) models of expertise and even reusable modeling steps. Further, its contents
                were significantly expanded, in what is probably the largest library of problem solving methods presently.
                In this article, we explain the rationale of the design of the CommonKADS library, the mechanisms defined
                to index the elements it stores, and the Library contents. An example of the use of a part of the Library
                concerned with problem-solving methods for assessment tasks is presented. The Library is not a finished
                product: not only because still many contents may need further verification and validation, but also because
                the Library is intended to accumulate and share practical experiences and advice on the process of
                modeling for actual applications. These and other shortcomings and lessons are discussed. (C) 1998
                Academic Press.
      Mots-clés
                model construction, systems
       Citations
                56
          ISSN
                1071-5819
        Adresse
     pour copies
                Valente, A
                UNIV SO CALIF
                INST INFORMAT SCI, 4676 ADMIRALTY WAY
                MARINA DEL REY, CA, USA
                90292 AP

 

   Preserving conceptual structures in design and implementation of industrial KBSs
        Auteurs
                Speel PH, Aben M
         Source
                INTERNATIONAL JOURNAL OF HUMAN-COMPUTER STUDIES
                Volume 49, no. 4, OCT
          Pages
                547 - 575
    Semaine CC
                9849
           Date
                1998
        Résumé
                Applying the best available knowledge at the right place and the right time is crucial for industries like
                Unilever. As one approach to careful knowledge management, we are developing knowledge-based
                systems (KBSs) to capture and exploit key knowledge. For this purpose, we have adopted and tailored
                the CommonKADS method as a standard to develop KBSs. In Speel and Aben (1997), we have
                reported our positive experiences in reusing problem-solving methods (PSMs). In this paper, we focus on
                the feasibility of another important technique called structure-preserving design and implementation (SPD).

                In the literature it is claimed that SPD leads to many benefits including improved maintenance and reuse of
                program code. In this paper, we discuss our experiences in applying SPD in an extensive case study. We
                have tested the validity of the scientifically claimed pros and cons during the development of four industrial
                KBSs. For these off-line diagnosis and assessment applications, we found that the SPD approach is
                feasible and improves maintainability, encourages reuse on all levels, contributes to improved
                understandability, documentation and explanation and promotes systematization. In addition, the off-line
                KBSs do not demonstrate any serious performance problems. (C) 1998 Academic Press.
      Mots-clés
                problem-solving methods, knowledge
       Citations
                18
          ISSN
                1071-5819
        Adresse
     pour copies
                Speel, PH
                UNILEVER RES LABS VLAARDINGEN
                POB 114
                VLAARDINGEN, NETHERLANDS
                NL-3130 AC BC
 


Modal Change Logic (MCL): Specifying the reasoning of knowledge-based systems
        Auteurs
                Fensel D, Groenboom R, deLavalette GRR
         Source
                DATA & KNOWLEDGE ENGINEERING
                Volume 26, no. 3, JUL
          Pages
                243 - 269
    Semaine CC
                9830
           Date
                1998
        Résumé
                We investigate the formal specification of the reasoning process of knowledge-based systems in this
                paper. We analyze the corresponding parts of the KADS specification languages KARL and (ML)(2) and
                deduce some general requirements. The essence of these languages is that they integrate a declarative
                specification of inferences with control information. The languages differ in the way they achieve this
                integration and each of them has shortcomings. We propose a unifying semantical framework that
                integrates the core of the different solutions and overcomes their problems. We define a semantics and
                axiomatization with the Modal Change Logic (MCL). The main contribution of the paper is not to
                introduce yet another specification language. Instead we aim at four goals: (1) defining a framework for
                describing the dynamic reasoning behavior of knowledge-based systems which integrates existing
                approaches; (2) defining a semantics for the specification of the dynamic reasoning behavior of a
                knowledge-based system within the stares as algebras setting that overcomes several shortcomings and ad
                hoc solutions of existing approaches; and (3) providing an axiomatization that enables the development of
                mechanized proof support. (4) Through conceptual and semantical clarity, we investigate the relationships
                to similar work in software engineering and database engineering opening possibilities for further
                cross-fertilization of these fields. (C) 1998 Elsevier Science B.V.
      Mots-clés
                modal logic, dynamic logic, formal specification, knowledge-based systems, axiomatic semantics
       Citations
                55
          ISSN
                0169-023X
        Adresse
     pour copies
                Fensel, D
                UNIV KARLSRUHE
                INST AIFB, KAISERSTR 12
                KARLSRUHE, GERMANY
                D-76128 BC

Designing knowledge based systems: the CommonKADS design model
        Auteurs
                Kingston JKC
         Source
                KNOWLEDGE-BASED SYSTEMS
                Volume 11, no. 5-6, NOV 23
          Pages
                311 - 319
    Semaine CC
                9904
           Date
                1998
        Résumé
                This paper describes the three-stage approach to KBS design recommended by the CommonKADS
                Design Model (choosing an overall approach to design, choosing ideal knowledge representation and
                programming techniques, and deciding how to implement the recommended techniques in the chosen
                software), as well as outlining possible sources of guidance for making good selections of knowledge
                representations and inference techniques. It then illustrates the use of the Design Model for two systems
                developed by AIAI(1), one for machine fault diagnosis and one for mortgage application assessment. (C)
                1998 Elsevier Science B.V. All rights reserved.
      Mots-clés
                commonkads, knowledge based system design, knowledge engineering
       Citations
                16
          ISSN
                0950-7051
        Adresse
     pour copies
                Kingston, JKC
                Univ Edinburgh
                80 S Bridge
                Edinburgh, Midlothian, Scotland
                EH1 1HN AC

Modelling expert knowledge with knowledge-based systems to design decision aids -
The example of a knowledge-based model on grazing management
        Auteurs
                Girard N (girard@avignon.inra.fr), Hubert B
         Source
                AGRICULTURAL SYSTEMS
                Volume 59, no. 2, FEB
          Pages
               123 - 144
    Semaine CC
                9915
           Date
                1999
        Résumé
                In order to develop decision support tools that will assist farmers in managing their farms effectively, farm
                advisors need to gain a clear understanding of the way these farms function. Formalising the knowledge to
                be used in gaining this understanding is therefore of crucial importance in building such tools. Achieving this
                is the objective of the knowledge-based system (KBS) presented here, which is based on the expert
                knowledge and understanding scientists and extension agents have of grazing management on suckler
                sheep farms in southeastern France. The resulting model abstracts from a description of the farmer's
                practices the coherence underlying his decisions which we have named his 'strategic pattern'. This
                understanding enables us to highlight the diversity of the management styles of farmers corresponding to
                different types of decision aid. Based on our experience in developing this model, we argue that KBS is an
                effective methodology for formalising the knowledge used by advisors in their understanding of a farm
                which mainly rests on its qualitative features. In this respect, KBS shows potential in designing decision
                assistance which is relevant to the farmer's objectives. (C) 1999 Elsevier Science Ltd. All rights reserved.
      Mots-clés
                knowledge-based systems, management, grazing, decision aid, expert knowledge, modelling,
                simulation-model, support system, sheep, agriculture, acquisition, farmers, irrigation, australia, adoption,
                improve
       Citations
                74
          ISSN
                0308-521X
        Adresse
     pour copies
                Girard, N
                French Natl Inst Agr Res INRA
                Site Agroparc
                Avignon, France


A case study on the use of model-based systems for electronic fault diagnosis
        Auteurs
                Cunningham P
         Source
                ARTIFICIAL INTELLIGENCE IN ENGINEERING
                Volume 12, no. 3, JUL
          Pages
                283 - 295
    Semaine CC
                9824
           Date
                1998
        Résumé
                A generic model-based system for fault diagnosis of switching-mode power supplies is described in this
                paper. The system contains a generic deep model that captures structural and behavioural knowledge
                about modules and components used in switching-mode power supplies. This deep model provides
                building blocks that can be instantiated to describe particular power-supply designs. Once this generic
                system has been developed, it is a simple matter to develop a diagnostic system for a particular circuit.
                This represents a considerable knowledge engineering advantage over the alternative shallow KBS
                approach. The generic nature of the diagnostic competence of the system suggests that it should be
                weaker than an alternative shallow system, specific to a particular circuit. However, the evaluation of the
                system shows that it can locate between 80% and 90% of faults. This evaluation is described in detail in
                the paper. (C) 1998 Elsevier Science Limited. All rights reserved.
      Mots-clés
                electronic fault diagnosis, knowledge-based systems, model-based systems, switching-mode power
                supplies
       Citations
                24
          ISSN
                0954-1810
        Adresse
     pour copies
                Cunningham, P
                UNIV DUBLIN TRINITY COLL
                DEPT COMP SCI, COLL GREEN
                DUBLIN, IRELAND
                2 AC

A knowledge-level testing method
       Auteurs
               HaoucheGingins C, Charlet J
        Source
               INTERNATIONAL JOURNAL OF HUMAN-COMPUTER STUDIES
               Volume 49, no. 1, JUL
         Pages
               1 - 20
    Semaine CC
               9836
          Date
               1998
       Résumé
               This paper aims at describing a testing method devoted to the study of knowledge-based systems (KBS)
               behaviour and results. Like any validation method, this method relies on specifications. Moreover, as the
               focus of the study is KBS behaviour, these specifications need to be expressed at high level. We distinguish
               between two kinds of specifications, system specifications and anomaly specifications, depending on the
               validation objectives. We adopt a knowledge-level perspective of validation since our first objective is to
               compare the KBS behaviour to expected behaviours that are a part of system specifications and described
               within a KADS conceptual model. In order to detect inconsistent results, we propose to add some kind of
               validation knowledge to the domain level knowledge. This method is illustrated by the validation of a
               prototype for natural language understanding. (C) 1998 Academic Press.
      Citations
               33
         ISSN
               1071-5819
       Adresse
    pour copies
               HaoucheGingins, C
               UNIV PARIS 06
               CHU PARIS, DIAM, SERV INFORMAT MED APHP, 91 BLVD HOSP
               PARIS, FRANCE
               F-75634 BC
              13 AC


 

Conceptual modelling: an essential pillar for quality software development
        Auteurs
                Ares J, Pazos J
         Source
                KNOWLEDGE-BASED SYSTEMS
                Volume 11, no. 2, OCT 12
          Pages
                87 - 104
    Semaine CC
                9902
           Date
                1998
        Résumé
                After many years of stressing the importance of the product and the process in software development,
                emphasis has now switched to the role played by the person. This paper, however, underlines the
                importance of understanding and modelling the problem, as this is a necessary, and often sufficient,
                condition for developing good quality software. Firstly, a formal definition is given of what the problem is
                and how it can be classified. In view of the confusion in the field of software development, where the word
                model is used very vaguely, an explanation is given of what modelling means, and a generally applicable
                form of modelling is briefly discussed. Finally, conceptualisation is defined, first declaratively and then
                procedurally, and a method of building conceptual models is presented which particularly stresses the
                information map as a visual overview of the entire process. (C) 1998 Elsevier Science B.V. All rights
                reserved.
      Mots-clés
                models, conceptual model, problem solving
       Citations
                18
          ISSN
                0950-7051
        Adresse
     pour copies
                Pazos, J
                Univ Politecn Madrid
                Campus Montegancedo,S-N
                Madrid, Spain
                E-28660 BC
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC


 

On the convergence of the lagged diffusivity fixed point method in total variation image
                restoration
        Auteurs
                Chan TF (chan@math.ucla.edu), Mulet P (mulet@math.ucla.edu)
         Source
                SIAM JOURNAL ON NUMERICAL ANALYSIS
                Volume 36, no. 2, MAR 5
          Pages
                354 - 367
    Semaine CC
                9918
           Date
                1999
        Résumé
                In this paper we show that the lagged diffusivity fixed point algorithm introduced by Vogel and Oman in
                [SIAM J. Sci. Comput., 17 (1996), pp. 227-238] to solve the problem of total variation denoising,
                proposed by Rudin, Osher, and Fatemi in [Phys. D, 60 (1992), pp. 259-268], is a particular instance of a
                class of algorithms introduced by Voss and Eckhardt in [Computing, 25 (1980), pp. 243-251], whose
                origins can be traced back to Weiszfeld's original work for minimizing a sum of Euclidean lengths [Tohoku
                Math. J., 43 (1937), pp. 355-386]. There have recently appeared several proofs for the convergence of
                this algorithm [G. Aubert et al., Technical report 94-01, Informatique, Signaux et Systemes de Sophia
                Antipolis, 1994], [A. Chambolle and P.-L. Lions, Technical report 9509, CEREMADE, 1995], and [D.
                C. Dobson and C. R. Vogel, SIAM J. Numer. Anal., 34 (1997), pp. 1779-1791]. Here we present a
                proof of the global and linear convergence using the framework introduced in [H. Voss and U. Eckhart,
                Computing, 25 (1980), pp. 243-251] and give a bound for the convergence rate of the fixed point
                iteration that agrees with our experimental results. These results are also valid for suitable generalizations of
                the fixed point algorithm.
      Mots-clés
                image restoration, total variation, weiszfeld's method, fixed point, recovery
       Citations
                13
          ISSN
                0036-1429
        Adresse
     pour copies
                Chan, TF
                Univ Calif Los Angeles
                405 Hilgard Ave
                Los Angeles, CA, USA
                90024 AP
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC

                Intelligent tutoring systems as design
        Auteurs
                Wu AKW, Lee MC
         Source
                COMPUTERS IN HUMAN BEHAVIOR
                Volume 14, no. 2, MAY
          Pages
                209 - 220
    Semaine CC
                9830
           Date
                1998
        Résumé
                To build intelligent tutoring systems (ITSs) is not easy. Over the years, various attempts had been made
                with many problems uncovered. This paper presents the situation and proposes the notion of ITS as
                design to engage ITS development with more rigor. With the various related issues, principles and
                characteristics of design described, the ITS design problem space is elaborated Implications from
                adopting such a perspective, including (a) a systems approach to ITS, (b) a paradigm hierarchy, (c) the
                emergence of an agent model, and (d) the need for a description language, are highlighted. We further
                argue that by such a rendering, more systematic collection of efforts on ITS can be achieved. It is
                envisaged that the discussions would help set the research of ITS in context and provide further intuition
                towards ITS development. (C) 1998 Elsevier Science Ltd All rights reserved.
      Mots-clés
                intelligent tutoring systems, its, design, its as design, paradigm hierarchy
       Citations
                32
          ISSN
                0747-5632
        Adresse
     pour copies
                Wu, AKW
                HONG KONG POLYTECH UNIV
                DEPT COMP
                KOWLOON, HONG KONG
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC

A knowledge-based system to represent spatial reasoning for fire modelling
        Auteurs
                Knight B, Taylor S, Petridis M (m.petridis@gre.ac.uk), Ewer J, Galea ER
         Source
                ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
                Volume 12, no. 2, APR
          Pages
                213 - 219
    Semaine CC
                9917
           Date
                1999
        Résumé
                This paper describes the architecture of the knowledge based system (KBS) component of Smartfire, a
                fire field modelling tool for use by members of the fire safety engineering community who are not expert in
                modelling techniques. The KBS captures the qualitative reasoning of an experienced modeller in the
                assessment of room geometries, so as to set up the important initial parameters of the problem. Fire
                modelling expertise is an example of geometric and spatial reasoning, which raises representational
                problems. The approach taken in this project is a qualitative representation of geometric room information
                based on Forbus' concept of a metric diagram. This takes the form of a coarse grid, partitioning the
                domain in each of the three spatial dimensions. Inference over the representation is performed using a
                case-based reasoning (CBR) component. The CBR component stores example partitions with key set-up
                parameters; this paper concentrates on the key parameter of grid cell distribution. (C) 1999 Elsevier
                Science Ltd. All rights reserved.
      Mots-clés
                spatial reasoning, metric diagrams, case-based reasoning, fire modelling, qualitative simulation
       Citations
                14
          ISSN
                0952-1976
        Adresse
     pour copies
                Petridis, M
                Univ Greenwich
                London, England
                SE18 6PF AC
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC

Causal domain model driven knowledge acquisition for expert diagnosis system development
        Auteurs
                Grundspenkis J
         Source
                JOURNAL OF INTELLIGENT MANUFACTURING
                Volume 9, no. 6, DEC
          Pages
                547 - 558
    Semaine CC
                9908
           Date
                1998
        Résumé
                Despite the successful operation of expert diagnosis systems in various areas of human activity these
                systems still show several drawbacks. Expert diagnosis systems infer system faults from observable
                symptoms. These systems usually are based on production rules which reflect so called 'shallow
                knowledge' of the problem domain. Though the explanation subsystem allows the program to explain its
                reasoning, deeper theoretical justifications of program's actions are usually needed. This may be one of the
                reasons why in recent years in knowledge engineering there has been a shift from rule-based systems to
                model-based systems. Model-based systems allow us to reason and to explain a system's physical
                structure, functions and behaviour, and thus, to achieve much better understanding of the system's
                operations, both in normal mode and under fault conditions. The domain knowledge captured in the
                knowledge base of the expert diagnosis system must include deep causal knowledge to ensure the desired
                level of explanation. The objective of this paper is to develop a causal domain model driven approach to
                knowledge acquisition using an expert-acquisition system-knowledge base paradigm. The framework of
                structural modelling is used to execute systematic, partly formal model-based knowledge acquisition, the
                result of which is three structural models-one model of morphological structure and two kinds of models of
                functional structures. Hierarchy of frames are used for knowledge representation in topological knowledge
                base (TKB). A formal method to derive cause-consequence rules from the TKB is proposed. The set of
                cause-consequence rules reflects causal relationships between causes (faults) and sequences of
                consequences (changes of parameter values). The deep knowledge rule base consists of
                cause-consequence rules and provides better understanding of system's operation. This, in turn, gives the
                possibility to construct better explanation facilities for expert diagnosis system. The proposed method has
                been implemented in the automated structural modelling system ASMOS. The application areas of
                ASMOS are complex technical systems with physically heterogeneous elements.
      Mots-clés
                structural modelling, causal domain model, expert diagnosis systems, fault diagnosis, knowledge
                acquisition
       Citations
                8
          ISSN
                0956-5515
        Adresse
     pour copies
                Grundspenkis, J
                Riga Tech Univ
                1 Kalku St
                Riga, Latvia
                LV-1658 BC
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC

A transformed input-domain approach to fuzzy modeling
        Auteurs
                Kim E, Park M, Kim S, Park M
         Source
                IEEE TRANSACTIONS ON FUZZY SYSTEMS
                Volume 6, no. 4, NOV
          Pages
                596 - 604
    Semaine CC
                9849
           Date
                1998
        Résumé
                This paper presents an explanation of a fuzzy model considering the correlation among components of
                input data. Generally, fuzzy models have a capability of dividing an input space into several subspaces
                compared to a linear model. But hitherto suggested fuzzy modeling algorithms have not taken into
                consideration the correlation among components of sample data and have addressed them independently,
                which results in an ineffective partition of the input space. In order to solve this problem, this paper
                proposes a new fuzzy modeling algorithm, which partitions the input space more effectively than
                conventional fuzzy modeling algorithms by taking into consideration the correlation among components of
                sample data. As a way to use the correlation and divide the input space, the method of principal
                component is used. Finally, the results of the computer simulation are given to demonstrate the validity of
                this algorithm.
      Mots-clés
                fuzzy model, kl transform, principal of component, identification, systems
       Citations
                21
          ISSN
                1063-6706
        Adresse
     pour copies
                Kim, E
                YONSEI UNIV
                DEPT ELECT ENGN
                SEOUL, SOUTH KOREA
                120749 AC
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC


 

KROL: a knowledge representation object language on top of Prolog
        Auteurs
                Shaalan K, Rafea M, Rafea A
         Source
                EXPERT SYSTEMS WITH APPLICATIONS
                Volume 15, no. 1, JUL
          Pages
                33 - 46
    Semaine CC
                9834
           Date
                1998
        Résumé
                This paper presents a knowledge representation object language (KROL) on top of Prolog. KROL is
                aimed at providing the ability to develop second-generation expert systems. The main aspects of KROL
                include multi-paradigm knowledge representation (first-order predicate logic, objects, rules), inference
                mechanisms at different levels of granularity, explanation facility, object-oriented database management
                module, and user-friendly interface. KROL has sufficient expressive power to be used in applying
                demanding knowledge based modeling methodologies, such as KADS and Generic Task, which are the
                major landmarks of the second-generation expert systems technology. Four successful agricultural expert
                systems have been developed in the last 6 years using KROL. To demonstrate the language capabilities,
                we present an example of disorder diagnosis. (C) 1998 Elsevier Science Ltd. All rights reserved.
       Citations
                34
          ISSN
                0957-4174
        Adresse
     pour copies
                Rafea, A
                CAIRO UNIV
                DEPT INFORMAT & COMP SCI, INST STAT STUDIES & RES, 5 THARWAT ST
                GIZA, EGYPT
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC

Dynamic case-based reasoning for process operation support systems
        Auteurs
                Xia QJ, Rao M (ming.rao@ualberta.ca)
         Source
                ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
                Volume 12, no. 3, JUN
          Pages
                343 - 361
    Semaine CC
                9928
           Date
                1999
        Résumé
                Modern process industry is faced with ever-increasing requirements for better quality, higher production
                profits, safer operation, and stringent environment regulation. New technologies are required to reduce the
                operator's cognitive load and achieve more consistent operations. Operation support systems, which help
                operators in obtaining effective and timely decisions, have attracted much attention. The research
                described here intended to develop an efficient reasoning method for operation support systems. It is
                pointed out that case-based reasoning (CBR), which is based on the concept that human memory is
                episodic in nature, is consistent with operator's problem solving. Despite their successful application to the
                solution of many problems, case-based reasoning methods are mostly static. Process operation support
                systems require a CBR method that can represent system dynamics and fault-propagation phenomena. To
                solve this problem, a new approach, namely dynamic case-based reasoning (DCBR), is developed.
                DCBR introduces a number of new mechanisms including time-tagged indexes, dynamic and composite
                features, and multiple indexing paths. As a result, it provides flexible case adaptation, timely and accurate
                problem solving, and an ability to incorporate other computational and reasoning methods. (C) 1999
                Elsevier Science Ltd. All rights reserved.
      Mots-clés
                case-based reasoning, intelligent systems, process operation support, fault diagnosis, pulp and paper,
                diagnosis, framework
       Citations
                30
          ISSN
                0952-1976
        Adresse
     pour copies
                Rao, M
                Univ Alberta
                Edmonton, AB, Canada
                T6G 2G6 AP
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC

                A case-based expert system approach for quality design
        Auteurs
                Suh MS, Jhee WC, Ko YK, Lee A
         Source
                EXPERT SYSTEMS WITH APPLICATIONS
                Volume 15, no. 2, AUG
          Pages
                181 - 190
    Semaine CC
                9839
           Date
                1998
        Résumé
                Quality design in an enterprise is to design the formulation and processing parameters of products in a
                cost-effective manner so that the resulting properties can meet the quality specifications given by
                customers. This design problem often lacks the exact formula for predicting quality properties, and highly
                depends upon human designers' past experiences. This paper presents a case-based expert system
                approach for developing new product properties that can effectively support all the steps in the design
                process from storing past cases, through retrieving similar cases, to adapting the retrieved case for the new
                product. Within this approach. case-based reasoning uses a hierarchical case indexing method for efficient
                case retrieval, and provides sophisticated similarity metrics for accurate case matching. When there is a
                discrepancy between the most similar case retrieved and new product, the case-based design process
                embraces the expert system technique to reconcile the discrepancy by using domain-specific knowledge
                on the relationships between design parameters and quality properties. Such integration of case-based
                reasoning and expert system provides a systematic procedure for design engineers to retrieve past cases
                quickly and accurately, and to effectively accumulate their expertise for design adaptation. (C) 1998
                Elsevier Science Ltd. All rights reserved.
      Mots-clés
                case-based design, hierarchial indexing, knowledge-based adaption, integration of case-base reasoning
                with expert system, model
       Citations
                23
          ISSN
                0957-4174
        Adresse
     pour copies
                Jhee, WC
                HONG IK UNIV
                DEPT IND ENGN, 72-1 MAPOKU SANGSUDONG
                SEOUL, SOUTH KOREA
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC


Data analysis by positive decision trees
        Auteurs
                Makino K, Suda T, Ono H, Ibaraki T
         Source
                IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS
                Volume E82D, no. 1, JAN
           Web
                http://www.ieice.or.jp/
          Pages
                76 - 88
    Semaine CC
                9914
           Date
                1999
        Résumé
                Decision trees are used as a convenient means to explain given positive examples and negative examples,
                which is a form of data mining and knowledge discovery. Standard methods such as ID3 may provide
                non-monotonic decision trees in the sense that data with larger values in all attributes are sometimes
                classified into a class with a smaller output value. (In the case of binary data, this is equivalent to saying
                that the discriminant Boolean function that the decision tree represents is not positive.) A motivation of this
                study comes from an observation that real world data are often positive, and in such cases it is natural to
                build decision trees which represent positive (i.e., monotone) discriminant functions. For this, we propose
                how to modify the existing procedures such as ID3, so that the resulting decision tree represents a positive
                discriminant function. In this procedure, we add some new data to recover the positivity of data. which the
                original data had but was lost in the process of decomposing data sets by such methods as ID3. To
                compare the performance of our method with existing methods, we test (1) positive data, which are
                randomly generated from a hidden positive Boolean function after adding dummy attributes, and (2) breast
                cancer data as an example of the real-world data. The experimental results on (1) tell that, although the
                sizes of positive decision trees are relatively larger than those without positivity assumption, positive
                decision trees exhibit higher accuracy and tend to choose correct attributes, on which the hidden positive
                Boolean function is defined. For the breast cancer data set, we also observe a similar tendency; i.e.,
                positive decision trees are larger but give higher accuracy.
      Mots-clés
                decision trees, id3, mid, extensions, positive functions, quasi-positive functions, data mining, knowledge
                discovery and data analysis, knowledge
       Citations
                17
          ISSN
                0916-8532
        Adresse
     pour copies
                Makino, K
                Osaka Univ
                Toyonaka, Osaka, Japan
                5608531 AP
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC

RIEVL: Recursive induction learning in hand gesture recognition
        Auteurs
                Zhao M, Quek FKH, Wu XD
         Source
                IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
                Volume 20, no. 11, NOV
          Pages
                1174 - 1185
     Semaine CC
                9849
           Date
                1998
        Résumé
                This paper presents a recursive inductive learning scheme that is able to acquire hand pose models in the
                form of disjunctive normal form expressions involving multivalued features. Based on an extended
                variable-valued logic, our rule-based induction system is able to abstract compact rule sets from any set of
                feature vectors describing a set of classifications. The rule bases which satisfy the completeness and
                consistency conditions are induced and refined through five heuristic strategies. A recursive induction
                learning scheme in the RIEVL algorithm is designed to escape local minima in the solution space. A
                performance comparison of RIEVL with other inductive algorithms, ID3, NewID, C4.5, CN2, and HCV,
                is given in the paper. In the experiments with hand gestures, the system produced the disjunctive normal
                form descriptions of each pose and identified the different hand poses based on the classification rules
                obtained by the RIEVL algorithm. RIEVL classified 94.4 percent of the gesture images in our testing set
                correctly, outperforming all other inductive algorithms.
       Mots-clés
                hand gesture, hand pose recognition, rule-based induction, feature detection, feature selection, disjunctive
                norm form, machine learning, variable-valued logic, mini
       Citations
                31
           ISSN
                0162-8828
        Adresse
     pour copies
                Zhao, M
                UNIV ILLINOIS
                DEPT NEUROSURG
                CHICAGO, IL, USA
                60612 AP
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC
 


ILA: an inductive learning algorithm for rule extraction
        Auteurs
                Tolun MR, AbuSoud SM
         Source
                EXPERT SYSTEMS WITH APPLICATIONS
                Volume 14, no. 3, APR
          Pages
                361 - 370
     Semaine CC
                9831
           Date
                1998
        Résumé
                In this paper we present a novel inductive learning algorithm called the inductive Learning Algorithm (ILA)
                for extracting production rules from a set of examples. We also describe application of the ILA to a range
                of data sets with different numbers of attributes and classes. The results obtained show that the ILA is
                more general and robust than most other algorithms for inductive learning. Most of the time, ILA appears
                to be comparable to other well-known algorithms, such as AQ and ID3, if not better. (C) 1998 Elsevier
                Science Ltd. All rights reserved.
       Mots-clés
                machine
       Citations
                25
           ISSN
                0957-4174
        Adresse
     pour copies
                Tolun, MR
                MIDDLE E TECH UNIV
                DEPT COMP ENGN
                ANKARA, TURKEY
                TR-06531 BC
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC

Rule-induction and case-based reasoning: Hybrid architectures appear advantageous
        Auteurs
                Cercone N (ncercone@uwaterloo.ca), An AJ (aan@uwaterloo.ca), Chan C (chan@cs.uregina.ca)
         Source
                IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
                Volume 11, no. 1, JAN-FEB
           Web
                http://www.computer.org
          Pages
                166 - 174
     Semaine CC
                9916
           Date
                1999
        Résumé
                Researchers have embraced a variety of machine learning (ML) techniques in their efforts to improve the
                quality of learning programs. The recent evolution of hybrid architectures for machine learning systems has
                resulted in several approaches that combine rule-induction methods with case-based reasoning techniques
                to engender performance improvements over more-traditional one-representation architectures. We briefly
                survey several major rule-induction and case-based reasoning ML systems. We then examine some
                interesting hybrid combinations of these systems, and explain their strengths and weaknesses as learning
                systems. We present a balanced approach to constructing a hybrid architecture, along with arguments in
                favor of this balance and mechanisms for achieving a proper balance. Finally, we present some initial
                empirical results from testing our ideas and draw some conclusions based on those results.
       Mots-clés
                case-based reasoning, rule induction, machine learning, classification, numeric prediction
       Citations
                26
           ISSN
                1041-4347
        Adresse
     pour copies
                Cercone, N
                Univ Waterloo
                Waterloo, ON, Canada
                N2L 3G1 AP
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC

An approach to case-based system for conceptual ship design assistant
        Auteurs
                Lee D (dklee@mailgw.kimm.re.kr), Lee KH
         Source
                EXPERT SYSTEMS WITH APPLICATIONS
                Volume 16, no. 2, FEB
          Pages
                97 - 104
     Semaine CC
                9916
           Date
                1999
        Résumé
                Designers heavily depend on their experience and existing ship data, when designing a skip. In preliminary
                design stage especially, decision making based on the designer's expertise and heuristic knowledge are
                very important factors to design process because available information is limited, and cannot be fully
                supported by formal design procedure and design sheet. To support these conceptual design environment,
                the designer's experience and heuristic knowledge are transformed into readable formats which can be
                operated on computer systems. The existing ship data are very useful and important in conceptual design.
                To use this data efficiently, it requires basically database of the existing ships and make practical
                application of it. In this article, intelligent system that can be a support to the conceptual design stage
                based on knowledge engineering was developed. Major design factors and parameters of the existing skip
                data were stored case base as design cases and the case base was connected with database for
                information exchange among them [Brown, A., Watson, I., & Filer, N. (1995). Separating the cases from
                the data: towards more flexible case-based reasoning. Proc. of International Conference on Case-Based
                Reasoning 95 (ICCBR-95), Sesimbra in Portugal]. To extract a good and suitable design case for a new
                ship design from case base, learning algorithm was adapted. The obtained knowledge from designers was
                used to compensate for the differences between the design case and a new design. The developed
                interactive intelligent conceptual design system (BASCON-IV) can be applied to commercial ships and
                bulk carriers. (C) 1999 Elsevier Science Ltd. All rights reserved.
       Mots-clés
                ship design, case-based system, knowledge-based system, conceptual design system
       Citations
                17
           ISSN
                0957-4174
        Adresse
     pour copies
                Lee, D
                Korea Res Inst Ships & Ocean Engn
                POB 101
                Taejon, South Korea
                305600 AC
       Transfert
                Format texte PC/UNIX || Format Endnote PC || Format texte MAC || Format Endnote MAC