This dissertation develops methods to minimize recommendation error costs when inputs to a rule-based expert system are prone to errors. The problem often arises in web-based applications where data are inherently noisy or provided by users who perceive some benefit from falsifying inputs. Prior studies proposed methods that attempted to minimize the probability of recommendation error, but did not take into account the relative costs of different types of errors. In situations where these differences are significant, an approach that minimizes the expected misclassification error costs has advantages over extant methods that ignore these costs. Building on the existing literature, two new techniques - Cost-Based Input Modification (CBIM) and Cost-Based Knowledge-Base Modification (CBKM) were developed and evaluated. Each method takes as inputs (1) the joint probability distribution of a set of rules, (2) the distortion matrix for input noise as characterized by the probability distribution of the observed input vectors conditioned on their true values, and (3) the misclassification cost for each type of recommendation error. Under CBIM, for any observed input vector v, the recommendation is based on a modified input vector v' such that the expected error costs are minimized. Under CBKM the rule base itself is modified to minimize the expected cost of error. The proposed methods were investigated as follows: as a control, in the special case where the costs associated with different types of errors are identical, the recommendations under these methods were compared for consistency with those obtained under extant methods. Next, the relative advantages of CBIM and CBKM were compared as (1) the noise level changed, and (2) the structure of the cost matrix varied. As expected, CBKM and CBIM outperformed the extant Knowledge Base Modification (KM) and Input Modification (IM) methods over a wide range of input distortion and cost matrices, with some restrictions. Under the control, with constant misclassification costs, the new methods performed equally with the extant methods. As misclassification costs increased, CBKM outperformed KM and CBIM outperformed IM. Using different cost matrices to increase misclassification cost asymmetry and order, CBKM and CBIM performance increased. At very low distortion levels, CBKM and CBIM underperformed as error probability became more significant in each method's estimation. Additionally, CBKM outperformed CBIM over a wide range of input distortion as its technique of modifying an original knowledge base outperformed the technique of modifying inputs to an unmodified decision tree.

A dissertation submitted to the Faculty of Engineering, University of the Witwatersrand, in fulfilment of the requirements of the degree of Master of Science Engineering / A framework designed to contain and manage the use of knowledge in a real-time knowledge based system for high level control of an industrial process is presented. A prototype of the framework is designed and implemented on a static objectorientated shell. Knowledge is stored in objects and in forward chaining rules. The knowledge has a well defined structure, making it easy to create and manage. Rules are used to recognize conditions and propose control objectives. The framework uses the knowledge to determine variables that if altered will meet the objectives. Control actions are then found to implement changes to these variables The use of explicit control objectives makes it possible to determine if an action worked as intended and if its use is suitable for the present conditions. This enables a learning mechanism to be applied in the expert system. The prototype operated adequately, but the knowledge required to drive the. system was found to be very detailed and awkward to create. / Andrew Chakane 2018

Nas atividades típicas de empresas de mineração que produzem pedra britada para construção civil, conhecidas como pedreiras, é comum a exposição dos empregados a diversos fatores de risco à saúde ou integridade física, exposição que, no Brasil, confere a estes empregados o direito a adicionais de insalubridade ou periculosidade sobre o salário, definidos em lei a partir de diferentes parâmetros técnicos. As controvérsias acerca do pagamento ou não destes adicionais entre os empregadores e os empregados, a exemplo dos demais conflitos entre capital e trabalho, são solucionados no Brasil pela Justiça do Trabalho, sendo a perícia técnica uma das principais ferramentas da qual lançam mão os seus juízes, visando a apuração da efetiva ocorrência da insalubridade ou periculosidade nos ambientes de trabalho. No presente estudo foi realizada a caracterização da insalubridade e periculosidade em pedreiras sob o enfoque pericial, abordando-se para tanto os conceitos e a tipificação legal afeitos ao tema, bem como os principais aspectos da prova pericial técnica e sua contextualização no processo judicial, com o levantamento e análise das características dos ambientes de trabalho e das atividades desenvolvidas pelos empregados neste tipo de mineração. Como resultado, constatou-se a potencial ocorrência de insalubridade e/ou periculosidade em diversas das funções desenvolvidas pelos funcionários de pedreiras, pelo que foram discutidas as possíveis formas de gerenciamento da insalubridade e periculosidade no âmbito da gestão global de segurança destes empreendimentos, culminando na proposta de utilização voluntária e sistemática, por parte das empresas de mineração, das técnicas de levantamento e análise pericial, de forma a contribuir para uma melhor concepção e execução dos diversos programas de segurança e medicina do trabalho exigidos pela legislação brasileira, bem como para a redução dos custos relacionados. / In the typical activities of mining companies that produce crushed stone for civil construction, known as \"quarries\", employees exposition to diverse risk factors, to the health or physical integrity, is common. This kind of exposition, in Brazil, gives to these employees unhealthy or hazard adds on the wage, defined in law from different technical parameters. The controversies concerning the payment of these adds between the employers and the employees, like the other conflicts between capital and labor, are solved in Brazil by the labor justice, and the forensic expert survey is one of the main tools used by the judges for the verification of the effective occurrence of the unhealthy or hazardous situations in labor activities. The present study focus the unhealthy or hazardous situations in quarries under the forensic expert survey, approaching the concepts and the legal definitions of the unhealthy or hazardous situations in labor activities, the main aspects of the expert evidence technique and the judicial proceeding, and the survey and analysis of the typical characteristics of work environments and quarries employees activities. The possible forms of management of the unhealthy or hazardous situations were also discussed, in the scope of the global safety management in quarries, as well as the contribution of raised data in forensic expert survey for the conception of security and health programs, equally demanded for the brazilian legislation.

A intensa adoção dos sistemas eletrônicos em praticamente todas as atividades humanas gera grande quantidade de evidências digitais que devem ser utilizadas pelos juízes para entender os fatos ocorridos e assim se sentirem mais seguros ao julgar os processos judiciais conforme o seu convencimento íntimo. Em contrapartida, a coleta e análise dessas evidências são tarefas cada vez mais caras e demoradas porque essas evidências se tornam rapidamente mais complexas, sem fronteiras e escondidas por recursos protetores de privacidade. São típicos os embates que ocorrem entre autoridades policiais, juízes, fabricantes e consumidores quando são frustradas as determinações judiciais para quebra do sigilo telemático em função de criptografia das comunicações e dos dados armazenados, situação que tende a se agravar em função de novas tecnologias como serviços em nuvem, Internet das Coisas, inteligência artificial, robôs e drones. Nesse contexto, os investigadores forenses precisam de laboratórios cada vez mais sofisticados, capazes de realizar a quebra judicial da criptografia por mecanismos como força-bruta ou chip-off, entre muitas outras técnicas de forense computacional. Ocorre que os laboratórios atuais não têm sido suficientes para atender a crescente demanda e, adicionalmente, eles estão concentrados nos principais centros econômicos, portanto distantes dos milhares de comarcas onde tramitam os processos judiciais. A volatilidade dos dados e a complexidade dos exames exigem a participação dos representantes dos autores e réus no processo judicial, para que possa ocorrer o imprescindível contraditório técnico e debate entre os especialistas das partes, evitando o cerceamento de defesa. Porém, esse acompanhamento é prejudicado e até impossibilitado pela produtividade insuficiente dos laboratórios frente à demanda e pela distância entre a comarca, os peritos aceitos pelo juiz e os laboratórios forenses. A velocidade dessa revolução e a escassez de laboratórios habilitados sobrecarregam os serviços periciais, provocam erros e confundem processos judicias que com isso correm o risco de absolver culpados ou, pior, condenar inocentes. Para enfrentar esse problema, propõe-se que os laboratórios forenses passem a adotar a tecnologia de laboratórios utilizáveis remotamente, denominada WebLab, e que esses laboratórios passem se integrar em federações de WebLabs Forenses, de modo a proporcionar ao delegado, juiz, promotor, advogados, perito nomeado pelo juiz e representantes das partes realizar ou acompanhar remotamente via Web os exames periciais forenses, maximizando a utilização dos laboratórios mais equipados nacionalmente e internacionalmente e proporcionando a concomitante e imprescindível fiscalização pelo Poder Judiciário e pelas partes nos processos judiciais. Em síntese, esta pesquisa propõe um modelo de referência que possibilita a utilização dos laboratórios forenses via Web. / The massive adoption of electronic systems in virtually every human activity generates a great amount of digital evidence, which must be used by judges to understand facts that took place and feel more confident when judging legal processes as per one\'s inner belief. In contrast, evidence collection and analysis are increasingly expensive and time-consuming activities, as said evidence quickly become more complex, borderless and safeguarded by privacy protection features. Conflicts between police authorities, judges, manufacturers and consumers are common when legal provisions for breaking telematics secrecy are thwarted due to encryption of communication and stored data, a situation that tends to escalate due to new technologies such as cloud services, Internet of Things, artificial intelligence, robots and drones. In this context, forensic investigators require increasingly sophisticated laboratories capable of carrying out the legal encryption breach through mechanisms such as brute force or chip-off, among many other computing forensics techniques. It happens that current laboratories are not sufficient to supply the increasing demand and, in addition, said laboratories are concentrated on the main economic centers, therefore, away from thousands of districts in which legal processes are treated. Data volatility and test complexity require participation from authors and defendants in the legal process, so that invaluable technical adversarial proceedings and debates between specialists from the parties may take place, avoiding curtailment of defense. However, this monitoring is hampered, and even prevented, by poor productivity of the laboratories in face of the demand and distance between the district, expert approved by the judge and forensic laboratories. The speed of this revolution and scarcity of qualified laboratories overburden expert services, cause mistakes and confusion among legal processes, leading to the risk of acquitting guilty parties or, even worse, convicting innocent parties. In order to address this issue, it is proposed that forensic laboratories adopt a technology for remotely usable laboratories, called WebLab, and that these laboratories are therefore integrated in federations of Forensic WebLabs, as to provide the police chief, judge, prosecutor, attorneys, expert assigned by the judge and representatives of the parties to perform or remotely follow expert forensic tests, maximizing national and international utilization of better equipped laboratories and providing simultaneous and crucial supervision by the Judiciary and the parties of legal processes. In summary, this study proposes a reference model that enables use of forensic laboratories via the Web.

This dissertation studies an economy where efficiency depends on the correct match between projects and two experts. A low-skill expert has low fixed cost and low productivity, so he is more efficient in handling low-potential or low-difficulty projects. And the opposite is true for a high-skill expert. The dissertation studies the effectiveness of markets and organizations in overcoming asymmetric information issues, experts' incentives to acquire project information at a cost, and how experts use the information to facilitate the correct match. The first chapter studies a market where a referring expert privately knows a project's potential and may refer it at any price. Inspection benefits the referred expert. First, it allows him to find out the project's potential before accepting the referral offer. Second, it allows him to tailor production effort to the project's potential for maximum efficiency. In equilibrium, the referring expert pools projects into subsets and refers each subset at a different price. A higher price signals a subset of projects with higher potentials. The referred expert almost always inspects and then uses the information to make the acceptance decision. Each subset must be small enough to incentivize the referral at a price, but also large enough to incentivize inspection by the referred expert. The second chapter studies contract design within an organization. A principal has to rely on the two experts to learn about projects' difficulties. If information cost is small, the principal can implement the first best by an optimal mechanism with the low-skill expert acting as a gatekeeper. The low-skill gatekeeper expert is incentivized to acquire information and report it truthfully. Subsequently, the principal efficiently assigns the project based on the report. The third chapter studies a market where each of the two experts can exert a variable effort to acquire project information imperfectly. In the first best, experts coordinate their information acquisition efforts. In the market, either one or both experts acquire information. The two experts may fail to coordinate because one acquires information for efficient match but the other acquires information again to protect himself.

by Hung Cheung Kwok. / Thesis (M.S.)--Chinese University of Hong Kong, 1993. / Includes bibliographical references (leaves [34]). / ACKNOWLEDGEMENTS / ABSTRACT / TABLE OF CONTENTS / Chapter 1. --- INTRODUCTION --- p.3 / Chapter 1.1 --- PROVISION OF NETWORK SERVICES --- p.3 / Chapter 1.2 --- THE CURRENT ENVIRONMENT FOR SERVICE PROVISION --- p.3 / Chapter 1.3 --- THE COMING ENVIRONMENT FOR SERVICE PROVISION --- p.6 / Chapter 1.4 --- THE ROLE OF THE ASSISTANT SYSTEM --- p.8 / Chapter 1.5 --- OVERVIEW OF THE THESIS --- p.10 / Chapter 2 . --- OVERALL DESIGN --- p.12 / Chapter 2.1 --- INFERENCE ENGINE --- p.14 / Chapter 2 .2 --- KNOWLEDGE BASES --- p.14 / Chapter 2.3 --- COMMAND GENERATOR --- p.15 / Chapter 2.4 --- USER INTERFACE --- p.16 / Chapter 2.5 --- HOST INTERFACE --- p.16 / Chapter 3 . --- KNOWLEDGE BASE DESIGN --- p.18 / Chapter 4. --- INFERENCE ENGINE DESIGN --- p.21 / Chapter 5. --- IMPLEMENTATION --- p.24 / Chapter 5.1 --- SYSTEM DESIGN --- p.24 / Chapter 5.2 --- KNOWLEDGE BASE MAINTENANCE --- p.24 / Chapter 5.3 --- MANUAL MODE --- p.26 / Chapter 5.4 --- AUTOMATIC MODE --- p.27 / Chapter 5.5 --- INTERFACE FUNCTIONS --- p.29 / Chapter 6. --- CONCLUSIONS --- p.31 / REFERENCES / LIST OF FIGURES / Chapter F1. --- THE SCHEMATICS OF THE INTELLIGENT SYSTEM / LIST OF TABLES / Chapter T1. --- STRUCTURE OF SERVICE INFORMATION FILE (SIF) / Chapter T2. --- STRUCTURE OF WORK CODE FILE (WCF) / Chapter T3. --- STRUCTURE OF MAPPING FILE (SIF-PRF´ØMAP) / Chapter T4. --- DATABASE RECORDS OF SIF一PRF.MAP / APPENDICES / Chapter A1. --- KNOWLEDGE BASE LIST FOR FEATURE ANALYSIS (CHK_SIF) / Chapter A2. --- KNOWLEDGE BASE LIST FOR WORK CODE INTERPRETATION (CHK一WCF) / Chapter A3. --- KNOWLEDGE BASE LIST FOR FILE SENDING MONITORING (CHK 一SND) / Chapter A4. --- A SAMPLE RUN OF THE SYSTEM

by Chang Yu. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1995. / Includes bibliographical references (leaves 111-115). / Acknowledgement --- p.i / Abstract --- p.ii / List of Tables --- p.vii / List of Figures --- p.viii / Chapter 1 --- INTRODUCTION --- p.1 / Chapter 1.1 --- Motivation and Objectives --- p.1 / Chapter 1.2 --- Outline of the Work of This Thesis --- p.4 / Chapter 1.3 --- Organization of the Thesis --- p.5 / Chapter 2 --- REVIEW OF RELATED WORKS --- p.6 / Chapter 2.1 --- Deduce2 --- p.6 / Chapter 2.2 --- ARES --- p.8 / Chapter 2.3 --- VAGUE --- p.10 / Chapter 2.4 --- Fuzzy Sets-Based Approaches --- p.12 / Chapter 2.5 --- Some General Remarks --- p.14 / Chapter 3 --- A FUZZY DATABASE QUERY LANGUAGE --- p.18 / Chapter 3.1 --- Basic Concepts of Fuzzy Sets --- p.18 / Chapter 3.2 --- The Syntax of the Fuzzy Query Language --- p.21 / Chapter 3.3 --- Fuzzy Operators --- p.25 / Chapter 3.3.1 --- AND --- p.27 / Chapter 3.3.2 --- OR --- p.27 / Chapter 3.3.3 --- COMB --- p.28 / Chapter 3.3.4 --- POLL --- p.28 / Chapter 3.3.5 --- HURWICZ --- p.30 / Chapter 3.3.6 --- REGRET --- p.31 / Chapter 4 --- SYSTEM DESIGN --- p.35 / Chapter 4.1 --- General Requirements and Definitions --- p.35 / Chapter 4.1.1 --- Requirements of the system --- p.36 / Chapter 4.1.2 --- Representation of membership functions --- p.38 / Chapter 4.2 --- Overall Architecture --- p.41 / Chapter 4.3 --- Interface --- p.44 / Chapter 4.4 --- Knowledge Base --- p.46 / Chapter 4.5 --- Parser --- p.51 / Chapter 4.6 --- ORACLE --- p.52 / Chapter 4.7 --- Data Manager --- p.53 / Chapter 4.8 --- Fuzzy Processor --- p.57 / Chapter 5 --- IMPLEMENTION --- p.59 / Chapter 5.1 --- Some General Considerations --- p.59 / Chapter 5.2 --- Knowledge Base --- p.60 / Chapter 5.2.1 --- Converting a concept into conditions --- p.60 / Chapter 5.2.2 --- Concept trees --- p.62 / Chapter 5.3 --- Data Manager --- p.64 / Chapter 5.3.1 --- Some issues on the implementation --- p.64 / Chapter 5.3.2 --- Dynamic library --- p.67 / Chapter 5.3.3 --- Precompiling process --- p.68 / Chapter 5.3.4 --- Calling standard --- p.71 / Chapter 6 --- CASE STUDIES --- p.76 / Chapter 6.1 --- A Database for Job Application/Recruitment --- p.77 / Chapter 6.2 --- Introduction to the Knowledge Base --- p.79 / Chapter 6.3 --- Cases --- p.79 / Chapter 6.3.1 --- Crispy queries --- p.79 / Chapter 6.3.2 --- Fuzzy queries --- p.82 / Chapter 6.3.3 --- Concept queries --- p.85 / Chapter 6.3.4 --- Fuzzy Match --- p.87 / Chapter 6.3.5 --- Fuzzy operator --- p.88 / Chapter 7 --- CONCLUSION --- p.93 / Appendix A Sample Data in DATABASE --- p.96 / Bibliography --- p.111

by So Yuen Tai. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1990. / Bibliography: leaves [144]-[146]. / ACKNOWLEDGEMENTS / TABLE OF CONTENTS --- p.T.1 / ABSTRACT / Chapter 1. --- INTRODUCTION --- p.1.1 / Chapter 1.1 --- Inexact Knowledge in Medical Expert Systems --- p.1.1 / Chapter 1.2 --- Fuzzy Expert System Shells --- p.1.2 / Chapter 1.2.1 --- SPII-2 --- p.1.3 / Chapter 1.2.2 --- Fuzzy Expert System Shell for Decision Support System --- p.1.4 / Chapter 1.3 --- Medical Expert Systems --- p.1.6 / Chapter 1.3.1 --- EXPERT --- p.1.6 / Chapter 1.3.2 --- DIABETO --- p.1.8 / Chapter 1.4 --- Impact from Micro-computer --- p.1.10 / Chapter 1.5 --- Approach --- p.1.11 / Chapter 2. --- SYSTEM Z-ll --- p.2.1 / Chapter 2.1 --- General Description --- p.2.1 / Chapter 2.2 --- Main Features --- p.2.2 / Chapter 2.2.1 --- Fuzzy Concepts --- p.2.2 / Chapter 2.2.2 --- Fuzzy Certainty --- p.2.3 / Chapter 2.2.3 --- Fuzzy Comparison --- p.2.5 / Chapter 2.2.4 --- Rule Evaluation --- p.2.7 / Chapter 2.2.5 --- Certainty Factor Propagation --- p.2.9 / Chapter 2.2.6 --- Linguistic Approximation --- p.2.10 / Chapter 2.3 --- Limitations and Possible Improvements --- p.2.11 / Chapter 3. --- A FUZZY EXPERT SYSTEM SHELL (Z-lll) IN PC ENVIRONMENT --- p.3.1 / Chapter 3.1 --- General Description --- p.3.1 / Chapter 3.2 --- Programming Environment --- p.3.1 / Chapter 3.3 --- Main Features and Structure --- p.3.3 / Chapter 3.3.1 --- Knowledge Acquisition Module --- p.3.5 / Chapter 3.3.1.1 --- Object Management Module --- p.3.5 / Chapter 3.3.1.2 --- Rule Management Module --- p.3.6 / Chapter 3.3.1.3 --- Fuzzy Term Management Module --- p.3.7 / Chapter 3.3.2 --- Consultation Module --- p.3.8 / Chapter 3.3.2.1 --- Fuzzy Inference Engine --- p.3.8 / Chapter 3.3.2.2 --- Review Management Module --- p.3.11 / Chapter 3.3.2.3 --- Linguistic Approximation Module --- p.3.11 / Chapter 3.3.3 --- System Properties Management Module --- p.3.13 / Chapter 3.4 --- Additional Features --- p.3 14 / Chapter 3.4.1 --- Weights --- p.3.15 / Chapter 3.4.1.1 --- Fuzzy Weight --- p.3.16 / Chapter 3.4.1.2 --- Fuzzy Weight Evaluation --- p.3.17 / Chapter 3.4.1.3 --- Results of Adding Fuzzy Weights --- p.3.21 / Chapter 3.4.2 --- Fuzzy Matching --- p.3.24 / Chapter 3.4.2.1 --- Similarity --- p.3.25 / Chapter 3.4.2.2 --- Evaluation of Similarity measure --- p.3.26 / Chapter 3.4.3 --- Use of System Threshold --- p.3.30 / Chapter 3.4.4 --- Use of Threshold Expression --- p.3.33 / Chapter 3.4.5 --- Playback File --- p.3.35 / Chapter 3.4.6 --- Database retrieval --- p.3.37 / Chapter 3.4.7 --- Numeric Variable Objects --- p.3.39 / Chapter 3.5 --- Implementation Highlights --- p.3.41 / Chapter 3.5.1 --- Knowledge Base --- p.4.42 / Chapter 3.5.1.1 --- Fuzzy Type --- p.4.42 / Chapter 3.5.1.2 --- Objects --- p.3.45 / Chapter 3.5.1.3 --- Rules --- p.3.49 / Chapter 3.5.2 --- System Properties --- p.3.53 / Chapter 3.5.2.1 --- System Menu --- p.3.53 / Chapter 3.5.2.2 --- Option Menu --- p.3.54 / Chapter 3.5.3 --- Consultation System --- p.3.55 / Chapter 3.5.3.1 --- Inference Engine --- p.3.56 / Chapter 3.5.3.2 --- Review Management --- p.3.60 / Chapter 3.6 --- Comparison on Z-lll and Z-ll --- p.3.61 / Chapter 3.6.1 --- Response Time --- p.3.62 / Chapter 3.6.2 --- Accessibility --- p.3.62 / Chapter 3.6.3 --- Accommodation of Large Knowledge Base --- p.3.62 / Chapter 3.6.4 --- User-Friendliness --- p.3.63 / Chapter 3.7 --- General Comments on Z-lll --- p.3.64 / Chapter 3.7.1 --- Adaptability --- p.3.64 / Chapter 3.7.2 --- Adequacy --- p.3.64 / Chapter 3.7.3 --- Applicability --- p.3.65 / Chapter 3.7.4 --- Availability --- p.3.65 / Chapter 4. --- KNOWLEDGE ENGINEERING --- p.4.1 / Chapter 4.1 --- Techniques used in Knowledge Acquisition --- p.4.1 / Chapter 4.2 --- Interviewing the Expert --- p.4.2 / Chapter 4.3 --- Knowledge Representation --- p.4.4 / Chapter 4.4 --- Development Approach --- p.4.6 / Chapter 4.5 --- Knowledge Refinement --- p.4.7 / Chapter 4.6 --- Consistency Check and Completeness Check --- p.4.12 / Chapter 4.6.1 --- The Consistency and Completeness in a nonfuzzy rule set --- p.4.13 / Chapter 4.6.1.1 --- Inconsistency in nonfuzzy rule-based system --- p.4.13 / Chapter 4.6.1.2 --- Incompleteness in nonfuzzy rule-based system --- p.4.18 / Chapter 4.6.2 --- Consistency Checks in Fuzzy Environment --- p.4.20 / Chapter 4.6.2.1 --- Affinity --- p.4.21 / Chapter 4.6.2.2 --- Detection of Inconsistency and Incompleteness in Fuzzy Environment --- p.4.24 / Chapter 4.6.3 --- Algorithm for Checking Consistency --- p.4.25 / Chapter 5. --- FUZZY MEDICAL EXPERT SYSTEMS --- p.5.1 / Chapter 5.1 --- ABVAB --- p.5.1 / Chapter 5.1.1 --- General Description --- p.5.1 / Chapter 5.1.2 --- Development of ABVAB --- p.5.2 / Chapter 5.1.3 --- Computerisation of Database --- p.5.4 / Chapter 5.1.4 --- Results of ABVAB --- p.5.7 / Chapter 5.1.5 --- From Minicomputer to PC --- p.5.15 / Chapter 5.2 --- INDUCE36 --- p.5.17 / Chapter 5.2.1 --- General Description --- p.5.17 / Chapter 5.2.2 --- Verification of INDUCE36 --- p.5.18 / Chapter 5.2.3 --- Results --- p.5.19 / Chapter 5.3 --- ESROM --- p.5.21 / Chapter 5.3.1 --- General Description --- p.5.21 / Chapter 5.3.2 --- Multi-layer Medical Expert System --- p.5.22 / Chapter 5.3.3 --- Results --- p.5.25 / Chapter 6. --- CONCLUSION --- p.6.1 / REFERENCES --- p.R.1 / APPENDIX I --- p.A.1 / APPENDIX II --- p.A.2 / APPENDIX III --- p.A.3 / APPENDIX IV --- p.A.14

by C.H. Fung. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1989. / Bibliography: leaves [81-82]

by Lam Sze-sing. / "April 2000." / Thesis (Ph.D.)--Chinese University of Hong kong, 2000. / Includes bibliographical references 9p. 149-157). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. Ann Arbor, MI : ProQuest Information and Learning Company, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Mode of access: World Wide Web. / Abstracts in English and Chinese.