Thesis (Ph. D.)--State University of New York at Binghamton, Thomas J. Watson School of Engineering and Applied Science, Department of Bioengineering, Biomedical Engineering, 2009. / Includes bibliographical references.

Στην εργασία που ακολουθεί, ασχολούμαστε με την εφαρμογή μεθόδων Τεχνητής Νοημοσύνης σε ένα πραγματικό πρόβλημα, που αναφέρεται στην διάγνωση του βαθμού ασφαλισιμότητας ενός πελάτη μιας ασφαλιστικής εταιρείας. Η ανάγκη για την εφαρμογή αυτή προέκυψε από το γεγονός ότι πολλές φορές ο εμπειρογνώμονας της εταιρείας δεν είναι διαθέσιμος, αλλά και όταν είναι, χρειάζεται ένα συμβουλευτικό πρόγραμμα. Πιο συγκεκριμένα, για τη λύση του προβλήματος χρησιμοποιούνται: α) ένα ασαφές έμπειρο σύστημα υλοποιημένο με τη βοήθεια του εργαλείου FuzzyCLIPS, β) ένα έμπειρο σύστημα που χρησιμοποιεί κανόνες με συντελεστές βεβαιότητας τύπου MYCIN, γ) ένα έμπειρο σύστημα που χρησιμοποιεί κανόνες με συντελεστές βεβαιότητας τύπου weighted, υλοποιημένα και τα δύο με βάση το εργαλείο CLIPS και δ) ένα νευρωνικό δίκτυο υλοποιημένο με βάση το εργαλείο WEKA. Στο τέλος συγκρίνουμε τα παραπάνω συστήματα με βάση κάποιες μετρικές. Πριν να ξεκινήσουμε την ανάλυση του προβλήματός μας και των υλοποιήσεων των παραπάνω συστημάτων, αναλύουμε λίγο παραπάνω τους όρους και τα εργαλεία που ήδη αναφέραμε, δίνοντας περισσότερες πληροφορίες για την προέλευση τους, τα χαρακτηριστικά τους, τη χρησιμότητά τους, κτλ. Έτσι, αρχικά δίνουμε περισσότερα στοιχεία για τον τομέα της Τεχνητής Νοημοσύνης και πώς αυτός έχει εξελιχτεί στις τελευταίες δεκαετίες, και αναλύουμε τη συσχέτιση των Έμπειρων Συστημάτων με την Τεχνητή Νοημοσύνη, τα χαρακτηριστικά τους, τη δομή τους, τα πλεονεκτήματα και μειονεκτήματά τους. Στη συνέχεια, αναλύουμε τα τρία εργαλεία που θα χρησιμοποιήσουμε και τις δυνατότητες αυτών. Κι αφού δώσουμε περισσότερες πληροφορίες για το πρόβλημα της ‘Ασφαλισιμότητας’ και τον τρόπο που το αντιμετωπίζουμε, γίνεται παρουσίαση των παραπάνω ευφυών συστημάτων και των αποτελεσμάτων τους σε συγκεκριμένο σύνολο δεδομένων. Τέλος, προχωράμε σε σύγκριση και σχολιασμό των τιμών των μετρικών που προέκυψαν από τις προηγούμενες εφαρμογές, και εξαγωγή των συμπερασμάτων της σύγκρισης. / In the work that follows, we deal with the application of methods of Artificial Intelligence in a real problem, that is concerned with the diagnosis of degree of ‘how safe is to insure a customer’ in an insurance company. The need for this application resulted from the fact that many times over, the expert of the insurance company may not be available, but also when he is, he could use an advisory program. To be more exact, for the solution of the problem described above we use: a) a fuzzy expert system (in our case we use FuzzyCLIPS), b) an expert system that use rules with certainty factors as in the MYCIN tool, c) an expert system that uses rules with certainty factors as in the Weighted tool, both programmed using the CLIPS expert systems tool d) a neural network through WEKA neural network producer tool. Finally, we compare the above mentioned systems by calculating a set of metrics to conclude which method produces the most accurate results. Before analysing our problem and running the systems mentioned above, we analyze fatherly the terms and the tools that we use, providing more information on their characteristics, usefulness, etc. Thus, initially we give more information about Artificial Intelligence and how it has developed in the last decades, and we analyze the cross-correlation of Expert Systems with Artificial Intelligence, their characteristics, their structure, their advantages and disadvantages. After that, we analyze the three tools that we will use, and their possibilities, advantages and disadvantages. After giving more information on the problem of ‘how safe is to insure a customer’ and the way we deal with it, we present the above expert systems and their results in a specific dataset. Finally, we compare the metrics that were calculated from the previous applications, and comment on the conclusions of this comparison.

Σκοπός της συγκεκριμένης εργασίας είναι η έρευνα πάνω στον τομέα της αυτόματης παραγωγής έμπειρων συστημάτων, ανακαλύπτοντας γνώση μέσα σε σύνολα δεδομένων και αναπαριστώντας την με την μορφή κανόνων. Ουσιαστικά πρόκειται για μια μέθοδο επιτηρούμενης μάθησης όπως η εξόρυξη κανόνων ταξινόμησης, ωστόσο ο στόχος δεν είναι αποκλειστικά η ταξινόμηση, αλλά και η τήρηση σημαντικών προδιαγραφών ενός έμπειρου συστήματος όπως η επεξήγηση, η ενημέρωση για νέα δεδομένα κ.α. Στα πλαίσια της προπτυχιακής μου εργασίας αναπτύχθηκε ένα εργαλείο που είχε σκοπό την σύγκριση μεθόδων για συνδυασμό αβέβαιων συμπερασμάτων για το ίδιο γεγονός, στο μοντέλο των Συντελεστών Βεβαιότητας. Το εργαλείο έδινε την δυνατότητα να παραχθούν Έμπειρα Συστήματα (στη γλώσσα CLIPS) που χρησιμοποιούν τις παραπάνω μεθόδους. Σκοπός της παρούσας εργασίας ήταν η διερεύνηση του τομέα της μηχανικής μάθησης και η επέκταση του υπάρχοντος εργαλείου, ώστε να παράγει έμπειρα συστήματα με έναν πιο αυτόματο, αποδοτικό και λειτουργικό τρόπο. Πιο συγκεκριμένα τροποποιήθηκε η αρχιτεκτονική για την υποστήριξη μεταβλητών εξόδου με περισσότερες από δυο κλάσεις (Multiclass Classification). Επίσης έγινε επέκταση ώστε να μπορούν να εξαχθούν κανόνες για περισσότερες μεταβλητές του συνόλου δεδομένων (εκτός δηλαδή από την μεταβλητή εξόδου), για τις οποίες δεν χρειάζεται πλέον να γνωρίζει τιμές ο τελικός χρήστης του έμπειρου συστήματος. Η επέκταση αυτή δίνει την δυνατότητα να σχεδιαστούν πιο πολύπλοκες ιεραρχίες κανόνων, που ακολουθούν μια δενδρική δομή, εύκολα ερμηνεύσιμη από τον άνθρωπο. Το μοντέλο συντελεστών βεβαιότητας επανασχεδιάστηκε, ενώ πλέον προσφέρεται και ένας εναλλακτικός τρόπος υπολογισμού των συντελεστών βεβαιότητας των κανόνων ταξινόμησης ο οποίος βασίζεται στον ορισμό τους στο έμπειρο σύστημα MYCIN. Τα αποτελέσματα έδειξαν ότι σε μη ισορροπημένα σύνολα δεδομένων η μέθοδος αυτή ευνοεί την πρόβλεψη για την κλάση μειοψηφίας. Τεχνικές επιλογής υποσυνόλων χαρακτηριστικών, δίνουν την δυνατότητα αυτοματοποίησης σε μεγάλο βαθμό της διαδικασίας παραγωγής του έμπειρου συστήματος με τρόπο αποδοτικό. Άλλες προσθήκες είναι η δυνατότητα δημιουργίας συστημάτων που μπορούν να ενημερώνονται δυναμικά αξιοποιώντας νέα δεδομένα για το πρόβλημα, η παραγωγή κανόνων και συναρτήσεων για την αλληλεπίδραση με τον χρήστη, η παροχή γραφικού περιβάλλοντος για το παραγόμενο έμπειρο σύστημα κ.α. / The main objective of this thesis is to present a method for automatic generation of expert systems, by extracting knowledge from datasets and representing it in the form of production rules. We use a supervised machine learning method, resembling Classification Rule Mining, although classification is not our only goal. Important operational characteristics of expert systems, like explanation of conclusions and dynamic update of the knowledge base, are also taken into account. Our approach is implemented within an existing tool, initially developed by us to compare methods for combining uncertain conclusions about the same event, based on the uncertainty model of Certainty Factors. That tool could generate Expert Systems (in CLIPS language) that use the above methods. The main aim of this thesis is to do research mainly on the field of machine learning in order to enhance the above mentioned tool for generating Expert Systems in a more automatic, efficient and functional fashion. More specifically, the architecture has been modified to support output variables classified in more than two classes (Multiclass Classification). An extension of the system made it possible to generate classification rules for additional variables (apart from the output variable), for which the final user of the expert system cannot provide values. This gives the ability to design more complex rule hierarchies, which are represented in an easy-to-understand tree form. Furthermore, the certainty factors model has been revised and an additional method of computing them is offered, following the definitions in MYCIN’s model. Experimental results showed improved performance, especially for prediction of minority classes in imbalanced datasets. Feature ranking and subset selection techniques help to achieve the generation task in a more automatic and efficient way. Other enhancements include the ability to produce expert systems that dynamically update the certainty factors in their rules, the generation of rules and functions for interaction with the end-user and a graphical interface for the produced expert system.

Made available in DSpace on 2016-06-02T20:39:17Z (GMT). No. of bitstreams: 1 5099.pdf: 1564575 bytes, checksum: 02de59c122ee79d4ecdf2334d473c650 (MD5) Previous issue date: 2013-03-12 / This research analyzes the relations established by agents with the expert knowledge in the context of a public policy for rural development, by reconstructing the experiences of those who took part of its processes of formulation and implementation. These processes are often mediated by funding institutions and rural extension agencies. Analyzing social phenomena from the perspective of mediations, and more specifically, from the perspective of expertise, assumes the centrality of power dimensions, the identification of its asymmetries and constant disputes over the construction of views about ruralities. The object of this study is the Land Management III Project (Programa Estadual de Microbacias de São Paulo - PEMH), one of the first public policies with a territorial, decentralized and participative approach for the sustainable rural development of the State of Sao Paulo. This program was partially financed by the World Bank and executed between 2000 and 2008 by the State Secretary of Agriculture, through its rural extension body, the Coordination of Integral Technical Assistance (CATI). The problematization of this research was designed from Anthony Giddens concept of systems experts, from whom we also extracted the notions of reflexivity, agency, time and space. Manuel Castells concepts of space of flows and space of places also assisted in the interpretation, as well as Michel Foucault's formulations on discourse. The field researches required the contributions of Luc Boltanski and Erving Goffman on trust and interaction processes. Techniques of social qualitative research, such as interviews and document analysis, were employed. / Esta dissertação analisa as relações que os agentes estabelecem com a perícia técnicocientífica em uma política pública de desenvolvimento rural, a partir da reconstrução da experiência de participantes de sua formulação e operacionalização - processos quase sempre mediados por instituições financiadoras e órgãos de extensão rural. Analisar os fenômenos sociais pela perspectiva das mediações e, em específico, da perícia, pressupõe a centralidade da dimensão do poder, o apontamento de suas assimetrias e de constantes disputas pela construção de visões de mundo acerca do rural. O estudo adota como objeto o Programa Estadual de Microbacias Hidrográficas de São Paulo (PEMH), uma das primeiras políticas públicas com viés territorial, descentralizado e participativo voltadas ao desenvolvimento sustentável do rural paulista. Tal programa foi parcialmente financiado pelo Banco Mundial e executado, entre 2000 e 2008, pela Secretaria de Agricultura do Estado de São Paulo, por meio de seu órgão de extensão, a Coordenadoria de Assistência Técnica Integral (CATI). O problema desta pesquisa foi pensado a partir do conceito de sistemas peritos, de Anthony Giddens, do qual também foram extraídas as noções de reflexividade, agência, tempo e espaço. Os conceitos de espaço de fluxos e de espaço de lugares, de Manuel Castells, também auxiliaram na interpretação, assim como as formulações de Michel Foucault sobre discurso. A pesquisa de campo demandou o aporte das contribuições de Luc Boltanski e Erving Goffman no tocante a processos de interação e confiança. Técnicas qualitativas de pesquisa social, como a realização de entrevistas e a análise de documentos, foram empregadas.

Made available in DSpace on 2014-12-17T15:47:52Z (GMT). No. of bitstreams: 1 CassioHF_DISSERT.pdf: 1503543 bytes, checksum: f8c5d434affbf0ef65dde0a97d3eb071 (MD5) Previous issue date: 2010-02-22 / Despite the emergence of other forms of artificial lift, sucker rod pumping systems remains hegemonic because of its flexibility of operation and lower investment cost compared to other lifting techniques developed. A successful rod pumping sizing necessarily passes through the supply of estimated flow and the controlled wear of pumping equipment used in the mounted configuration. However, the mediation of these elements is particularly challenging, especially for most designers dealing with this work, which still lack the experience needed to get good projects pumping in time. Even with the existence of various computer applications on the market in order to facilitate this task, they must face a grueling process of trial and error until you get the most appropriate combination of equipment for installation in the well. This thesis proposes the creation of an expert system in the design of sucker rod pumping systems. Its mission is to guide a petroleum engineer in the task of selecting a range of equipment appropriate to the context provided by the characteristics of the oil that will be raised to the surface. Features such as the level of gas separation, presence of corrosive elements, possibility of production of sand and waxing are taken into account in selecting the pumping unit, sucker-rod strings and subsurface pump and their operation mode. It is able to approximate the inferente process in the way of human reasoning, which leads to results closer to those obtained by a specialist. For this, their production rules were based on the theory of fuzzy sets, able to model vague concepts typically present in human reasoning. The calculations of operating parameters of the pumping system are made by the API RP 11L method. Based on information input, the system is able to return to the user a set of pumping configurations that meet a given design flow, but without subjecting the selected equipment to an effort beyond that which can bear / Apesar do surgimento de outras t?cnicas de eleva??o artificial de petr?leo, sistemas de bombeio mec?nico mant?m-se hegem?nicos devido ? sua flexibilidade de atua??o e custo menor de investimento se comparada com outras t?cnicas de eleva??o desenvolvidas. Um dimensionamento de bombeio mec?nico bem sucedido necessariamente passa pelo atendimento da vaz?o prevista e pelo desgaste controlado dos equipamentos de bombeio utilizados na configura??o montada. Entretanto, a concilia??o destes elementos mostrase particularmente desafiadora, sobretudo para a maioria dos projetistas que lidam com este trabalho, que n?o possuem ainda a experi?ncia necess?ria para chegar a bons projetos de bombeio em tempo h?bil. Mesmo com a exist?ncia de diversos aplicativos computacionais no mercado com o objetivo de facilitar esta tarefa, eles precisam enfrentar um exaustivo processo de tentativa e erro at? chegar a combina??o mais adequada de equipamentos para instala??o no po?o. A proposta do presente trabalho consiste em desenvolver um sistema especialista no dimensionamento de sistemas de bombeio mec?nico. Ele tem a miss?o de guiar um engenheiro de petr?leo na tarefa de selecionar um conjunto de equipamentos apropriados ao contexto provido pelas caracter?sticas do ?leo que ser? produzido. Caracter?sticas como o n?vel de separa??o do g?s, presen?a de elementos corrosivos, possibilidade de produ??o de areia e de parafina??o s?o levados em considera??o na escolha da bomba de fundo, coluna de hastes e unidade de bombeio, bem como as caracter?sticas de opera??o dos mesmos. Ele ? capaz de aproximar seu processo de infer?ncia da forma do racioc?nio humano, o que gera resultados mais pr?ximos daqueles obtidos por um especialista. Para tanto, suas regras de produ??o foram elaboradas com base na teoria dos conjuntos fuzzy, capazes de modelar conceitos imprecisos tipicamente presentes no racioc?nio humano. Os c?lculos dos par?metros operacionais do sistema de bombeio s?o feitos por meio do m?todo API RP 11L. Com base em informa??es de entrada, o sistema ? capaz de retornar ao usu?rio um conjunto de configura??es de bombeio mec?nico que atendam uma determinada vaz?o de projeto, por?m sem submeter os equipamentos selecionados a um esfor?o al?m daquele que possam suportar

In an ever-changing and demanding world where short assembly and innovation times are indispensable, it is of paramount importance to ensure that the machinery used throughout the whole process of a product are in their best possible condition. This guarantees that the performance of each machine will be optimal, and hence, the process times will be the shortest possible, while the best quality products are obtained. Moreover, having a machine in an impeccable status permits making the necessary changes to it, in order to fulfil the requirements that a more advanced or complex product may have. Maintenance operations and their corresponding trainings have historically been time-consuming, and a vast amount of information has been transmitted from an expert to a newer operator. This means that there has been the need of working with experienced operators to secure that a good service is provided. However, different technologies like augmented reality (AR) have been shown to have a positive impact in the support and monitoring of operators in industrial maintenance operations.The present project gathers information in regard to the framework of AR, with the aim of supporting and monitoring operators in industrial environments. The proposed method consists on the development of an artefact, which would lead to a possible improvement of the already existing solutions. It is believed that the development of an AR application could grant the necessary aid to any operator in maintenance operations. The result of this suggestion is an AR application which superimposes visual information on the physical equipment.

Submitted by Reinaldo Almeida (reinaldo.almeida@ufba.br) on 2017-05-23T16:38:39Z No. of bitstreams: 1 Análise de Domínio na Aquisição de Conhecimentos - Ontologias para Sistemas Computacionais.pdf: 4033106 bytes, checksum: 207a133d5a98f624185f2ffd87870a6c (MD5) / Approved for entry into archive by Maria Auxiliadora da Silva Lopes (silopes@ufba.br) on 2017-05-25T18:25:37Z (GMT) No. of bitstreams: 1 Análise de Domínio na Aquisição de Conhecimentos - Ontologias para Sistemas Computacionais.pdf: 4033106 bytes, checksum: 207a133d5a98f624185f2ffd87870a6c (MD5) / Made available in DSpace on 2017-05-25T18:25:37Z (GMT). No. of bitstreams: 1 Análise de Domínio na Aquisição de Conhecimentos - Ontologias para Sistemas Computacionais.pdf: 4033106 bytes, checksum: 207a133d5a98f624185f2ffd87870a6c (MD5) / A partir do alinhamento entre as Semióticas desenvolvidas pelos filósofos e pensadores, Charles Sanders Peirce, Gilles Deleuze e Félix Guattari, e da atualização teórica para a atividade de Análise de Domínio, baseada nos pressupostos defendidos pelos pesquisadores da Royal School of Library and Information Science, da Dinamarca, com destaque para Birger Hjørland e Torkild Thellefsen, esta Tese disserta sobre os aspectos de cognição a serem observados para determinar o significado num universo do discurso referente a fatos de um domínio, com o objetivo de aumentar o grau de aproximação entre as realidades, dos fatos, entendida e significada. Deste modo, é feito um aprofundamento no processo de aquisição do conhecimento, com a crítica à abordagem atomista e estruturalista, na qual, termos e relações do universo de discurso são especificados a partir de uma relação direta entre signo e significado, de uma concepção onde a expressão supera o conteúdo, e a dimensão espaço prevalece sobre a dimensão tempo no processo de significância. O ambiente de estudo usado é aquele referente às ontologias computacionais, bases de conhecimentos apoiadas sobre redes semânticas e semióticas de frames, concentrado nas fases que vão do entendimento da realidade de um domínio até aquela onde a significância dos termos e relações é tratada a fim de se obter os seus respectivos significados. A pesquisa, na sua fase experimental, dentro do referencial proposto, analisou as etapas de desenvolvimento da ontologia EDXL-RESCUER, contrapondo as hipóteses tratadas na Tese e o processo de desenvolvimento da ontologia, tendo como resultados, uma abordagem crítica e uma fundamentação teórica correspondente, complementada por uma metodologia para Análise de Domínio capaz de atuar numa dimensão pós-estruturalista. O método de pesquisa aplicado é qualitativo, exploratório, envolvendo atualização do estado da arte para os conceitos apresentados, a partir da análise de um projeto de construção de ontologia. / ABSTRACT From the alignment between the Semiotics developed by the philosophers, Charles Sanders Peirce, Gilles Deleuze and Felix Guattari, and the theoretical update for the Domain Analysis activity, based on the assumptions defended by the researchers of the Royal School of Library and Information Science, from Denmark, notably Birger Hjørland and Torkild Thellefsen, this thesis discusses the aspects of cognition to be observed to determine the signified in a universe of discourse concerning at facts of a domain, with the aim of increasing the degree of approximation between the realities, of the facts, understood and signified. In this way, a deepening of the process of knowledge acquisition is made, with the criticism at the atomist and structuralist approach, in which terms and relations of the universe of discourse are specified from a direct relation between sign and signified, a conception where the expression exceeds the content, and the space dimension prevails over the time dimension in the process of significance. The study environment used is that referring to computational ontologies, knowledge bases supported on semantic networks and semiotic frames, focused on the phases that go from the understanding of the reality of a domain to that where the significance of terms and relations is treated in order to obtain their respective signified. The research, in its experimental phase, within the proposed reference, analyzed the stages of development of the EDXL-RESCUER ontology, opposing the hypotheses treated in the thesis and the process of development of the ontology, resulting in a critical approach and a corresponding theoretical foundation, complemented by a methodology for Domain Analysis capable of acting in a post-structuralist dimension. The applied research method is qualitative, exploratory, involving updating the state of the art to the presented concepts, from the analysis of an ontology construction project

The evaluation of software process is considered an important tool and commonly used to point out the way, where the organization must apply its efforts to improve processes. In particular, the assessment process proposed by the MA-MPS requires the presentation of objective evidence showing satisfaction of the expected results of processes and results of process attributes of a certain level of maturity. The problem of presentation of objective evidence boils down in the difficulty of the managers responsible for processes in interpreting these results described in the Spreadsheet of Indicators. In addition, there are cases where the implementer of the MPS model not has enough time to monitor the entire fill or revision of the spreadsheet, before submitting it for evaluation of processes. Thus, the occurrence of errors in filling can occur and affect the outcome of the evaluation. In this context, is presented the intelligent assistant, Expert Mentoring, which is designed to support, through questions, the managers responsible for processes in the interpreting the expected results of processes and results of attributes of process described in the Spreadsheet of Indicators, where the final of these questions will suggest direct and indirect indicators, which are more suitable for proof of these results. In this sense, the main results of the evaluation of the Expert Mentoring were: decreasing the number of errors in completing the spreadsheet, and recall the name of the indicator, even before the suggestion of the intelligent assistant. / Fundação de Amparo a Pesquisa do Estado de Alagoas / A avaliação de processo de software é considerada uma ferramenta importante e, comumente, usada para apontar o caminho, onde a organização precisa aplicar seus esforços em prol da melhoria dos processos. Em particular, a avaliação de processos proposta pelo MA-MPS exige a apresentação de evidências objetivas que comprovem a satisfação dos resultados esperados de processos e resultados de atributos de processos de determinado nível de maturidade. O problema da apresentação de evidências objetivas se resume na dificuldade dos gerentes responsáveis pelos processos em interpretar esses resultados descritos na Planilha de Indicadores. Além disso, há casos em que o implementador do modelo MPS não dispõe de tempo suficiente para acompanhar todo o preenchimento ou revisão da planilha, antes de submetê-la à avaliação de processos. Assim, a ocorrência de erros de preenchimento pode acontecer e comprometer o resultado da avaliação. Neste contexto, é apresentado o assistente inteligente, Expert Mentoring, cujo objetivo é apoiar, através de perguntas, os gerentes responsáveis pelos processos na interpretação dos resultados esperados de processos e resultados de atributos de processos descritos na Planilha de Indicadores, onde ao final destas perguntas sugere indicadores diretos e indiretos, que sejam mais adequados para a comprovação destes resultados. Nesse sentido, os principais resultados obtidos com a avaliação do Expert Mentoring foram: a diminuição do número de erros de preenchimento da planilha, e a recordação do nome do indicador, antes mesmo da sugestão do assistente inteligente.

Knowledge is now considered the most important organizational resource, surpassing other resources like land and capital. It has, therefore, been acknowledged that knowledge can play an important role in ensuring an organization’s competitive edge. The purpose of this study was to investigate if knowledge sharing is being used to improve service delivery at the National University of Lesotho’s Thomas Mofolo Library. The researcher held the view that Librarians at Thomas Mofolo Library have different sets of skills which, if combined, could improve service delivery. By not sharing and retaining this existing wealth of knowledge, the researcher claimed that when librarians retire or resign from work, they will certainly take with them the knowledge they possess and the result of this knowledge loss is that the Library may be plagued by an inability to learn from the past experiences, which leads to reinvented wheels, unlearned lessons and the pattern of repeated mistakes. Both qualitative and quantitative methods were employed in the case study design in order to allow for multiple methods of data collection. Data were collected by means of questionnaires and interviews. Questionnaires were administered to all librarians who were available at the time and purposive sampling was used to determine interview participants. Out of the 25 questionnaires administered, 15 were returned, providing a response rate of 60%. The data collected by means of questionnaires was processed using Microsoft Access and analyzed using the Statistical Package for Social Science (SPSS) software (Version 17). The results of analysis were exported into Microsoft Excel for visual presentation and reporting of the results. The data from the interview sessions was analyzed manually by content analysis, using the notes that were taken by the researcher from the respondents during the interview sessions. The findings pointed to the fact that knowledge sharing does occur at TML, although mostly in an informal manner. This was largely due to a number of impediments such as lack of trust and the absence of motivations and rewards. The study concluded by recommending a number of initiatives that could be implemented in order to retain knowledge within the Library. The recommendations included developing a knowledge management strategy and formalizing knowledge sharing by formulating the desired policies. / Information Science / M.A. (Information Science)

O presente trabalho descreve um modelo híbrido para sistemas especialistas classificatórios que Integra os mecanismos de representação de conhecimento de frames e de redes neurais, assimilando qualidades inerentes de cada um dos dois paradigmas. O mecanismo de frames fornece tipos construtores flexíveis para a modelagem do conhecimento do domínio, enquanto as redes neurais possibilitam tanto a automação da aquisição de conhecimento, a partir de uma base de casos, quanto a implementação de aprendizado indutivo e dedutivo. Empregam-se dois tipos de frames para representar o conhecimento relativo aos problemas de classificação: frames de achado e frames de diagn6stico. Os primeiros descrevem os objetos elementares da aplicação que podem influenciar na detecção de certos diagnósticos. Os frames de diagnóstico, por sua vez, são descritos em termos de achados e correspondem as descrições dos problemas de classificação tratados pelo sistema especialista. E utilizado o Modelo Neural Combinatório (MNC), caracterizado por sua capacidade de aprendizado heurístico através do reconhecimento de regularidades observadas no mundo exterior. A integração do mecanismo de representação de frames e do MNC se da ao nível dos frames de diagn6stico, armazenando representações simbólicas do conhecimento das redes neurais que explicitam o conhecimento armazenado nas camadas intermediarias das Últimas. O motor de inferência do sistema opera de maneira integrada sobre ambos os paradigmas, o conexionista e o simbólico. A máquina de aprendizado fornece meios para a automação da aquisição de conhecimento a partir de um banco de casos, bem como para o refinamento do conhecimento do sistema. A máquina de aprendizado dividida em dois componentes principais: o componente de aprendizado indutivo e o componente de aprendizado dedutivo. O primeiro se volta para o componente indutivo e utiliza os métodos de aprendizado por casos de treinamento, descritos para o MNC, para refinar o conhecimento das redes neurais. O componente dedutivo faz use de uma estratégia de reorganização das redes neurais, para gerar novos modelos de resolução de problemas. 0 sistema implementado, denominado HYCONES (the HYbrid CONnectionist Expert System) é apresentado corn detalhamento da integragao entre os paradigmas simbOlico e conexionista. Para validar HYCONES, foram construfdas e testadas 5 versoes de uma base de conhecimento hfbrida para o diagnostico das 3 cardiopatias congenitas mais freqfjentes no Institute de Cardiologia - RS. A primeira delas (B1) baseou-se no conhecimento extrafdo de 22 especialistas em cardiologia pediatrica, de 4 instituicaes brasileiras. Os grafos medios de conhecimento destes especialistas foram diretamente traduzidos para redes neurais MNC, dando origem a primeria versao do sistema. A segunda versao (B2) baseou-se no refinamento da base de conhecimento da versao B1, atraves da reorganizacao e treinamento das redes neurais dos especialistas, corn aplicagao dos metodos de aprendizado dedutivo. A terceira versão (B3) baseou-se no conhecimento de 23 não especialistas, cardiologistas gerais, na sua maioria oriundos do Institute de Cardiologia - RS. A mesma técnica aplicada para a construção da versão B1 foi utilizada para construir esta versão da base de conhecimento hibrida. A quarta versao (B4) corresponde ao refinamento e reorganização da base de conhecimento da versão B3, com as mesmas técnicas de aprendizado dedutivo, aplicadas na construção da versão B2. Finalmente, na quinta versão (B5), a base de conhecimento foi construída, automaticamente, a partir de base de casos, valendo-se do método de explicitação automática de conhecimento do MNC. Os resultados obtidos na validação do sistema comprovaram a eficiência dos métodos de explicitação automática de conhecimento, a partir do banco de casos e de aprendizado dedutivo, por reorganização e treinamento das redes neurais. As principais contribuições deste trabalho foram: a definição do mecanismo para integração de frames a redes neurais; a especificação dos métodos de aquisição automática do conhecimento, a partir de um banco de casos e seu posterior treinamento e aprendizado, por métodos indutivos e dedutivos; finalmente, a incorporação das medidas de especificidade e sensibilidade no processo de explicitação automática do conhecimento do banco de casos. / This dissertation describes a hybrid model for classification expert systems - HYCONES (HYbrid CoNnectionist Expert System), which integrates the knowledge representation mechanism of frames and neural networks, incorporating qualities inherent to each of the knowledge representation paradigms. The mechanism of frames provides flexible constructs for modeling the domain knowledge. The neural networks provide the means for automatic knowledge acquisition and enable the implementation of deductive and inductive learning. The frames mechanism makes use of semantic modeling concepts to make the new model more generic and to provide flexible constructs for knowledge representation. It employs two basic types of frames to represent the knowledge involved in classification problems: finding frames and diagnosis frames. The finding frames describe the elementary objects of the application world that can influence in the detection of certain diagnoses. Diagnosis frames, on the other hand, are described in terms of findings and correspond to the descriptions of the classification problems addressed by the expert system. The Combinatorial Neural Model (CNM) was the architecture chosen for the neural network environment. This model is characterized by its capacity of learning through the recognition of regularities observed in the outside world. The integration of the knowledge representation mechanism of frames with the CNM happens at the diagnosis frames level. The diagnosis frames store symbolic representations of knowledge contained in the neural networks. The system's inference machine operates under an integrated manner with both knowledge representation paradigms. The learning machine provides the means for automatic knowledge acquisition and for knowledge refinement. It is composed by two modules: the inductive and the deductive learning component. The first one uses CNM methods to refine the neural network knowledge. The latter follows a strategy that reorganizes the neural networks, generating new problem solving models. The implemented HYCONES system is presented, detailing the integration between the symbolic and the connectionist approach. To validate HYCONES, five versions of a hybrid knowledge base (HKB) for the diagnosis of the three most common congenital heart diseases at the Institute de Cardiologia - RS were constructed and tested. The first version of the HKB, named B1 for validation purposes, is based on the mean knowledge graphs of 22 experts in congenital heart diseases, from four important Brazilian cardiological institutions. These KG were directly translated to CNM neural networks. The second version of the HKB, named B3 for validation purposes, contains the mean knowledge graphs of 23 non-experts, i.e., general cardiologists, from the Institute of Cardiology of RS. The same method used to built B1 was invoked here: direct translation from the knowledge graphs to CNM neural networks. The second and fourth versions of HKB, named B2 and B4 respectively, consist of trained versions of B1 and B3. To create these trained versions, the CNM neural networks of B1 and B3 were submitted to deductive learning procedures. Finally, the fifth version of HKB, B5, was automatically created, based on the training of the case database. The results obtained and presented in this document point out to the efficiency of the automatic knowledge acquisition methods and the deductive learning techniques implemented, through the reorganization and training of the neural networks. The main contributions of this research are the design of a mechanism for the integration of frames and neural networks, the specification of the learning methods for the hybrid model and the analysis of factors such as specificity and sensibility in the process of automatic knowledge acquisition.