IRLIST Digest April 1, 1991 Volume VIII, Number 12 Issue 55 ********************************************************** I. NOTICES A. Meetings Announcements/Calls for Papers 1. 4th Advanced Course on Artificial Intelligence (ACAI-91) July 1-12, 1991 Bilbao, Basque Country, Spain IV. PROJECT WORK B. Bibliographies 1. Selected IR-Related Dissertation Abstracts ********************************************************** I. NOTICES I.A.1. Fr: acai@labein.es. Re: 4th Advanced Course on Artificial Intelligence (ACAI-91) July 1-12, 1991 Bilbao, Basque Country, Spain INTRODUCTION The European Coordinating Committee for Artificial Intelligence (ECCAI) organizes biannual Advanced Courses on Artificial Intelligence (ACAI) open to interested personnel from all over the world. ACAI-91 will cover the topic of some Issues In Reasoning, through six lectures as explained later in the Program. The speakers of ACAI-91 are well known within the field of AI, even renown references in their areas of expertise; a fair description of their work can not be given within the limited space of the brochure, therefore we are limited to a small presentation of the speakers. INTENDED AUDIENCE The 4th ACAI (1991) is aimed at people who are interested in learning about several advanced topics in Artificial Intelligence, organized around the general topic of Issues in Reasoning. It will be useful to managers who want an overview of the field and its future directions, to researchers who are starting to develop their own systems, and to scholars in related disciplines such as Psychology and Cognitive Science. COURSE SITE Bilbao is the economic capital of the Basque Country, within easy reach of public transportation from all over Europe. It is rated as the Spain's third largest economic nucleus. Bilbao's population is close to one million people and its tradition as a shipping and trading centre goes right back to the beginning of the 14th century. Bilbao is endowed with an excellent communications system. Modern highways and an intricate network of roads and railways connect Bilbao with the rest of the Basque Country, the Mediterranean, Madrid and France. Air communication is provided through Sondica International Airport which is located 8 km from the center of Bilbao and Vitoria- Gasteiz (40 km); also nearby is Biarritz (120 km) in the French Basque Country. The Basque Country is a picturesque green and hilly region located by the Gulf of Biscay, and is rich in local traditions and culture including one of the oldest languages still alive and unrelated (so far) to any other language known. The seaside, mountains, small villages, traditions and pleasant weather mix to provide enjoyable surroundings for leisure and weekend journeys, some of which will be arranged for the participants. PROGRAM COMMITTEE I$aki Etxebarria, LABEIN (IDEIA Dept.) Mikel Emaldi, LABEIN (IDEIA Dept.) JosFrancisco Garijo, Telef"nica I + D Ram"n L"pez de M ntaras, Centre d'Estudis Avan CEAB-CSIC Juan Pazos, Univ. PolitTim Smithers, University of Edinburgh PROGRAM ISSUES ON REASONING Course 1.- TRENDS IN COOPERATIVE DISTRIBUTED PROBLEM SOLVING Victor R. Lesser University of Massachusetts at Amherst SUMMARY Cooperative distributed problem solving (CDPS) research investigates the territory where the fields of AI and distributed processing overlap. The primary goal of CDPS research is to develop the knowledge and reasoning techniques that a loosely-coupled network of sophisticated problem-solving nodes needs to cooperate effectively in solving a complex, distributed problem. A problem well-suited to a CDPS approach consists of a set of interdependent subproblems that arise because of spatial, temporal, and functional distributed of data, knowledge, and processing capabilities. Typical CDPS application areas include distributed interpretation, distributed planning and control, cooperating expert systems, and computer-supported human cooperation. We survey major CDPS approaches and current research directions. CURRICULUM VITAE Victor R. Lesser received his B.A. in Mathematics from Cornell University in 1966, and the M.S. and Ph.D. degrees in Computer Science from Stanford University in 1969 and 1972, respectively. He is currently a professor of Computer Science at the University of Massachusetts/Amherst. His major research focus is on the control and organization of complex AI systems. He is considered a leading researcher in the areas of blackboard systems, distributed AI, and real-time AI. He has also made contributions in the areas of diagnostics, plan recognition, and intelligent user interfaces. Course 2.- CASE BASED REASONING AND LEARNING Introduction to CBR Methods Janet L. Kolodner Georgia Institute of Technology SUMMARY Case-based reasoning (CBR) means solving problems by adapting known solutions to previous problems. It is an alternative to rule-based reasoning, which solves problems by chaining many independent rules together, reasoning "from first principles," so to speak. CBR is of interest in psychology and cognitive science, where research has suggested that people understand and reason with fairly large "chunks" or templates of knowledge, and that experts learn from and refer to prior cases when approaching new problems. For real experts, few problems are totally new. CBR also has a number of attractive features for implementing practical expert systems, in the areas of (1) knowledge acquisition (entering sample solutions is easier than constructing the rules that would generate them), (2) speed (a two-step process of retrieve and adapt replaces the combinatoric chaining of rules), and (3) explanation (solutions can be justified by showing the prior cases they are based on). The goal of this introductory tutorial is to describe how CBR works, what its strengths and weaknesses are, and where current research in the field is heading. NOTE: Earlier versions of this tutorial have been presented by Janet Kolodner and Chris Riesbeck at IJCAI-89 and AAAI-90 to audiences of approximately 150. It has also been presented by Janet Kolodner as part of Georgia Tech's Continuing Education program on Knowledge-Based Systems and at several private locations in the US. CURRICULUM VITAE Janet L. Kolodner is a Professor at the Georgia Institute of Technology, where she has directed the development of numerous case-based reasoning projects, including NEGOTIATOR, PERSUADER, CAS, and JULIA, in domains ranging from psychiatric diagnosis to contract negotiation to catering. She is the author of "Retrieval and Organizational Strategies in Conceptual Memory", co-editor (with Riesbeck) of "Experience, Memory and Reasoning", and editor of the proceedings of the DARPA-sponsored workshop on Case-based Reasoning in May 1988, at Clearwater Beach, Florida. She was program chair for that workshop, and on the program committee for the AAAI-88 workshop on Case-based Reasoning in St. Paul, Minnesota. She has written dozens of papers, mostly on issues related to case-based reasoning and is currently writing a textbook on CBR. She co-taught the CBR tutorial at IJCAI-89 and AAAI-90 and has taught it in several private locations as well. Course 3.- TEMPORAL REASONING Murray P. Shanahan (1) & Yoav Shoham (2) Imperial College (1), Stanford University (2) SUMMARY The subject of Temporal Reasoning will be introduced through the Situation Calculus, using the Blocks World as sample domain, to be followed by several attempts to solve the frame problem, the Hanks- McDermott problem and several attempts to its solution (Shoham and Lifschitz). In the second half, the approach will shift from the logical to the ontological issues, studying those that try to remove the restrictions of the Situation Calculus, namely those of Allen, McDermott, Hayes'histories and Kowalski and Sergot's Event Calculus. CURRICULUM VITAE Murray Patrick Shanahan, B. Sc. in Computer Science (Imperial College 1984), Ph. D. (University of Cambridge 1988). His Ph.D. research concerned the use of dependencies to control search and inference. In particular, selective backtracking techniques for constraint satisfaction and their application to theorem proving. Currently his line of reseach is Temporal Reasoning. He has designed and built a system for explanation (reasoning from effects to causes) which uses abduction, and has extended Kowalski and Sergot's Event Calculus to deal with continuous change. He is currently working on the relationship between qualitative and temporal reasoning. Yoav Shoham received his PhD from Yale University in 1987, and later joined the Computer Science Department at Stanford University as an Assistant Professor. He has recently been awarded the first Finmeccanica Chair at Stanford university. Shoham's research interests include knowledge representation, temporal reasoning, control and multi-agent systems, and he has published extensively on these topics. Shoham's second book, Artificial Intelligence Programming in Prolog, is scheduled to appear this summer. Course 4.- GENERAL COGNITIVE ARCHITECTURES Luc Steels and Walter Van de Velde VUB - Free University of Brussels SUMMARY The goal of research on General Cognitive Architectures is to design a small set of mechanisms that can account for a wide variety of aspects of intelligence. In a strong sense the aim is to model and explain the mental mechanisms that underlie human behaviour. In a more general sense, and this is the one we adopt here, the human mind is not the major standard of success. Instead, coverage and scope of functionality as well as practical value are the main criteria for evaluation. These architectures are general architectures for intelligence, not necessarily based on the human model. In this tutorial we provide a brief overview of the field's goals and history. We then focus on some concrete architectures. Our goal is to illustrate the variety of underlying goals, principles and assumptions, and the way in which these are reflected in the actual designs. We also investigate the potential role of general cognitive architecture in real- world AI applications, and examine current trends in this rapidly evolving research area. CURRICULUM VITAE Luc Steels studied Computer Science at M.I.T. after a doctorate in computational linguistics from the University of Antwerp (UIA). He was a researcher at the Artificial Intelligence Laboratory at M.I.T. after which he joined Schlumberger research labs (U.S.); currently he is professor of Computer Science and Artificial Intelligence at the Free University of Brussels (VUB) and director of the AI Lab which he founded in 1983. He is president of the Belgian Association for Artificial Intelligence and active in various international organizations and was program chairman of the largest European AI Conference (ECAI) in 1986. His research topics span from natural language, knowledge representation, expert systems, robot architectures and subsymbolic computation. Walter Van de Velde received a degree in mathematics from the University of Antwerp (UIA 1980) and a Ph. D. in sciences from the Free University of Brussels (VUB 1988) with his thesis "Learning from Experience" (awarded the IBM-Belgium prize). He is currently associated to the Artificial Intelligence Laboratory of the Vrije Universiteit Brussel. His major focus of research is in Machine Learning, Second Generation Expert Systems and General Architectures of Intelligence. During the summer of 1989 he spent four months at Carnegie-Mellon University, and from November 1989 one year as invited scientist at the CSIC Center for Advanced Studies of Blanes, Spain. Walter Van de Velde has an extensive list of publications, is editor of "Towards Learning Robots" (MIT Press, 1991), co-author of various research proposals, and chairs the 1991 European Summer School on Machine Learning (Belgium). Course 5.- VALIDATION OF KNOWLEDGE BASED SYSTEMS Enric Plaza and Pedro Meseguer Centre d'Estudis Avan SUMMARY The validation of expert systems raises new issues that set them apart from mere software program development. The roles played by logic deduction, the acquired expertise, the representation scheme used and the inference regimes are reviewed first. Then the validation of KBS's is characterized in three types: validation of structural properties of the Knowledge Base (KB), validation of the dynamic properties of KBS performance, and evaluation of KBS characteristics. The more representative of the currently developed techniques and systems for KBS validation are explained in all of the three provinces of validation. CURRICULUM VITAE Enric Plaza received his Ph.D. in Computer Science from Universitat Politecnica de Catalunya in 1987. Research Scientist since 1988 in CSIC (Spanish Scientific Research Council) at the AI Lab in Centre d'Estudis Avanseveral papers on knowledge acquisition, machine expert systems validation are published in journals and books. He is project leader of the ESPRIT-II VALID Project "Validation Methods and Tools for Knowledge-based Systems" and teaches postgraduate AI courses on Machine Learning and on Expert Systems in Universitat Politecnica de Catalunya (Barcelona) and Universidad de Deusto (Bilbao). Pedro Meseguer received his M. Sc. in Physics from Universidad de Zaragoza in 1980 and his M. Sc. in Computer Science from Universitat Politecnica de Catalunya in 1987. He is currently a researcher in the CSIC (Spanish Scientific Research Council) at the IA Lab in the Centre d'Estudis Evanresearch topics are constraint satisfacti system validation, on which he has published several papers. He is currently involved in the ESPRIT-II VALID project: "Validation Mehtods and Tools for Knowledge-Based Systems" Course 6.- KNOWLEDGE ELICITATION TECHNIQUES Peter Smith Sunderland Polytechnic SUMMARY It is generally recognised that the major bottleneck in the development of knowledge based systems is that of knowledge elicitation. It is during the knowledge elicitation phase that the majority of the problems and difficulties can arise. However, it is this very phase which is probably the most important for the development of a successful knowledge based system. This lecture will introduce the major problems of knowledge elicitation and present a review of current techniques and tools. In particular, the approaches of interviews, repertory grids, protocol analysis simulation and machine induction will be presented and the advantages and disadvantages of each technique discussed. Current tools to support these techniques will also be reviewed. CURRICULUM VITAE Peter Smith received his BSc (Honors) in Combined Studies of Science in 1978 and PhD in Modelling and Simulation in 1981, both at the Sunderland Polytechnic, where he is currently Reader. Dr. Smith has over seventy publications in the areas of Expert Systems, Software Engineering, Decision Support Systems, and others, and has been involved in the last five areas in Research and Consultancy in several areas of Modelling, Expert Systems and Knowledge Elicitation. FOR FURTHER INFORMATION CONTACT ******************************** LABEIN (ref. ACAI-91) IDEIA Dept. P.O. Box 1234 E-48080 Bilbao - SPAIN Phone: + 34 - 4 - 441 93 00 Fax: + 34 - 4 - 441 17 49 E-mail: acai@labein.es ********************************************************** IV. PROJECT WORK IV.B.1. Fr: Susanne Humphrey Re: Selected IR-Related Dissertation Abstracts The following are citations selected by title and abstract as being related to Information Retrieval (IR), resulting from a computer search, using BRS Information Technologies, of the Dissertation Abstracts Online database produced by University Microfilms International (UMI). Included are UMI order number, title, author, degree, year, institution; number of pages, one or more Dissertation Abstracts International (DAI) subject descriptors chosen by the author, and abstract. Unless otherwise specified, paper or microform copies of dissertations may be ordered from University Microfilms International, Dissertation Copies, Post Office Box 1764, Ann Arbor, MI 48106; telephone for U.S. (except Michigan, Hawaii, Alaska): 1-800-521-3042, for Canada: 1-800-268-6090. Price lists and other ordering and shipping information are in the introduction to the published DAI. An alternate source for copies is sometimes provided. Dissertation titles and abstracts contained here are published with permission of University Microfilms International, publishers of Dissertation Abstracts International (copyright by University Microfilms International), and may not be reproduced without their prior permission. AN University Microfilms Order Number ADG90-19409 AU NOFAL, ADIL MOHAMMED. TI THE USE OF INFORMATION SOURCES BY FACULTY IN THE PHYSICAL SCIENCES AND SOCIAL SCIENCES AT KING ABDULAZIZ UNIVERSITY. IN University of Pittsburgh Ph.D. 1989, 185 pages. DE Library Science. Education, Curriculum and Instruction. AB The purpose of this study was to investigate the use of formal and informal sources of information by the Saudi faculty at King Abdulaziz University and to assess whether there was a difference in information sources use by the physical science and social science faculty members at three stages of research and two stages of teaching activities. The three stages of research activities were preliminary or proposal stage, data collection stage, and data analysis stage while the two stages of teaching were developing a new course and lecture preparation. Data were collected by means of self-administered questionnaires to 254 faculty members, and face-to-face interviews with six other faculty members. Data analysis was based on 159 valid questionnaires, 97 from the physical sciences and 62 from the social sciences and all six interviews. Data from the returned questionnaires were analyzed using descriptive statistics (means, standard deviation and proportions), inferential statistics (t-test) which was used to test the hypothesis at the.05 level of significance. The findings of this study showed that faculty members in the physical sciences and social sciences use both formal and informal sources of information, but they rely more on formal sources. Among the formal sources, they rely heavily on personal files, journals, and books, while personal contact with colleagues is considered as the most important informal source of information. Also it was found that the 2 groups of faculty members use similar sources of information at different stages of a research project. Another important finding of this study was that faculty members in the physical sciences and social sciences differ significantly from each other in their use of formal sources of information at the preliminary or proposal stage, the data collection stage of a research project, and when developing a new course. AN University Microfilms Order Number ADG90-33260 AU RAZZAGHI, FARZANEH. TI HOW SELECTED FACULTY IN SEVEN MEDICAL SCHOOLS IN TEXAS MEET THEIR INFORMATION NEEDS. IN Texas Woman's University Ph.D. 1990, 121 pages. DE Library Science. Health Sciences, Education. Education, Higher. AB The use of seven medical libraries by members of the faculties of seven medical schools in Texas was studied by questionnaire. The study's results describe the population of users and the kinds of sources they use in an information search, and identify some of the factors which influence the population's choice of kinds of sources, and their use of the medical libraries in their institutions. The results explore primarily the information searching habits of medical faculty in certain rank with certain degree. Most respondents found the information they sought through a personal trip to the library. It was found that the most frequently sought information related to research interests. Interestingly, in spite of the fact that the majority of the responses referred to the use of formal printed sources (usually journals), the indicated use of informal information sources would justify further research in this area. The use to which the information was put was most often for research rather than for clinical activity, teaching, or administration. The high use of informal sources suggest that the type of work may influence the method of search, especially the use of co-workers in informal communication. There was a low dependence on librarians, but a high dependence on librarians, but a high dependence on the card catalog. Of those who went to the library, more than half used Index Medicus, other on-line indexes, and journal collections as their major access points. Users indicated that journals should be shelved accessibly, with Index Medicus nearby, that books might be shelved less accessibly, and that provision of adequate photocopying facilities was important. AN University Microfilms Order Number ADG90-33272 AU RUDDY, MARY KAREN. TI A SOCIOMETRIC ANALYSIS OF INFORMATION-SEEKING BEHAVIOR, INFORMATION SOURCES, AND INFORMATION NETWORKS IN BOARDS, COMMITTEES AND COMMISSIONS IN A SMALL RURAL IOWA COMMUNITY. IN Texas Woman's University Ph.D. 1990, 148 pages. DE Library Science. Information Science. Mass Communications. Political Science, General. AB The sources ultimately used by information-seeking individuals may be understood by examining influences present in the information network in which these persons are embedded. That understanding is expanded when the invisible links in this system are studied, beyond the first-choices of the information-seeker. Roger's and Kincaid's Network Analysis techniques allow the study of these links using sociometric diagrams. The viability of this application is tested by charting the choices made by individuals elicited from a population of randomly selected boards, committees and commissions in Emmetsburg, Iowa. In this study, these diagrams reveal the network "roles" played by these participants. Identifying Opinion Leaders, Isolates and Liaisons reveals the invisible information network which leads from individual to individual to ultimate information "stores". Previous research studies in Library Science chronicle the increasing jeopardy faced by libraries in our society. Previous research in Communication Studies found that average citizens get information from other people when seeking answers to questions that will enlighten them and aid in their decision-making. The current study questions both of these positions by revealing the linkage from people back to the print-on-paper or electronic "stores" of literature collected in public, personal, professional, governmental, medical and law "libraries". Those seeking information are generally not aware that the information desired ultimately comes from collected bodies of literature through the "information chain" of individuals. A present danger is that these sources may be bypassed in budgetary considerations because they are not perceived as important and essential to the provisions of accurate useful information for public policy decision-making. ********************************************************** IRLIST Digest is distributed from the University of California, Division of Library Automation, 300 Lakeside Drive, Oakland, CA. 94612-3550. Send subscription requests to: LISTSERV@UCCVMA.BITNET Send submissions to IRLIST to: IR-L@UCCVMA.BITNET Editorial Staff: Clifford Lynch lynch@postgres.berkeley.edu calur@uccmvsa.bitnet Mary Engle engle@cmsa.berkeley.edu meeur@uccmvsa.bitnet Nancy Gusack ncgur@uccmvsa.bitnet The IRLIST Archives will be set up for anonymous FTP, and the address will be announced in future issues. These files are not to be sold or used for commercial purposes. Contact Mary Engle or Nancy Gusack for more information on IRLIST. The opinions expressed in IRLIST do not represent those of the editors or the University of California. Authors assume full responsibility for the contents of their submissions to IRLIST.