IRList Digest Monday, 1 December 1986 Volume 2 : Issue 65 Today's Topics: Abstracts - NSF IST Awards for Fiscal Year 1986 - Part 3 of 5 News addresses are ARPANET: fox%vt@csnet-relay.arpa BITNET: foxea@vtvax3.bitnet CSNET: fox@vt UUCPNET: seismo!vtisr1!irlistrq ---------------------------------------------------------------------- Date: Fri, 21 Nov 86 18:38:14 est From: vtopus!fox (Ed Fox) Subject: Information on NSF awards, sent by J. Deken at NSF Fiscal Year 1986 Research Projects Funded by the Information Science Program (now Knowledge and Database Systems Program) Part 3 of 5 IST-8512108 $99,394 - 12 mos. Hans Kamp University of Texas at Austin Logic Representation of Attitudes for Computer Natural Language Understanding - - - This project develops a theory of attitudes based on logical and computational models. Modeling attitudes is important to information science because automated natural language processing systems have proved unable to understand unrestricted narrative and exposition without additional contextual information such as the attitudes of speaker and listener. In particular, the dynamics of attitudes, the mechanisms by which they change in response to information input, are being illuminated. _____ IST-8644864 $35,721 - 12 mos. Abraham Kandel Florida State University Analysis and Modeling of Imprecise Information in Uncertain Environments - - - A problem of considerable inportance in the development of expert systems is the ability to represent imprecise data. This research is based upon the representation of imprecise data by possibility distributions. The PI is interested in developing a theory of statistics in the framework of possibility distributions. Particular emphasis in this research is given to the development of a "typical" value associated with a possibility distribution. This value is analagous to the expected value in the framework of probability theory. The PI will study the properties of this typical value and its usefulness solving differential equations. Just as the expected value provides a means for summarizing probabilistic information the significance of this research lies in the fact that the so called "typical" value will provide a means for summarizing possibilistic data. This in turn will lead to simplification in handling imprecise information of the type represented by possibility distributions. _____ IST-8542811 $65,414 - 12 mos. R.L. Kashyap Purdue University Research on Inference Procedures with Uncertainty - - - The development of inference procedures which can be used with uncertain information is an important issue in the construction of expert systems. Ad hoc and intuitive approaches to inference are used in many expert systems; this research investigates several systematic and rational inference procedures which could be used in expert systems. The theoretical foundations include Bayesian theory, Dempster and Shafer's theory of evidence, and fuzzy reasoning. A working expert system for structure damage assessment is used to test the inference procedures. The significance of this research is that it will contribute to the important problem of handling uncertain information in expert systems by developing a rigorous approach to uncertainty. _____ IST-8644676 $74,752 - 12 mos. George J. Klir State University of New York at Binghamton Possibilistic Information: Theory and Applicability - - - Information theory based on the Shannon entropy has been very successful in dealing with many systems problems, but its limitations are becoming increasingly clear. A development of new information theory is proposed, one which is based on possibility theory rather than probability theory. Such a theory is fundamentally different from the Shannon information theory since it involves imprecision that is not statistical in nature. As such, it is more suitable for dealing with the wide range of systems problems that are often referred to as problems of organized complexity (as contrasted with problems of disorganized complexity, where statistical methods perform well). In analogy with its probabilistic counterpart, possibilistic information is defined in terms of the underlying notion of possibilistic uncertainty. The latter is a measure that possesses counterparts of all properties of the Shannon entropy and, in addition, has an important property of monotonicity. The proposed re- search includes a mathematical investigation of some fundamental issues involv- ing the possibilistic information, a development of basic laws and principles of the possibilistic information theory, and a study of its applicability in a broad range of relevant areas. The ultimate goal is to demonstrate the fundamental complementarity of the two information theories and to characterize the two domains for which application of one or the other information theory is superior. This will be accomplished analytically, when feasible, or by experiments simulated on the computer. _____ IST-8552925 $54,250 - 12 mos. Richard E. Korf University of California at Los Angeles Presidential Young Investigator Award : Machine Learning - - - Richard Korf's research focuses on mechanisms by which intelligent computer systems can learn the large amounts of specific knowledge necessary to solve realistic problems. In particular, the research explores the ways in which specific knowledge can help to reduce the amount of trial-and-error search needed by a computer for problem solving. A central thrust of this research is machine learning - the processes by which relevant knowledge is automatically acquired by the computer rather than being programmed by a human. The significance of machine learning and problem solving for the future use of computers can hardly be overstated. Practical autonomous computer systems of the future will need far more knowledge than humans can explicitly transfer, and so the computers themselves will have to learn independently. _____ IST-8518307 $15,750 - 12 mos. Donald H. Kraft Louisiana State University Travel to the ACM Conference on Research and Development in Information Retrieval: Pisa, Italy; September 8-10, l986 - - - This award supports travel funds for fifteen U.S. participants to the l986 ACM Conference on Research and Development in Information Retrieval in Pisa, Italy on September 8-10, l986. This conference is sponsored by the Consiglio Nazionale delle Ricerche (Italian National Research Council), in cooperation with the Association for Computing Machinery Special Interest Group on Informa- tion Retrieval and several other professional organizations. The conference scope includes theory, methodology and applications of information retrieval, with particular emphasis on new emerging application areas. This conference provides an opportunity for strengthening the quality of information retrieval research, as well as facilitating international communication in this area. _____ DCR-8602665 $45,720 - 12 mos. Benjamin J. Kuipers University of Texas at Austin Knowledge Representations for Expert Causal Models - - - The goal of this research is to design and build a computer program to under- stand a realistic description of a medical case, construct a detailed model of the internal state of the patient and its evolution, and answer questions about the pathophysiology of the case, concentrating on the domain of salt and water balance disorders in the kidney. During the past year, progress has been made in several directions. The theoretical study of qualitative simulation has been completed. QSIM, a new version of the qualitative simulation algorithm based on the theoretical work, has been implemented, runs very efficiently and is being distributed to interested researchers. The Evoker, an associative hypothesis-generator, has been redesigned and reimplemented to make it a more general-purpose tool for knowledge engineering. A preliminary version of the interaction between QSIM and the Evoker that constitutes the core of the RENAL program has been designed and implememted. The size and scope of the knowledge bases for both QSIM and the Evoker are steadily being augmented. This project focuses on completing the knowledge bases, codifying a set of principles for acquiring and validating its contents. The pattern of interactions are being completed between QSIM and the Evoker, concentrating particularly on the matcher between the different descriptive terms used by the two reasoners, to make RENAL into a complete diagnostic system. _____ RII-8600412 $10,000 - 12 mos. Jill H. Larkin University of California at Berkeley Developing the Instructional Power of Modern Personal Computing - - - This project will provide the principal investigator with the opportunity to conduct advanced research not possible at her home institution and to interact with students and colleagues through teaching, counseling, and mentoring activities in a way that demonstrates her successful career in science. (Visiting Professorships for Women) _____ IST-8600412 $10,000 - 12 mos. Wendy G. Lehnert University of Massachusetts at Amherst Presidential Young Investigator Award: Natural Language Computing Systems - - - Advanced computer information processing systems which manipulate natural language (normal human verbal discourse) are composed of complex sets of interacting sub-programs. Among these are programs which dynamically create new data structures and new sub-programs according to rules for manipulating the initial input. Such dynamic "learning" systems are inherently difficult to understand and maintain; especially for those not involved in the original de- sign. The goals of this project are to develop tools for automatically analyzing the data dependencies and tracking interactions and isolating potential effects due to system modifications. The importance of the effort is in its potential to facilitate reliable and practical natural language information systems. _____ IST-8603697 $5,000 - 12 mos. Michael E. Lesk Bell Communications Research Workshop on Document Generation Principles - - - This award is for a three day workshop on document generation principles. The workshop brings together four groups of researchers with closely interrelated, but historically separately investigated interests: The generation of natural language documents by computer; The production of manuals and other documents by people; The search of natural language texts by computer; and The evaluation of human) writing for clarity and readability. This workshop, as an interdisciplinary forum for the interaction of questions and methods about natural language production and searching, provides a signifi- cant perspective beyond document generation systems to more general "knowledge transfer" systems. Artificial intelligence programs of the near future will require such knowledge transfer capability as a widespread infrastructure. _____ IST-8602765 $76078 - 12 mos. R. Duncan Luce Harvard University Measurement: Axiomatic and Meaningfulness Studies - - - Information from phenomena in the real world cannot be managed unless it is converted into a representation, for use by people and computers. This conversion from phenomena to representation is the fundamental process of quantification or measurement. Since quantification is essential, it is of paramount importance for information systems that it be done well. This research investigates the properties of some mathematical systems of quantification, in order to illuminate the concept of "meaningfulness." In particular, the research shows how quantification must be managed in many cases, so that seemingly equivalent choices of "representation" do not inadvertently lead the people or computers which use these representations into unexpectedly different reasoning and conclusions. The research is significant in that structural and analytic studies of "meaningfulness" may well illuminate the empirical calibration of the representation and reasoning schemes used by many modern computing systems. _____ IST-8444028 $62,500 - 12 mos. David Maier Oregon Graduate Center Presidential Young Investigator Award: Foundations of Knowledge Management Systems - - - The goal of this project is to integrate recent findings from research on database structure, knowledge representation, programming languages, expert systems, graphics, and information retrieval in order to build a foundation for building knowledge management systems (KMS). A KMS is an intelligent database system which allows extraction and manipulation of information without reference to storage format, mode of representation (text, graphics, audio, etc.) or the state of organization of the data. The significance of the project will be a substantial improvement in ease of use and intelligence capability of data management systems. _____ IST-8604977 $46,956 - 12 mos. David Maier Oregon Graduate Center Automatic Generation of Interactive Displays - - - This project seeks a way to construct user-system interfaces with less effort than is currently required and at the same time to achieve a higher degree of uniformity among interfaces than is currently found. The approach taken is to relate a source object (such as some information in an application program) and a view object (such as a graphical image on the computer's screen) by means of a "filter". By this term is meant a program which can satisfy some constraint on the two objects it relates. Filters are to be bidirectional: if the source changes, the view must be updated; if the view is edited, the system must find a corresponding change to the source to keep the constraint satisfied. An important property of filters is that they can be composed in a variety of ways, so that new filters can be defined using other filters as parts. One objective of the research is a formal theory which will allow transformations between equivalent systems of filters. The results are to be tested by implementing a prototype system for generating user interfaces. The research is carried out in collaboration with Alan H. Borning, of the University of Washington, and Ralph London, of Tektronix, Inc. _____ IST-8642813 $25,500 - 12 mos. Gerald S. Malecki Office of Naval Research Committee on Human Factors - - - This award is for the National Science Foundation to join with other agencies of the Feceral Government in providing sponsorship support for ongoing activities of the Committee on Human Factors of the National Academy of Sciences. This Committee was established in October l980 and is currently sponsored jointly by various agencies of the Department of Defense and by the National Aeronautics and Space Administration. The Committee's principal objectives are to provide new perspectives on important theoretical and methodological issues to strengthen the scientific bases of human factors research and development. _____ IST-8606187 $19,650 - 12 mos. James L. McClelland Carnegie-Mellon University Workshop of Parallel Distributed Processing in Information and Cognitive Research (Washington, D.C. ; February 28 - March 1, 1986) - - - This project is a workshop for researchers from several areas supported by Information Science. The workshop explores the interaction of issues in artificial intelligence, cognitive psychology, and cognitive neuroscience with parallel distributed process (PDP) models. (PDP models are ones in which information processing is accomplished through the interaction of a large number of simple, highly interconnected processing units). The primary goal of the workshop is to develop an agenda for researchers in Cognitive/Information Science. The goal is achieved in the following way: At the workshop researchers provide an overview of the relevant issues, examine existing PDP models in light of recent data, and consider challenges posed by high-level phenomena studied in the area of artificial intelligence. The workshop will benefit the study of Information Science by bringing together top researchers and allowing them to look at newly emerging issues and currently unanswered questions. ------------------------------ END OF IRList Digest ********************