IRList Digest Monday, 1 December 1986 Volume 2 : Issue 66 Today's Topics: Abstracts - NSF IST Awards for Fiscal Year 1986 - Part 4 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 4 of 5 IST-8451438 $37,500 - 12 mos. Kathleen R. McKeown Columbia University Presidential Young Investigator Award: Natural Language Interfaces - - - This research investigates computer generation of natural language, focusing on problems which arise in the context of interactive dialogues with users of computer-based information systems. The primary goal is to develop methods for tailoring explanations for users of expert systems. A secondary goal is to investigate problems involved in the choice of words and syntactic structures. In both cases information about users and their goals is used to constrain and guide the output. The significance of this research lies in its potential to improve user interfaces to computer based information systems. _____ IST-8520217 $115,220 - 12 mos. Douglas P. Metzler University of Pittsburgh An Expert System Approach to Syntactic Parsing and Information Retrieval - - - This research examines the role which discovering language structure beyond the meaning of isolated words can play in improving the effectiveness of computerized searching for information. This enhanced language use would be a substantial improvement over present day information retrieval systems, where the task of finding documents or other relevant source materials is largely done based on the match (or combination of matches) between individual words in a searcher's query and words in the target materials. To extract linguistic information beyond the simple meaning of individual words, the principal investigator's approach to automated language processing is implemented, using an "expert-system" architecture in which parallel processors independently analyze textual material while sharing information through a "blackboard." Language information discovered by this expert system is then analyzed and evaluated for its effectiveness in assisting information retrieval tasks. This research is significant both in the light it will shed on the utility of syntac- tic information in enhancing information retrieval, and in the ancillary devel- opment of parsers and other linguistic tools in advance LISP-based environments. _____ IST-8512736 $137,503 - 24 mos. David Mumford Harvard University The Parsing of Images - - - Processes in the human visual system are often divided into two categories, local processes which are data-driven and global processes which are knowledge-driven. Some early visual processing, however, seems to involve analyzing shapes and structure without regard to this categorization; analyses are made which are global but not knowledge driven. An example of a computer system which tests this hypothesis is Marr's system in which the Gestalt laws of organization are used to generate a global analysis in a data-driven way. In this project three closely related problems which arise when computer systems are constructing visual analyses are investigated: (1) The development of an algorithm to divide an image into maximally homogeneous regions with the shortest possible edges; (2) the development of methods of analyzing texture by classifying the micro-shapes out of which it is composed and the links between shapes given by repetion and similarity; (3) the development of computer algorithms implementing the Gestalt laws. The research will further understanding of the human visual system and ultimately will aid in the development of computer vision systems. _____ IST-8604282 $2,447 - 12 mos. Kent Norman University of Maryland at College Park Developing an Effective User Evaluation Questionnaire for Interactive Systems - - - The purpose of this research is to develop a standard test for measuring the satisfaction users derive from a given interactive computer system. The test is designed to measure subjective ratings of system quality and performance as well. Additionally, the test will be valid for a wide range of systems. The test is constructed by defining ease-of-use as being composed of a large number of underlying facets such as the quality of the visual display and the speed at which the system responds to users. Questions are developed to reflect each of these aspects. The questionnaire is a tool to aid the design of computer systems. It also is a tool for assessing potential areas of system improvement and for comparing the ease-of-use of a set of computer systems. This research therefore contributes to the development of more sophisticated computer systems. _____ IST-8645347 $80,492 - 12 mos. Donald E. Nute University of Georgia Discourse Representation for Natural Language Processing - - - The purpose of this research is to integrate two recently developed theories of discourse, discourse representation and conversational scorekeeping, and to develop from them computationally feasible procedures for natural language understanding. The procedures developed are applied in experimental computer programs for automated extraction of data from texts and in natural language front ends for database construction and querying. The integration of these two theories of discourse should make possible considerable more powerful techniques for discourse understanding then have so far been applied to computer processing of natural language. _____ IST-8645348 $49,211 - 12 mos. Donald E. Nute University of Georgia Hypothetical Reasoning and Logic Programming - - - This research uses work in the area of philosophical logic to enhance the theoretical basis of logic programming systems. Non-classical intensional conditional logic is used to represent subjunctive conditionals and reason about them nonmonotonically. The theoretical results are incorporated into two existing logic programming systems, PROWESS and N-PROLOG. The significance of this research is in its potential for improving the techniques of knowledge representation and automated reasoning used in artificial intelligence systems, such as expert systems. _____ IST-8642477 $86,969 - 12 mos. Robert N. Oddy Syracuse University Representations for Anomalous States of Knowledge in Information Retrieval - - - A major goal of information retrieval research is the discovery of effective ways of identifying the relatively few documents in a large collection which an inquirer will find useful in solving a problem. The ASK (Anomalous State of Knowledge) hypothesis is that an information need arises from an anomaly in an individual's state of knowledge and that in many cases a user can not specify what is needed to resolve the anomaly. This project investigates representations of ASK's by collecting statements from system users, analyzing them automatically, comparing the representations to the perceptions of the users, and determining the relationship between them and the structures of both individual documents and collections of documents. The implications of the results for strategy choices in information retrieval systems are also considered. The significance of this research lies in its potential to improve the performance of information retrieval systems. _____ IST-8609201 $44,905 - 12 mos. Daniel Osherson Syracuse University A Computational Approach to Decision-Making - - - The purpose of this collaborative research is to develop a theory of how humans make decisions. The theory is developed as a computational theory, using the representations and processes currently prevalent in cognitive science and artificial intelligence research. The theory explains the formation of complex concepts. Additionally, it accounts for people's incorrect estimates of the likelihood with which events will occur. The theory is tested through psychological experimentation. Knowledge about how people make decisions and estimate probabilities is useful to all decision makers. Additionally, a computational theory of the process allows the development of computer based decision-making aids. _____ IST-8544976 $197,055 - 12 mos. Charles Parsons and Isaac Levi Columbia University The Structure of Information in Science: Fact Formulas and Discussion Structures in Related Subsciences - - - The purpose of this research is to develop a method for reducing the information contained in a scientific article including data presentations, discussions, and conclusions to a system of formulas adequate to describe the information for any given subscience. This system of formulas can be regarded as a "grammar" or informational structure of a subscience. This work is possible because methods now exist for studying the word-combinations in scientific writing, which carry almost all the information in science. This permits the discovery of fact-structures, argument structures, and the relation of facts to arguments and the like. These methods are not semantic. Because they are based on observed work-combinations they are objective, testable by others, applicable in the real world, and eventually amenable to computer processing. The reduction of the information to formulas has many advantages for computerized information storage, manipulation, and retrieval besides the obvious one of reduced storage space. It will permit the manipulation of the information in order to retrieve facts, inferences, and hypotheses not previously possible. The significance of this work lies in its ability to provide a basis for the computer processing of the specific information in scientific reports. The research team including Zellig Harris have made numerous and important contributions to the field and the environment at Columbia provides a supportive setting for such research. _____ IST-8642841 $12,000 - 12 mos. William J. Rapaport State University of New York - System Office Logical Foundations for Belief Representation - - - This research investigates the theoretical foundation for the design of computer systems capable of representing and reasoning about the cognitive attitudes of intelligent agents. The agents include users, other artificial intelligence systems, and the system itself. The cognitive attitudes include beliefs, knowledge, goals, and desires. Particular attention is given to the ability to handle nested beliefs and self-referential beliefs. The theoretical approach used is based upon recent work in philosophical logic. The significance of this research lies in its contribution to the development of intelligent computer based systems. _____ IST-8644984 $62,500 - 12 mos. James A. Reggia University of Maryland at College Park Presidential Young Investigator Award: Abductive Inference Models in Artificial Intelligence - - - Abductive inference refers to a mode of reasoning in which a plausible solution to a problem is derived from related knowledge on a nondeductive basis. A domain-independent model of abductive inference, parsimonious covering, is proposed. The theory of parsimonious covering is being extended to address a number of issues not currently handled, including causal chaining, hierarchial knowledge structures, and certainty measures. The use of this theory in diagnostic expert systems and other artificial intelligence systems is being evaluated by measuring performance on real-world problems and by assessing the ability of the systems to model human reasoning. The significance of this research is in its potential to improve the performance of intelligent computer-based systems. _____ IST-8644983 $123,221 - 12 mos. Whitman A. Richards Massachusetts Institute of Technology Natural Computation: A Computational Approach to Visual Information Processing - - - Natural computation is the study of how man can derive reliable representations of useful aspects of his environment. These representations may be visual, auditory, or consist of programs for touch sensing and motor control. At present, the data under study are the vast arrays of image intensities cast upon the eye or camera. Given these data, the task is to derive rather simple but reliable descriptions of objects and events in a three-dimensional world. The problems involve choice of representation, constraint analysis and explicit implementation (algorithm) to show that reliable and efficient descriptions can indeed be computed. The research may be divided roughly into four general problem areas: (1) 3D surface reconstruction; (2) property-based representations, (3) integrating depth maps obtained from stereo, structure-from-notion, and shading; and (4) computing spatial relations between "parts" and their combination into "object" representations. The significance of this research is that it will help provide a basic understanding of the issues involved in the very difficult problem of computer vision. _____ IST-8604530 $49,953 - 12 mos. Fred S. Roberts Rutgers University Scales of Measurement and the Limitations they Place on Information Processing - - - Information processing relies fundamentally on the process of quantification: Events in the real world must be translated into numbers and other symbols. These symbols (information) can then be transformed and otherwised processed by computers and other information processing systems. This research is a mathematical investigation of how that critical first step, quantification, limits and influences the way in which information can be used. In particular, the ways in which different quantifications can lead to different conclusions in decision making are analyzed. This research is significant as a foundation for information science. Too often, existing quantifications are taken for granted in information science or decision analysis, without any theoretical insight as to the range of significant alternatives. _____ IST-8603407 $27,000 - 12 mos. Robert D. Rodman North Carolina State University Dialog Processing for Voice Interactive Problem Solving (see description of award to Biermann) _____ IST-8640925 $198,800 - 12 mos. Naomi Sager New York University Language As a Database Structure - - - The proposal is based on past work of the Principal Investigator and associates, which include: the development of a computer grammar and program for analyzing free-text documents and converting their contents into a structured form; methods of identifying the fact-structures and text- wide regularities in specific subject domains ("sublanguages"); and experimental results in mapping the computer-structured textual information into databases of the network and relational types. The database experiments showed that certain necessary features of language information did not fit the existing models. This result, along with the accumulated knowledge of the major mechanisms of language information, have led to this project to develop a new database model in which the informational relations of language have a direct correspondence in constructs of the model. Using the techniques of database design, generalized Linguistic Schemata (LS's) will be developed for sample computer-structured language databases; the LS's will be specified in a Linguistic Schemata Decription Language (LSDL) to be designed for this purpose. Implementation of some of the schema elements will be undertaken in the form of enhancements to a relational data model, in particular, for treating the recursive connective structure of language information and the time-order relations in narrative. ------------------------------ END OF IRList Digest ********************