Browsing by Author "Nutter, J. Terry"

Now showing 1 - 9 of 9
  • Thumbnail Image
    AI, Science, and Intellectual Processes: Preliminary Remarks andArguments
    Nutter, J. Terry (Department of Computer Science, Virginia Polytechnic Institute & State University, 1990)
    This paper argues that trying to answer questions like the precise relationship of AI to existing disciplines (psychology, philosophy, linguistics, etc.) is both premature and potentially harmful to all concerned. This is not to say that we cannot say anything useful on the underlying questions which plague those who question its status; but the answers that can currently be given will probably fail to satisfy critics and proponents alike. This dissatisfaction--and indeed much of the debate--results from a view of science which takes as its model mature, developed sciences, and ignores facts about necessary phases in their development. This paper also argues that the question whether AI is a science is usually standing surrogate for concerns that have nothing whatever to do with science, and which should be addressed on their own grounds. The fundamental thesis here is that understanding the current status of AI, and so understanding the relationship between the various proposed approaches, requires adopting a more sophisticated approach to the status and development of intellectual disciplines, and that such an approach can contribute substantially to a broad area of current disputes, including most notably the "traditionalist/connectionist" controversy. This paper is divided into four questions: (1) "What is AI?"; (2) "Is AI science?"; (3) "What does all this say about AI, cognitive science and art forms?" The discussion of the fourth question will deal with consequences concerning the relationship between "traditional" and connectionist AI. I outline my positions on these four questions.
  • Thumbnail Image
    Building a Lexicon from Machine-Readable Dictionaries for Improved Information Retrieval
    Nutter, J. Terry; Fox, Edward A.; Evens, Martha W. (Department of Computer Science, Virginia Polytechnic Institute & State University, 1990)
    Information retrieval systems have a tremendous potential for contributing to research in virtually all areas. To date, this potential has not been fully realized, largely because of problems with controlling retrieval. One way to viewing these problems is that retrieval systems use keywords as indices to retrieve texts, as opposed to understanding the words in requests. We describe a project for creating a lexicon from machine-readable dictionaries, which information retrieval systems can use to go beyond present indexing methods, bringing the actual performance of such systems closer to their potential.
  • Thumbnail Image
    Implications of Natural Categories for Natural Language Generation
    Cline, Ben E.; Nutter, J. Terry (Department of Computer Science, Virginia Polytechnic Institute & State University, 1989)
    Psychological research has shown that natural taxonomies contain a distinguished or basic level. Adult speakers use the names of these categories most frequently and can list a large number of attributes for them. They typically can list many attributes for superordinate categories and list few additional attributes for subordinate categories. Because natural taxonomies are important to human language, their use in natural language processing systems appears well founded. In the past, however, most AI systems have been implemented around uniform taxonomies in which there is no distinguished level. It has recently been demonstrated that natural taxonomies enhance natural language processing systems by allowing selection of appropriate category names and by providing the means to handle implicit focus. We propose that additional benefits from the use of natural categories can be realized in multi-sentential connected text generation systems. After discussing the psychological research on natural taxonomies that relates to natural language processing systems, the use of natural categorizations in current natural language processing systems is presented. We then describe how natural categories can be used in multiple sentence generation systems to allow the selection of appropriate category names, to provide the mechanism to help determine salience to aid in the selection of discourse schema. to provide for the shallow modeling audience expertise, and to increase the efficiency of taxonomy inheritance.
  • Thumbnail Image
    Knowledge Representation Issues in Default Reasoning
    Nutter, J. Terry (Department of Computer Science, Virginia Polytechnic Institute & State University, 1989)
    Most existing approaches to reasoning in uncertainty and with incomplete information appeal to formal theories, with relatively little attention to the phenomena they are intended to capture. This has had two major consequences. First, it has led to the spurious disputes, in which participants criticize alternative approaches in the belief that they are competing, when in fact they are investigating different aspects of related phenomena, and should ultimately be viewed as cooperative efforts. Second, it has led to wasted efforts of models which fail to reflect important aspects of kinds of reasoning which they are trying to capture, because the representational requirements have not been adequately spelled out. This paper delineates several different kinds of reasoning in uncertainty, establishes some directions within the field, and attempts to begin setting some ground rules for representational adequacy.
  • Thumbnail Image
    A Lexical Relation Hierarchy
    Nutter, J. Terry (Department of Computer Science, Virginia Polytechnic Institute & State University, 1989)
    An extensive literature now exists documenting various lexical relations for representing information about words. This report summarizes the lexical relations recognized in a variety of sources. In addition, we claim that lexical relations themselves form not a class but a taxonomy, with a rich hierarchical structure. We present the outlines of this taxonomy, organize relations identified in a number of works under the taxonomy, and then give a condensed report of over 100 relations derived from the compendium, organized by their hierarchical status.
  • Thumbnail Image
    Natural Categories for More Natural Generation
    Cline, Ben E.; Nutter, J. Terry (Department of Computer Science, Virginia Polytechnic Institute & State University, 1990)
    Psychological research has shown that natural taxonomies contain a distinguished or basic level. Adult speakers use the names of these categories most frequently and can list a large number of attributes for them. They typically cannot list many attributes for superordinate categories and few list additional attributes for subordinate categories. Because natural taxonomies are important to human language, their use in natural language processing systems appears well founded. In the past, however, most AI systems have been implemented around uniform taxonomies in which there is no distinguished level. It has recently been demonstrated that natural taxonomies enhance language processing systems by allowing selection of appropriate category names and by providing the means to handle implicit focus. In previous research, we have argued that benefits from the use of natural categories can be realized in multi-sentential connected generation systems. We briefly summarize the psychological research on natural taxonomies that relates to natural language processing systems, the use of natural categorizations in current natural language processing systems, and the results of our previous research in which we show how natural categories can be used in multiple sentence generation systems to allow the selection of appropriate category names, to provide a mechanism to help determine salience, and to provide for the shallow modeling of audience expertise. We then describe additional benefits of natural categories in generation systems by demonstrating that natural categories provide a mechanism that aids selection of discourse schemes and increase the efficiency of inheritance.
  • Thumbnail Image
    Representing Knowledge About Words
    Nutter, J. Terry (Department of Computer Science, Virginia Polytechnic Institute & State University, 1989)
    Most on-line lexicons contain only semantic information. Semantic information is usually stored elsewhere, in a form consistent with representation of the syntactic information. This paper reports on research toward developing a large on-line lexicon from machine-readable dictionaries, which contains both syntactic and semantic information in uniform style. The fundamental theory is that of one of the relational lexicon; we describe relational lexicons, discuss our extensions to the usual theory of relational lexicons, rehearse very quickly some of the relations we are dealing with, and show how information for some simple entries is stored.
  • Thumbnail Image
    Sign for a Fully Transportable Natural Language Front-end to Database Management Systems
    Nutter, J. Terry; Safigan, Steve J.; Diaz, Angel M. (Department of Computer Science, Virginia Polytechnic Institute & State University, 1989)
    Natural language front-ends to database management systems represent a major improvement in accessibility for non-expert users. Unfortunately, such interfaces usually require extensive customizing not only of the front-end, but also of the data manager and hence of the DBMS itself. Developing such customized systems represents a huge investment of time and resources. In the 1980s, research has centered on making such systems portable at least across data bases, and in some cases across data base management systems. This report describes an architecture for complete transportability with minimal reprogramming, which has been partially implemented in a prototype system called TIPS. The TIPS architecture is compared briefly with other recent architectures, with attention to ease of portability, amount and locus of reprogramming needed, and extent of coverage both linguistically and in terms of database operations.
  • Thumbnail Image
    Uncertainty and Probability
    Nutter, J. Terry (Department of Computer Science, Virginia Polytechnic Institute & State University, 1986)
    Advocates of probability theory as a primary tool for reasoning in contexts of uncertainty and incomplete information have increased in number in recent years. At the same time, opponents have put forward a variety of arguments against using probabilities in this field. This paper examines the relationship between probability theory and reasoning in uncertainty, and argues that (contra opposing views) probability theory does have a place, but that its place is more restricted than many of its advocates claim. In particular, two major theses are presented and argued for. (1) Reasoning from probabilities works well in domains which permit a clear analysis in terms of events over outcome spaces and for which either relatively large bodies of evidence or long periods of "kaining" are available; but such domains are relatively rare, and even there, care must be taken in interpreting probability results. And (2) some generalizations with which AI applications must concern themselves am not statistical in nature, in the sense that statistical generalizations neither capture their meanings nor even preserve their truth values. For these contexts, different models will be needed.