leff@smu.CSNET (Laurence Leff) (05/23/87)
Below are the abstracts of forthcoming publications from USC/Information Sciences Institute. If you wish to receive copies of any of these reports, please contact: Diane Speekman USC/Information Sciences Institute 4676 Admiralty Way, Ste. 1001 Marina del Rey, CA 90292-6695 Copies of all USC/ISI reports may be obtained by writing (about three months after publication) to National Technical Information Service, Springfield, VA 22151. ---------------------------------------------------------------------------- ANTITHESIS: A STUDY IN CLAUSE COMBINING AND DISCOURSE STRUCTURE William C. Mann Sandra A. Thompson ISI/RS-87-171 April 1987 approx. 30 pages AI research in text generation needs a strong linguistically justified descriptive theory as a basis for creating methods by which programs can write multiparagraph texts. This paper sketches Rhetorical Structure Theory, which has been designed to support text generation, and then applies RST to describing a particular class of discourse constructs. There is no consensus as to the status of clause combining relations relative to larger texts. This paper demonstrates a clause combining relation that is also found as part of larger text structures, and shows how this fact can be used to explain cases in which contrastive clause combining appears between clauses that are not in fact in contrast. The appropriate generalization is that the relations of clause combining and the relations of general text structure are the same. Use of this generalization should make AI text planning and text generation significantly easier. ------------------------------------------------------------------------------- NOTES ON THE ORGANIZATION OF THE ENVIRONMENT OF A TEXT GENERATION GRAMMAR Christian Matthiessen ISI/RS-87-177 April 1987 approx. 52 pages One of the tasks in designing a text generation system is to organize the environment of the grammatical component of the generation system in such a way that it supports the grammatical resources in generation. This report discusses the methods used for the Penman generation system to infer aspects of the organization of the knowledge base and other components of the environments of the Nigel grammar of the Penman system. It is shown how the design task can be broken down into a number of very explicit demands on the environment. In the main part of the report, the results of application of such an approach is sketched, with particular emphasis on the general organization of the knowledge base and the discourse model parts of the environment. ------------------------------------------------------------------------------- SYSTEMIC GRAMMAR AND FUNCTIONAL UNIFICATION GRAMMAR - and - REPRESENTATIONAL ISSUES IN SYSTEMIC FUNCTIONAL GRAMMAR Christian Matthiessen Robert Kasper ISI/RS-87-179 April 1987 approx. 55 pages SYSTEMIC GRAMMAR AND FUNCTIONAL UNIFICATION GRAMMAR: Systemic Functional Grammar (SFG) and Functional Unification Grammar (FUG) are superficially very different approaches to grammatical knowledge, but they share an underlying comparability that runs very deep. FUG shares with systemic descriptions an emphasis on the functions of linguistic objects, and an explicit representation of feature choices. This paper explores how a systemic grammar can be represented in FUG notation, as a step toward creating a grammatical analysis program for English. Because FUG has been developed as a computational tool, expressing a systemic grammar in FUG notation allows new computational techniques to be applied to it. Among other benefits, this program will make it possible to study how much the grammatical functions of sentences are recoverable from them. It will also provide a method to test the amount of ambiquity implicit in a systemic description, a topic which has so far been inaccessible. This use of FUG as an alternate representation for SFG may have some additional benefits for both frameworks. It provides some solutions to problems in systemic notation which are described by Matthiessen (in this volume). Several extensions to the FUG framework are also suggested by this study. REPRESENTATIONAL ISSUES IN SYSTEMIC FUNCTIONAL GRAMMAR: Nigel is a large diverse computational grammar for text generation. Its framework is an implementation of Systemic Functional Theory of grammar and it constitutes a context in which the representation of systemic theory can be explored and studied. This paper surveys the representational devices used in the Nigel grammar and the representational issues that they raise in relation to systemic theory. These issues are diagnosed in the light of the metafunctional differentiation of systemic theory. ------------------------------------------------------------------------------- ROUTING AND ADDRESSING PROBLEMS IN LARGE METROPOLITAN-SCALE INTERNETWORKS Gregory G. Finn ISI/RR-87-180 March 1987 approx. 61 pages Digital packet networking technology is spreading rapidly into the commercial sector. Currently, most networks are isolated local area networks. This isolation is counterproductive. Within the next twenty years it should be possible to connect these networks to one another via a vast internetwork. A metropolitan internetwork must be capable of connecting many thousand networks and a national one several million. It is difficult to extend current internetworking technolgy to this scale. Problems include routing and host mobility. This report addresses these problems by developing an algorithm that retains robustness and has desirable commercial characteristics. ------------------------------------------------------------------------------- TEXT GENERATION: THE PROBLEM OF TEXT STRUCTURE William C. Mann ISI/RS-87-181 March 1987 approx. 22 pages One of the key problems in AI text generation is text organization. A poorly organized text can be unreadable or even misleading. Two AI approaches to text organization are compared: McKeown's TEXT system and Rhetorical Structure Theory (RST). They share many assumptions about the nature of text yet they are found to be in strong contrast. TEXT identifies text organization with whole-text nonrecursive structures, while RST uses small recursive ones. RST has an elaborate apparatus of relations between parts of texts, and of the "nuclearity" of particular parts; TEXT has no correlates for these. RST works with a wide range of relation types, TEXT with just one of them. TEXT has no correlates for these. RST works with a wide range of relation types, TEXT with just one of them. TEXT is an implemented system, whereas RST is developmental. Most important, TEXT develops text organizations so that they resemble patterns extracted from previous text, while RST strives for an organization which is justifiable as meeting the goals of the text being generated. This contrast raises many of the key issues discussed in this paper about the nature of text organization and how it can be created by programs. ------------------------------------------------------------------------------- A REVIEW OF CONCEPTUAL STRUCTURES: INFORMATION PROCESSING IN MIND AND MACHINE Stephen Smoliar ISI/RS-87-182 March 1987 approx. 8 pages This is a review of the book Conceptual Structures: Information Processing in Mind and Machine, by John F. Sowa, published in 1984 by the Addison-Wesley Publishing Company. ------------------------------------------------------------------------------- THE STRUCTURE OF DISCOURSE AND "SUBORDINATION" Christian Matthiessen Sandra A. Thompson ISI/RS-87-183 April 1987 The use and nature of clause combining in natural discourse are explored in this paper. First, a theory of text structure, Rhetorical Structure Theory, is introduced and illustrated for a number of short texts. Then, it is shown how the grammar of clause combining can be explained in terms of the structuring of text. The paper focuses on one particular way of combining clauses and shows how it is used to express a nuclear-satellite structuring of text identified by Rhetorical Structure Theory. -------------------------------------------------------------------------- RELIABILITY AND PERFORMANCE MODELLING OF HYPERCUBE-BASED MULTIPROCESSORS Walid Najjar Jean-Luc Gaudiot ISI/RS-87-184 April 1987 New technologies of integration now enable the design of computing systems with dozens and even hundreds of independent homogeneous Processing Elements which can cooperate on the solution of the same problem for a corresponding improvement in the execution time. However, as the number of Processing Units increases, concerns for reliability and continued operation of the system in the face of failures must be addressed. A commonly used network topology, the hypercube, is analyzed in this paper. It is shown how the disconnection of a subset of the machine would yield a total system failure. Overall reliability measures as well as computational availability, etc., are derived by a combination of analytical modelling and simulation approaches (Monte Carlo simulation). Finally, a comparison with other topologies (mesh connected network, etc.) shows the influence of the connectivity of the network on the reliability of the system. ------------------------------------------------------------------------------- RHETORICAL STRUCTURE THEORY: A FRAMEWORK FOR THE ANALYSIS OF TEXTS William C. Mann Sandra A. Thompson ISI/RS-87-185 April 1987 Rhetorical Structure Theory is a theory of text organization which provides a framework for an analysis of text. The theory is based on the understanding that a text is not merely a string of clauses, but consists instead of hierarchically organized groups of clauses that stand in various relations to one another. These "rhetorical relations" can be described functionally in terms of the purposes of the writer and the writer's assumptions about the reader. They hold between two adjacent parts of a text, where, typically, one part is "nuclear" and one a "satellite." An analysis of a text consists in identifying the relations holding between successively larger parts of the text, yielding a natural hierarchical descriptions of the rhetorical organization of the text. The paper informally outlines RST's mechanisms and applications, which include studies of clause combining, coherence and assertional effects of discourse structure. RST characteristically provides comprehensive analyses rather than selective commentary. RST is insensitive to text size, and has been applied to a wide variety of sizes of text. ------------------------------------------------------------------------------- INTERPRETATION IN GENERATION Eduard H. Hovy ISI/RS-87-186 April 1987 The computer maxim "garbage in, garbage out" is especially true of language generation. When a generator slavishly follows its input topics, it usually produces bad text. In order to find more appropriate forms of expression, generators must be given the ability to interpret their input topics. Often, newly-formed interpretations can help generators achieve their pragmatic goals with respect to the hearer. Since interpretation requires inference, generators must exercise some control over the inference process. Some general strategies of control and some specific techniques, geared toward achieving pragmatic goals, are described here. ------------------------------------------------------------------------------- A UNIFICATION METHOD FOR DISJUNCTIVE FEATURE DESCRIPTIONS Robert Kasper ISI/RS-87-187 April 1987 Although disjunction has been used in several unification-based grammar formalisms, existing methods of unification have been unsatisfactory for descriptions containing large quantities of disjunction, because they require exponential time. This paper describes a method of unification by successive approximation, resulting in better average performance. ------------------------------------------------------------------------------- SEMANTICS FOR A SYSTEMIC GRAMMAR: THE CHOOSER AND INQUIRY FRAMEWORK Christian Matthiessen ISI/RS-87-189 May 1987 This report describes the semantic interface between a systemic functional grammar for text generation and the environment the grammar operates in. The grammar is organized as a network of choice points and the semantic interface provides a method for making the grammatical choices in a purposeful way. Each grammatical choice point is equipped with its own semantic procedure for choosing: one or more questions are addressed to one of the components of the environment, such as the knowledge base, so that the information needed to select the appropriate choice alternative can be obtained. The paper presents the framework as a kind of semantics for systemic grammars and also relates it to other semantic approaches.