Theoretical Issues In Natural Language Processing最新文献

{"title":"Parallel Distributed Processing and Role Assignment Constraints","authors":"James L. McClelland","doi":"10.3115/980304.980320","DOIUrl":"https://doi.org/10.3115/980304.980320","url":null,"abstract":"I believe that parallel-distributed processing models (i.e., conneelionist models which make use of distributed representations) provide the mechanisms that are needed for these lasks. Argument altachments and role assignments seem to require a consideration of the relative merits of competing possibilities (Marcus, 1980; Bates and MacWhinney, 1987; MaeWhinney, 1987), as deles lexical dlsambigualion. Conuectionist models provide a very natural substrate for these kinds of competition processes (Cottrell, 1985; Wallz and Pollack, 1985).","PeriodicalId":439225,"journal":{"name":"Theoretical Issues In Natural Language Processing","volume":"209 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115563017","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Reference and Reference Failures","authors":"Bradley A. Goodman","doi":"10.3115/980304.980343","DOIUrl":"https://doi.org/10.3115/980304.980343","url":null,"abstract":"Introduction Reference in the real world differs greatly from the reference processes modelled in current natural ~ language systems. A speaker in the real world is a rational agent who must make a decision about his description in a limited time, with limited resources, knowledge, and abilities. In particular, the speaker's perceptual and communicative skills are imperfect or his model of the listener is erroneous or incomplete. Additionally, a speaker can also be sloppy in his description. Since the speaker's goal in the reference process is to construct a description that \"works\" for the listener, the listener, from his viewpoint, must take these imperfections into account when trying to interpret the speaker's utterances. Yet, listeners, too, have imperfect perceptual or communicative skills and can be sloppy. Hence, they must be prepared to deal with their own imperfections when performing reference identification. In real reference, listener's often recover from initial misunderstandings with or without help from the speaker. Natural language understanding systems must do this, too. Therefore, in performing the reference process, a system should assume and expect problems.","PeriodicalId":439225,"journal":{"name":"Theoretical Issues In Natural Language Processing","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116315909","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Words and Worlds","authors":"R. A. Amsler","doi":"10.3115/980304.980307","DOIUrl":"https://doi.org/10.3115/980304.980307","url":null,"abstract":"WORDS AND WORLDS Robert A. Amsler Artificial Intelligence and Information Science Research Bell Communications Research Morristown, New Jersey 07960 For several years now I have been concerned with how artificial intelligence is going to build the substitute for human world knowledge needed in performing the task of text understanding. I continue to believe that the bulk of this knowledge will have to be derived from existing machine-readable texts produced as a byproduct of computer typesetting and word processing technologies which have overtaken the publishing industries. However, there are many obstacles to the acquisition of world knowledge from text. There are some, I am sure, who would argue that world knowledge of the form needed in text understanding will have to be hand-coded and cannot be derived from existing reference books or other texts. My basic argument against those who hold this view is that they are ignor- ing the magnitude of the task ahead. Whether measured in terms of bytes or man-years, the sum of recorded knowledge is so massive that it is unlikely to be capable of being recoded in anything less than man-centuries. Put another way, there currently exist sizeable publishing empires in this country which every day employ hundreds of people involved directly in the coding of information for new refer- ence texts and revised editions of older reference works. To attempt a recoding of world knowledge solely for use in AI would eventually become an attempt to parallel this effort. It would become a major industry in itself. Thus, it is more likely that, instead of a new knowledge-base industry, we will see an evolutionary change in the methods used by the existing publishing empires to record knowledge in a manner that is of use in producing text both for hu- man consumption and as knowledge bases for computers. Researchers in AI and computational linguistics therefore have some responsibility to determine how the existing printed knowledge can evolve into usable computational world knowledge Now, of course, I do admit there are subclasses of world knowledge that evidence to date has not shown to exist in print at all. Jerry Hobbs is attempting to codify one such subclass in his work on TACITUS (Hobbs et al. 1986). There are others as well, such as some forms of linguistic knowledge. However, I am concerned about the very large body of knowledge that we try to communicate to people through books, newspapers and other texts. This knowledge of the out- side world, of experiences in which the individual has not and in fact may never personally be in- volved, is nevertheless shared knowedge known to all of us through reading and listening to the words of others. Another assumption, and one that has been guiding my work for many years now, is that natural language systems cannot understand text for which they do not possess the lexicon. This seems so elemental an assumption that I find it hard to see how to ignore the fact that we d","PeriodicalId":439225,"journal":{"name":"Theoretical Issues In Natural Language Processing","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126114969","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Situated Perspective on Natural-Language Processing","authors":"S. Stucky","doi":"10.3115/980304.980337","DOIUrl":"https://doi.org/10.3115/980304.980337","url":null,"abstract":"Use the word 'situated' these days and all anyone can think of is situation semantics, t That 's unfortunate from my point of view, because [ would like to explore an idea tha t follows from ideas handily referred to as \"a si tuated perspective.\" (Not surprisingly, both situation semantics and its companion logic situation theory are by and large consistent with this point of view, but they aren't the only way of working out a theory compatible with it.) Viewing computation, language, and inference through this perspective, [ will maintain, suggests a conception of naturallanguage processing both more complicated and more realistic than that underlying much of current practice. No system yet exists under this conception, though there is one under design Isee note 4): therefore all [ can reasonably hope to do is engage your interest and convince you that the overall conception merits consideration. This \"situated perspective\" [ have in mind is best thought of as a cluster of mutual ly reinforcing assumptions. As implied, these assumptions don't constitute a theory in and of themselves; they're merely indicative of one. or several Furthermore. a curious fact is that many of the underlying assumptmns are unobjectionable. They're the sort of thing about which you find yourself ~.hinking well, who could argue with that. Take the claim that language is efficient--Barwise and Perry's way of referring to the fact that you can use the same expression over and ,)vet\" again to talk about different things. That 's as familiar a notion as the productivity of language (Barwise and Perry, 1983}. Thus, to claim that language is efficient is to claim that interpretation depends on context: to know what the interpretation of the word 'you' is (i.e., who is being referred to) on a particular occasion of use depends on who you are talking to. ([ will be using the word ' interpretation' to mean the actual stuff referred to, the properties it is said to have etc.} Yet, if you pair the claim about the context-dependency of interpretation of language with the equally reasonable claim that computational processes similarly have context-dependent interpretations, you are all set for the more complicated perspective on naturallanguage processing [ mentioned. 2 At least, this is what [ am going to argue. Another central assumption of the \"situated perspective\" is the idea that meaning is relational. A kingpin of situation semantics, this assumption follows quite","PeriodicalId":439225,"journal":{"name":"Theoretical Issues In Natural Language Processing","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122318979","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"What is Special About Natural Language Generation Research?","authors":"W. Mann","doi":"10.3115/980304.980355","DOIUrl":"https://doi.org/10.3115/980304.980355","url":null,"abstract":"Abstract : Since the guidance given to the panel was more provocative than regulative, I have organized my statement around just one of the questions: What is special about Text Generation relative to NL Understanding? This breaks down conveniently into parts: 1. Are there foundational ideas that generation and understanding work share? 2. What are the technical distinctives of generation? 3. Are those distinctives real? 4. What are the special characteristics of generation as a research task?","PeriodicalId":439225,"journal":{"name":"Theoretical Issues In Natural Language Processing","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133951723","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Position Paper on Common-sense and Formal Semantics","authors":"G. Nunberg","doi":"10.3115/980304.980334","DOIUrl":"https://doi.org/10.3115/980304.980334","url":null,"abstract":"I'm not sure what I'm doing on this panel, but I thought it would be helpful if we could start at the beginning. It's interesting to note that both the dictionary and common sense were eighteenth-century inventions. This is no coincidence; in fact, it's entirely appropriate that the most celebrated blow that was ever struck on behalf of common sense was delivered by Dr. Johnson himself, when he kicked a stone in refutation of Berkeley's idealism. If you'll indulge me, I'd like briefly to explore this connection, with the promise that I'll wind up by drawing a moral about n_atural-language understanding. First for \"common sense.\" To be sure, the phrase had been in use since the fourteenth century, but it was not until the eighteenth century that it acquired three important philosophical uses, which we confuse at our peril. First, there was the understanding of common sense as the faculty of judgement possessed by the average person as part of his birthright. This is the sense of the phrase that Priestly had in mind when he wrote (1775): \"Common sense...in common acceptation...has long been appropriated...to that capacity for judging of common things that persons of middling capacities are capable of.\" In this sense, common sense is variously a general or universal faculty, \"the same in every time and clime,\" as Hume (somewhere) put it; to avoid confusion here, I'll sometimes refer to this as the faculty of common sense. Now for many philosophers of the period, it was an article of doctrine that natural and moral law were accessible to common sense, and hence that the body of common man could achieve both the degree of understanding of the world and the refinement of moral judgment necessary to self-government. This was the view of Hume, of Johnson, and of course of Thomas Paine, who entitled his famous treatise on natural rights Common Sense. The second use of the phase was more narrowly associated with the Scottish School of phi losophy(also called the \"Philosophers of Common Sense\"); on this understanding it is a universal endowment, not just of critical faculties, but of beliefs. As the philosopher Hamilton put it, common sense is in this interpretation is \"the complement of those cognitions or convictions which we receive from nature; which all men possess in common .... \" This group made common sense the test of all philosophical doctrines, and in particular referred to the common-sense belief in the reality of t he physical world in the course of their Berkeley.bashing (a","PeriodicalId":439225,"journal":{"name":"Theoretical Issues In Natural Language Processing","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121183137","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"World Knowledge and Word Meaning","authors":"Jerry R. Hobbs","doi":"10.3115/980304.980308","DOIUrl":"https://doi.org/10.3115/980304.980308","url":null,"abstract":"We use words to talk about the world. Therefore, to understand what words mean, we must have a prior explication of how we view the world. In a sense, efforts in the past to decompose words into semantic primitives were attempts to link word meaning to a theory of the world, where the set of semantic primitives constituted the theory of the world. With the advent of naive physics and research programs to formalize commonsense knowledge in a number of areas in predicate calculus or some other formal language, we now have at our disposal means for building much richer theories of varlous aspects of the world, and consequently, we are in a much better position to address the problems of le~cal semantics. In the TACITUS project for using commonsense.knowledge in the understanding of texts about mechanical devices and their failures, we have been developing various commonsense theories that are needed to mediate between the way we talk about the behavior of such devices and causal models of their operation (Hobbs et M., 1986). The theories cover a number of areas that figure in virtually every domain of discourse, such as scalar notions, granularity, structured systems, time, space, material, physical objects, causality, functionality, force, and shape. Our approach has been to construct core t h e o r i e s of each of these areas. These core theories may use English words as their predicates, but the principal criterion for adequacy of the core theory is elegance, whatever that is, and this can usually be achieved better using predicates that are not lexically realized. It is easier to achieve elegance if one does not have to be held responsible to linguistic evidence. Predicates that are lexically realized are then pushed to the periphery of the theory. A large number of lexical items can be defined, or at least characterized, in terms provided by the core theories. The hypothesis is that once these core theories have been formulated in the right way, it will be straightforward to explicate the meanings of a great many words. The phrase \"in the right way\" is key in this strategy. The world is","PeriodicalId":439225,"journal":{"name":"Theoretical Issues In Natural Language Processing","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128590829","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"On Formal Versus Commonsense Semantics","authors":"David J. Israel","doi":"10.3115/980304.980335","DOIUrl":"https://doi.org/10.3115/980304.980335","url":null,"abstract":"There is semantics and, on the other hand, there is seraan~ics. And then there is the theory of meaning or content. I shall speak of pure mathematical semantics and real semantics. I have very little idea what \"formal\" means in \"formal semant ics\"--unless it simply means semantics done rigorously and systematically. 1 I have even less idea what is meant by \"commonsense semantics\". I shall not speak much of the theory of meaning. The distinction between these two modes of semantics, the mathematical and the real, is not meant to be a hard and fast dist inct ion--nor, most assuredly, is it intended to be rigorous or systematic. As I see it, the distinction has primarily to do with purposes or goals, derivatively, with constraints on the tools or conceptual resources considered available to realize those goals. In particular, real semantics is simply pure mathemat ical semantics with certain goals in mind and thus operating under certain additional constraints. Ano the r way to put the point: some work in pure mathematical semantics is in fact a contribution to real semantics; however, it does not have to be such to make a genuine contribution to pure mathematical semantics. 2 Hence, since real semantics can be executed with the same degree of rigor and systematicity as must all of pure mathemat ical semantics, it should be. Have I made myself clear? Not entirely, perhaps. Let 's t ry a more systematic approach. Pure mathematical semantics is either a part of or an application of mathemat ical logic. Real semantics, even though an application of mathematical logic, is a part of the theory of meaning or content. Contributions to real semantics had better cast some light on natural ly occurring phenomena within the purview of a theory of meaning-on such properties and relations as t ruth analyticity, necessity, implication. Traditionally (indeed, until Montague, almost undeviatingly) the techniques of pure mathematical semantics were deployed for formal or artificial languages. But this by itself is of no importance. These languages were invented, and are of interest only because, or insofar as, they are plausible and il luminating models, in the intuitive sense, of real phenomena in thought and in informal language. Consequently, the nature of the languages studied need not make an essential difference. 3 What does make a difference is the purpose","PeriodicalId":439225,"journal":{"name":"Theoretical Issues In Natural Language Processing","volume":"143 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129251678","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Toward Connectionist Semantics","authors":"G. Cottrell","doi":"10.3115/980304.980318","DOIUrl":"https://doi.org/10.3115/980304.980318","url":null,"abstract":"Much of the study of language has centered around the study of syntax, to the detriment of semantics and pragmatics. Part of the reason for this may be akin to the motivation of the besotted gentleman on his hands and knees beneath a streetlamp, who, when queried as to why he is looking on the sidewalk for the keys he lost in the alley, replies: \"Because the light is better here!\" I believe it is time to start mucking about in the alley; the keys are there. I also think we have a new flashlight: Parallel Distributed Processing 2. PDP mechanisms allow us to build machines whose fundamental operations include best fit search, constraint relaxation and automatic generalization. These are useful properties for processing language. I think the application of these models to NLP will change our view of what constitutes \"semantics'. I will argue that in order to deal with meaning seriously, we have to move beyond the folkpsychological level of symbols, and represent the microstructure of symbols. This is more than a granularity issue. It also has to do with the grounding of meaning in perception. It is on the level of microfeatures that I believe this grounding occurs, and PDP gives us a way to express this interface between language and perception. My discussion of these issues will take the following course 3. First I describe my previous work on word sense disambiguation in a PDP framework as a springboard for the rest of the discussion, and to give a sense of how lexical semantics might fit into an overall parsing model. Next I motivate a new model of word meanings through an example. I try to show that PDP has a natural way of expressing these meanings, and I give a sketch of how connectionist semantics could be learned. Finally, I briefly discuss metaphor.","PeriodicalId":439225,"journal":{"name":"Theoretical Issues In Natural Language Processing","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133436440","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Whither Discourse and Speech Acts?","authors":"B. Grosz","doi":"10.3115/980304.980323","DOIUrl":"https://doi.org/10.3115/980304.980323","url":null,"abstract":"The panelists papers and a glance through the proceedings of the previous two T I N L A P meet ings suggests that we have in fact learned a good deal and that research in discourse and speech act theory has become much more focused in the last five or so years. Several theories of discourse have been proposed to account for the structure and processing of ex tended sequences of ut terances. In a recent paper [Grosz&Sidner86], Candy Sidner and I present one such theory and discuss a range of alternative ones. Other research has provided partial accounts for the interpretat ion and generat ion in discourse of individual expressions, clauses, and interclausal connections (e.g., [McKeown85], [Appelt85], [Groszeta183], [Mann&Thompson83] [Sidner83], [Webber83] , [Hobbs79]) . These theories differ in the predictions they • make about discourse structure and coherence, and about the processing of various types of linguistic expressions (e.g., cue phrases, definite descriptions, and p ronouns ) . Initial psychological investigations into various aspects of these theories are being carried out.","PeriodicalId":439225,"journal":{"name":"Theoretical Issues In Natural Language Processing","volume":"242 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115223275","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}