DRAFTSpeech and Language Processing: An Introduction to Natural Language Processing,Computational Linguistics and Speech Recognition: Second Edition, Daniel Jurafsky & JamesH. Martin. Copyrightc! 2006, All rights reserved. Draft of August 28, 2006. Do notcite.16REPRESENTINGMEANINGISHMAEL: Surely all this is not without meaning.Herman Melville, Moby DickThe approach to semantics introduced here, and elaborated on in the next fourchapters, is based on the notion that the meaning of linguistic utterances can becaptured in formal structures, which we will call meaning representations. Cor-MEANINGREPRESENTATIONSrespondingly, the frameworks that are used to specify the syntax and semantics ofthese representations will be called meaning representation languages. TheseMEANINGREPRESENTATIONLANGUAGESmeaning representations play a role analogous to that of the phonological, mor-phological, and syntactic representations introduced in earlier chapters.The need for meaning representations arises when neither the raw linguisticinputs, nor any of the structures derivable from them by any of the transducerswe have studied thus far, facilitate the kind of semantic processing that is desired.More specifically, what we need are representations that bridge the gap from lin-guistic inputs to the non-linguistic knowledge of the world needed to perform tasksinvolving the meaning of linguistic inputs. To illustrate this notion, consider thefollowing everyday language tasks that require some form of semantic processingof natural language:• Answering an essay question on an exam;• Deciding what to order at a restaurant by reading a menu;• Learning to use a new piece of software by reading the manual;• Realizing that you’ve been insulted; and• Following a recipe.Simply having access to the phonological, morphological, and syntactic represen-tations that we have discussed thus far will not get us very far on accomplishingDRAFT2 Chapter 16. Representing Meaningany of these tasks. These tasks require access to representations that link the lin-guistic elements involved in the task to the non-linguistic knowledge of the worldneeded to successfully accomplish them. For example, some of the world knowl-edge needed to perform the above tasks would include the following:• Answering and grading essay questions requires background knowledge aboutthe topic of the question, the desired knowledge level of the students, and howsuch questions are normally answered.• Reading a menu and deciding what to order, giving advice about where to goto dinner, following a recipe, and generating new recipes all require knowl-edge about food, its preparation, what people like to eat and what restaurantsare like.• Learning to use a piece of software by reading a manual, or giving adviceabout how to do the same, requires knowledge about current computers, thespecific software in question, similar software applications, and knowledgeabout users in general.In the representational approach explored here, we take linguistic inputs andconstruct meaning representations that are made up of the same kind of stuff that isused to represent this kind of everyday commonsense knowledge of the world. Theprocess whereby such representations are created and assigned to linguistic inputsis called semantic analysis.SEMANTIC ANALYSISTo make this notion a bit more concrete, consider Fig. 16.1, which showssample meaning representations for the sentence I have a car using four repre-sentative meaning representation languages. The first row illustrates a sentence inFirst-Order Logic, which will be covered in detail in Section 16.4; the graph inthe center illustrates a Semantic Network, which will be discussed further in Sec-tion 16.6; the third row contains a Conceptual Dependency diagram, discussed inmore detail in Ch. 18, and finally a Frame-Based representation, also covered inSection 16.6.While there are non-trivial differences among these approaches, at an abstractlevel they all share as a common foundation the notion that a meaning represen-tation consists of structures composed from a set of symbols, or representationalvocabulary. When appropriately arranged, these symbol structures are taken to cor-respond to the objects, properties of objects and relations among objects in somestate of affairs being represented. In this case, all four representations make use ofsymbols corresponding to the speaker, a car, and relations denoting the possessionof one by the other.It is important to note that these representations can be viewed from at leasttwo distinct perspectives in all four of these approaches: as representations of themeaning of the particular linguistic input I have a car, and as representations ofDRAFT3∃x, y Having(x) ∧ Haver(Speaker, x) ∧ HadT hing(y, x) ∧Car(y)HavingHaver Had-ThingSpeaker CarCar Having⇑POSS-BY Haver: SpeakerSpeaker HadThing: CarFigure 16.1 A list of symbols, two directed graphs, and a record structure: a sam-pler of meaning representations for I have a car.the state of affairs in some world. It is this dual perspective that allows these rep-resentations to be used to link linguistic inputs to the world and to our knowledgeof it.The structure of this part of the book parallels that of the previous parts. Wewill alternate discussions of the nature of meaning representations with discus-sions of the computational processes that can produce them. More specifically,this chapter introduces the basics of what is needed in a meaning representation,while Ch. 17 introduces a number of techniques for assigning meanings to linguis-tic inputs. Ch. 18 explores a range of complex representational issues related tothe meanings of words. Ch. 19 then explores some robust computational methodsdesigned to exploit these lexical representations.Since the focus of this chapter is on some of the basic requirements for mean-ing representations, we will defer a number of extremely important issues to laterchapters. In particular, the focus of this chapter is on representing what is some-times called the literal meaning of sentences. By this, we have in mind represen-LITERAL MEANINGtations that are closely tied to the conventional meanings of the words that are usedto create them, and that do not reflect much of the context in which they occur. Theshortcomings of such representations with respect to phenomena such as idiomsand metaphor will be discussed in the next two chapters, while the role of contextin ascertaining the
View Full Document
Unlocking...