Send us a comment.
1.1 The origins of natural language processing 1.2 The imposition of structure 1.3 The representation of meaning 1.4 The role of knowledge 1.5 The emergence of a new technology 1.6 Using Prolog for natural language procesing
This is a textbook that aims to teach the reader about useful, well-understood techniques for natural language processing (NLP). In the chapters that follow, we will rarely have much to say about the history of the field, about approaches whose utility is doubtful or about techniques that are, as yet, poorly understood. Occasionally, the further reading sections that end each chapter will provide some pointers into the history, but that is not their primary function. Accordingly, this introductory chapter provides a thumbnail sketch of the history of NLP so that those new to the field can get some slight sense of what has gone on in the first three decades of work in computational linguistics.
We begin by looking at the origins of NLP on machines less powerful than the pocket calculators of today, at early applied computational linguistics and at early machine translation. We observe the move from numbers to strings and thence to structures, and at the parallel progress from procedural ways of thinking embodied in, for example, augmented transition networks, towards the declarative formalisms and data structures that have emerged in the 1980s. En route, we note how ambiguity has emerged as the problem for NLP and see the directions in which people have sought solutions to it.
Then we turn to the representation of meaning: classic procedural work from the 1970s, the emergence of network formalisms and the gradual resurgence of logicism in the 1980s after a decade in which it was deeply unfashionable. Ambiguity rears its head again and we glance at a semantic technique for reducing it by imposing selectional restrictions. Such semantic techniques, it turns out, are insufficient for satisfactory NLP and so we look at the role of knowledge, especially real-world knowledge, and examine its relevance to participants' goals and beliefs, and hence also their speech acts, conversation and the structure of their discourse.
This is a book about NLP techniques, not about their application. However, after surveying the short history of the field in this introductory chapter, we take an even briefer look at the emergence of the new technology of current NLP applications and tools, including machine translation, database front ends, grammar development environments, intelligent text processing and articulate expert systems.
In subsequent chapters, our discussion will be illustrated with programming examples and exercises that use Prolog. Although the program code used is intended to be straightforward, this book is not intended to teach the basics of programming, nor even the basics of Prolog. It is assumed that readers already have some facility in Prolog, or are being taught it in parallel with an NLP course based on this book. A bit more is said about the particular characteristics of Prolog and its suitability for NLP in the penultimate section of this chapter.
This chapter concludes, in a final further reading section, by suggesting sources, some of which provide more information on the history of NLP than has been given here, and others of which survey current areas of application in detail. There are a number of good textbooks on Prolog programming and these are also listed in this section.
Send us a comment.