Издательство Marcel Dekker, 2000, -962 pp.The discipline of Natural Language Processing (NLP) concerns itself with the design and implementation of computational machinery that communicates with humans using natural language. Why is it important to pursue such an endeavor? Given the self-evident observation that humans communicate most easily and effectively with one another using natural language, it follows that, in principle, it is the easiest and most effective way for humans and machines to interact; as technology proliferates around us, that interaction will be increasingly important. At its most ambitious, NLP research aims to design the language input- output components of artificially intelligent systems that are capable of using language as fluently and flexibly as humans do. The robots of science fiction are archetypical: stronger and more intelligent than their creators and having access to vastly greater knowledge, but human in their mastery of language. Even the most ardent exponent of artificial intelligence research would have to admit that the likes of HAL in Kubrick's 2001: A Space Odyssey remain firmly in the realms of science fiction. Some success, however, has been achieved in less ambitious domains, where the research problems are more precisely definable and therefore more tractable. Machine translation is such a domain, and one that finds ready application in the internationalism of contemporary economic, political, and cultural life. Other successful areas of application are message-understanding systems, which extract useful elements of the propositional content of textual data sources such as newswire reports or banking telexes, and front ends for information systems such as databases, in which queries can be framed and replies given in natural language. This handbook is about the design of these and other sorts of NLP systems. Throughout, the emphasis is on practical tools and techniques for implementable systems; speculative research is minimized, and polemic excluded.Part I Symbolic Approaches to NLP.
Symbolic Approaches to Natural Language Processing.
Tokenisation and Sentence Segmentation.
Lexical Analysis.
Parsing Techniques.
Semantic Analysis.
Discourse Structure and Intention Recognition.
Natural Language Generation.
Intelligent Writing Assistance.
Database Interfaces.
Information Extraction.
The Generation of Reports from Databases.
The Generation of Multimedia Presentations.
Machine Translation.
Dialogue Systems: From Theory to Practice in TRAINS-96.
Part II Empirical Approaches to NLP.
Empirical Approaches to Natural Language Processing.
Corpus Creation for Data-Intensive Linguistics.
Part-of-Speech Tagging.
Alignment.
Contextual Word Similarity.
Computing Similarity.
Collocations.
Statistical Parsing.
Authorship Identification and Computational Stylometry.
Lexical Knowledge Acquisition.
Example-Based Machine Translation.
Word-Sense Disambiguation.
Part III Artificial Neural Network Approaches to NLP.
NLP Based on Artificial Neural Networks: Introduction.
Knowledge Representation.
Grammar Inference, Automata Induction, and Language Acquisition.
The Symbolic Approach to ANN-Based Natural Language Processing.
The Subsymbolic Approach to ANN-Based Natural Language Processing.
The Hybrid Approach to ANN-Based Natural Language Processing.
Character Recognition with Syntactic Neural Networks.
Compressing Texts with Neural Nets.
Neural Architectures for Information Retrieval and Database Query.
Text Data Mining.

Text and Discourse Understanding: The DISCERN System.

Издательство Marcel Dekker, 2000, -962 pp.自然语言处理（NLP）学科关注的是设计和实现使用自然语言与人类交流的计算机器。为什么要进行这样的努力呢？鉴于不言而喻的观察，即人类使用自然语言进行交流是最容易和最有效的，因此，从原则上讲，它是人类和机器互动的最简单和最有效的方式；随着技术在我们周围扩散，这种互动将变得越来越重要。在其最雄心勃勃的时候，NLP研究的目的是设计人工智能系统的语言输入-输出组件，能够像人类一样流利和灵活地使用语言。科幻小说中的机器人是典型的：比他们的创造者更强大、更聪明，并能获得更多的知识，但他们对语言的掌握是人类的。即使是最热衷于人工智能研究的人也不得不承认，像库布里克的《2001：太空漫游》中的HAL这样的机器人仍然牢牢地属于科幻小说的范畴。然而，在不那么雄心勃勃的领域已经取得了一些成功，那里的研究问题可以更精确地定义，因此也更容易解决。机器翻译就是这样一个领域，它在当代经济、政治和文化生活的国际主义中找到了现成的应用。其他成功的应用领域是信息理解系统，该系统提取文本数据源（如新闻报道或银行电报）的命题内容的有用元素，以及信息系统（如数据库）的前端，在这些系统中可以用自然语言进行查询并给出答复。本手册是关于这些和其他类型的NLP系统的设计。在整个过程中，重点是可实施系统的实用工具和技术；尽量减少投机性研究，并排除论战。

自然语言处理的符号方法。

符号化和句子分割。

词法分析。

解析技术。

语义分析.

话语结构和意图识别。

自然语言生成.

智能写作帮助.

数据库接口.

信息提取.

从数据库生成报告。

多媒体演示文稿的生成。

机器翻译.

对话系统。TRAINS-96中从理论到实践.

第二部分 NLP的经验方法.

自然语言处理的经验方法.

数据密集型语言学的语料库创建.

语音部分标记.

对齐。

语境词的相似性。

计算相似性。

搭配.

统计学解析.

作者身份识别和计算风格测量.

词汇知识获取.

基于例子的机器翻译.

词义辨析.

第三部分 NLP的人工神经网络方法.

基于人工神经网络的NLP: 介绍.

知识表示.

语法推理，自动归纳，和语言获取.

基于ANN的自然语言处理的符号方法。

基于ANN的自然语言处理的亚符号方法。

基于ANN的自然语言处理的混合方法。

用句法神经网络识别字符.

用神经网络压缩文本.

信息检索和数据库查询的神经架构.

文本数据挖掘.

文本和话语理解: DISCERN系统.