About the Book
The lexicon has emerged from the study of computational linguistics as a fundamental resource that enables a variety of linguistic processes to operate in the course of tasks ranging from language analysis and text processing to machine translation. Lexicon acquisition, therefore, plays an essential part in getting any natural language processing system to function in the real world. Computers that process natural language require a variety of lexical information in addition to what can be found in standard dictionaries. Moreover, machine-readable dictionaries of the conventional sort have been found to be inadequate for fully supporting realistic natural language processing tasks. This volume describes corpus processing techniques that can be used to extract the additional lexical information required. Bringing together a balanced blend of the theoretical and practical, the contributions provide the most recent look at lexical acquisition techniques and practices. These include coping with unknown lexicalizations, task-driven lexical induction, categorization of lexical units, lexical semantics from corpus analysis, and measuring lexical acquisition. The problems addressed reflect a host of topics including recognition of open compounds, incremental acquisition of meanings from sentence usages, recognition of new senses of existing words, sense disambiguation, recognition of specific classes of works, and recognition and annotation of patterns of word use, each of them important to the overall language analysis process, and each employing text analysis techniques in a useful and theoretically motivated way. Language, Speech, and Communication series
Table of Contents:
Part 1 Introduction: issues in text-based lexicon acquistion - the problem of lexical knowledge acquistion; text-based lexicon acquisition, Branimir Boguraev, James Pustejovsky. Part 2 Coping with unknown lexicalizations: internal and external evidence in the identification and semantic categorization of proper names - introduction, internal versus external evidence, procedure overview - delimit, classify, record, the setting for the process, walking through an example, David D. McDonald; identifying unknown proper names in newswire text - introduction, approaches to name identification, proper names - syntax and semantics, overall algorithm, mention generator, knowledge sources, representation of uncertainty, appositives, coreference, Inderjeet Mani, T. Richard MacMillan; categorizing and standardizing proper nouns for efficient information retrieval - introduction, proper noun boundary identification, proper noun classification scheme, use of proper nouns in matching, performance evaluation, system comparisons, future directions, Woojin Paik, Elizabeth D. Liddy, Edmund Yu, Mary McKenna. Part 3 Task-driven lexicon induction: customizing a lexicon to better suit a computational task - introduction, creating categories from WORDNET, a topic labeler, augmenting categories with relevant terms, combining distant categories, Marti A. Hearst, Hinrich Schutze; towards building contextual representations of word senses using statistical models - contextual representations, acquiring topical context, an upper bound for classifier performance, acquiring local context, Claudia Leacock, Geoffrey Towell, Ellen M. Voorhees. Part 4 Categorization of lexical units: a context driven conceptual clustering method for verb classification - introduction, CIAULA - an algorithm to acquire word clusters, basic level categories, Roberto Basili, Maria-Teresa Pazienza, Paola Velardi; distinguished usage - introduction, information extraction, functionality in lexical semantics, integrating syntactic with semantic constraints, patterns, the current state of pattern acquisition, structural similarity clustering, lexical clustering using edit distance, context clustering, context method results, Scott A. Waterman. (Part contents).
