TFB 32: Automatische Exzerption: Corpusbasierte Materialbeschaffung für die Lexikographie

词典会以相对较短的时间间隔定期更新，以保持最新。在更新过程中，词典编写者阅读报纸、书籍等，并注意到所有与词典相关的语言现象。这项任务将通过该项目得到NLP软件(自然语言处理)的支持：--这些工具将对大量文本(通常超过1亿字的新闻文本)进行分析，以寻找新的单词和单词组合；--将根据语言和词典标准对文本语料库中发现的语言现象进行分类；--文本中的材料将与正在修订的词典的内容进行比较，结果将在互动工具中显示。该项目基于合作研究中心340(“Theoretische Grundlagen für dieComputerLanguistik”)的词汇习得技术。特别是在2002年，出版合作伙伴Duden BiFab AG(Mannheim)和Langenscheidt KG(München)开发和评估了识别和分类多词表达的新程序。此外，已经指定了用于词典编纂的系统架构和图形用户界面(GUI)&gt；用于部分所需功能的第一个图形用户界面原型已经可用。

Dictionaries are regularly updated, in relatively short intervals, tokeep them up to date. In the updating process, dictionary authors readnewspapers, books, etc. and they note all linguistic phenomenarelevant for their dictionary. This task will be supported, throughthe project, by means of NLP software (Natural Language Processing):-- The tools will analyse large amounts of text (usually over 100million words of news texts) to find new words and word combinations;-- linguistic phenomena found in the text corpora will be classifiedaccording to linguistic and lexicographic criteria; -- material found in the texts will be compared with the contents of the dictionary under revision, and the result will be displayed in an interactive tool. The project is based on lexical acquisition technology from thecollaborative research centre 340 ("Theoretische Grundlagen für dieComputerlinguistik"). In 2002, in particular new procedures for theidentification and classification of multiword expressions have beendeveloped and evaluated by the publishing partners, Duden BIFAB AG(Mannheim) and Langenscheidt KG (München). In addition, the system architecture and the Graphical User Interface (GUI) for lexicography have been specified> A first GUI prototype for part of the functions needed is available.