Technology for Automatic Dictionary Compilation

本研究实现了包含拉丁字母人名及其片假名音译的双语人名词典的自动编纂。汇编过程是完全自动化的，包括以下内容：(1)从日文文本中自动收集片假名人名，(2)自动反音译片假名人名，(3)自动选择标题，(4)自动选择每个标题的例子(全名翻译)，以及(5)以书本形式自动排版词典。这本词典收录了150,000个全名翻译。

This study achieved automatic compilation of bilingual person-name dictionary, which contains person names in Latin alphabet and their Katakana transliterations. The compilation process is fully automated, which consists of the followings :(1) automatic collection of Katakana person-names from Japanese texts,(2) automatic back-transliteration of Katakana person-names,(3) automatic selection of headwords,(4) automatic selection of examples(translations of full-names) for every headword, and(5) automatic typesetting of the dictionary in book-style. This dictionary contains 150, 000 full-name translations.