Automatic multilingual annotation of EU legislation with Eurovoc descriptors

Abstract

Automatic document annotation from a controlled conceptual thesaurus is useful for establishing precise links between similar documents. This study presents a language independent document annotation system based on features derived from a collocation segmentation method. Using the multilingual conceptual thesaurus EuroVoc, we evaluate the method, comparing it against other language independent methods based on single words and bigrams. Testing the method against the manually tagged multilingual corpus Acquis Communautaire 3.0 (AC) using all descriptors found there, we attain improvements in keyword assignment precision from 50.7 to 57.6 percent over three diverse languages (English, Lithuanian and Finnish) tested. We found high correlation between automatic assignment precision against document length and language features such as inflectiveness and compounding.