Different indexing strategies for multilingual web retrieval: experiments with the EuroGOV corpus

Abstract

Experiments with a multi-lingual web collection are presented. The EuroGOV corpus is the first multi-lingual web corpus for retrieval evaluation. We show how indexes based on words and n-rams are developed for different document parts. Different indexes werde based on the full document content, partial content and the title. The best results were achieved for a title only index based on words.