UAIC: Participation in INFILE@CLEF Task

Abstract

This year marked UAIC’s first participation at the INFILE@CLEF competition. The purpose of this campaign is the evaluation of cross-language filtering systems, which is to successfully build an automated system that separates relevant from non-relevant documents written in different languages with respect to a given profile. For the batch filtering task, participants are provided with the whole document collection and must return the list of relevant documents for each topic. We achieved good results in filtering documents, also obtaining the highest originality score, when having English as target language. Our team was also the only one who submitted runs for cross-lingual and multilingual batch filtering, with French and English/French as target languages. A brief description of our system, including presentation of the Parsing, Indexing and Filtering modules is given in this paper, as well as the results of the submitted runs.