Learning Inter-Lingual Document Representations via Concept Compression

Marc Lenz, Tsegaye Misikir Tashu, Tomás Horváth

Abstract: In this work, we proposed a novel approach to derive inter-lingual document representations. The introduced methods aim to enhance the quality of content-based Multilingual Document Recommendation and information retrieval Systems. The main idea centers around creating inter-lingual representations by using mappings to align monolingual representation spaces. According to the experimental results carried out on JRC-Acquis and EU bookshop multilingual corpora, the proposed concept compression approach has outperformed the traditional cross-lingual retrieval and recommendations methods.