Extraction of alignment relationships in comparable corpora based on Singular Value Decomposition

With the wide availability of information from the Internet nowadays, the process of automatically establishing alignment relationships between bilingual pairs has become a hot research topic. In this study, an alignment algorithm based on Singular Value Decomposition is proposed for handling crossing relationships between structurally dissimilar language pairs in comparable corpora. Portuguese and Chinese are used as example languages and different features are defined to model the relationship between them. Experiments results show that the proposed approach enhances the alignment accuracy compared with the baseline approach. © 2011 Asian Network for Scientific Information.