Exploring Translation Similarities for Building a Better Sentence Aligner

摘要

The approaches previously used for sentence alignment (sentence length, word correspondence and cognate matching) take into account different aspects of similarity between the source and the target language sentences. In this paper we discuss various aspects of similarity in translated texts that can be used for sentence alignment. We then describe a customizable method for combining several approaches that can exploit these aspects of similarity. This method also includes a novel way of using sentence length for alignment. Finally, the results of evaluation for this composite method (overall and at various stages) are presented. These results are compared with those for some previous approaches.