A Simple Approach to Use Bilingual Information Sources for Word Alignment

Resumen

In this paper we present a new and simple method for using sources of bilingual information for word alignment between parallel segments of text. This method can be used "on the fly", since it does not need to be trained. In addition, it can also be applied on comparable corpora. We compare our method to the state-of-the-art tool GIZA++, widely used for word alignment, and we obtain very similar results.