Document Matching

We propose a completely unsupervised method for mining parallel sentences from quasi-comparable bilingual texts which have very different sizes, and which include both in-topic and off-topic documents. We discuss and …...