Gigaword Collections

Most work on extracting parallel text from comparable corpora depends on linguistic re- sources such as seed parallel documents or translation dictionaries. This paper presents a simple baseline approach for …...