Task Alternation in Parallel Sentence Retrieval for Twitter Translation
Felix Hieber, Laura Jehl, Stefan Riezler · 2013
We present an approach to mine com-parable data for parallel sentences us-ing translation-based cross-lingual infor-mation retrieval (CLIR). By iteratively al-ternating between the tasks of retrieval and translation, an initial general-domain model is allowed to adapt to in-domain data. Adaptation is done by training the translation system on a few thousand sen-tences retrieved in the step before. Our setup is time- and memory-efficient and of similar quality as CLIR-based adaptation on millions of parallel sentences. 1