Approach to Automatic Translation Template Acquisition Based on Unannotated Bilingual Grammar Induction

In this paper, we propose a new approach which can automatically acquire translation templates from the unannotated bilingual spoken language corpora in the domain of travel information accessing. In the approach, two basic algorithms named grammar induction algorithm and dynamic programming algorithm are adopted. Our approach is an unsupervised, statistical, data-driven method which avoids the parsing procedure. Firstly, the semantic groups and the phrasal structure groups are extracted from both the source language and the target language. Then, the dynamic programming algorithm is used to align these phrasal structure groups. The aligned phrasal structure groups are treated as the translation templates. And the preliminary experimental result is also desirable.