Paper: RNN-based Derivation Structure Prediction for SMT

ACL ID P14-2126
Title RNN-based Derivation Structure Prediction for SMT
Venue Annual Meeting of the Association of Computational Linguistics
Session Main Conference
Year 2014
Authors

In this paper, we propose a novel deriva- tion structure prediction (DSP) model for SMT using recursive neural network (RNN). Within the model, two steps are involved: (1) phrase-pair vector represen- tation, to learn vector representations for phrase pairs; (2) derivation structure pre- diction, to generate a bilingual RNN that aims to distinguish good derivation struc- tures from bad ones. Final experimental results show that our DSP model can sig- nificantly improve the translation quality.