Exact Decoding of Phrase-Based Translation Models through Lagrangian Relaxation

This paper describes an algorithm for exact decoding of phrase-based translation models, based on Lagrangian relaxation. The method recovers exact solutions, with certificates of optimality, on over 99% of test examples. The method is much more efficient than approaches based on linear programming (LP) or integer linear programming (ILP) solvers: these methods are not feasible for anything other than short sentences. We compare our method to MOSES (Koehn et al., 2007), and give precise estimates of the number and magnitude of search errors that MOSES makes.

• Type: Journal
• Year: 2011
• Where: EMNLP
• Authors: Yin-Wen Chang, Michael Collins
• Keywords: EMNLP 2011, Natural Language Processing, Number And Magnitude, Precise Estimates, Search Errors