Machine translation method

Abstract

The invention provides a machine translation method, comprising the steps as follows: 1) source language strings are analyzed to gain share compressed syntax forests of the source language strings; step 2) according to the translation rule set between the known source language and a destination language, the syntax forests are matched so as to gain the share compressed translation forests; step 3) a search algorithm is used to look over the translation forests to generate final translation results. The machine translation method utilizes the share compressed forest to guide the translation, can search the translation results from a plurality of trees, and far exceeds the search space when N-best tree is independently used. On 2.23 millions of parallel bilingual data sets, compared with models decoded by 30-best, the translation speed of machine translation method is 1.4 times faster and the translation performance thereof is 1.7 BLEU points higher.

Description

technical field [0001] The invention belongs to the technical field of natural language processing, in particular, the invention relates to the technical field of tree-based statistical machine translation. Background technique [0002] Syntax-based statistical machine translation models have become the current mainstream translation methods. According to different inputs, they can be divided into string-based models and tree-based models (for tree-based models, please refer to Yang Liu, Qun Liu, and Shouxun Lin.2006 .Tree-tostring alignment template for statistical machine translation. In Proceedings of COLING-ACL, pages 609-616, Sydney, Australia, July. and Liang Huang, Kevin Knight, and Aravind Joshi. 2006. Statistical syntax-directed translation with extended domain of locality. In Proceedings of AMTA.). Compared with the string-based model, the tree-based model uses a syntax tree as input. The advantages are: fast decoding speed, concise model, and no need for binariza...

AI Technical Summary

Problems solved by technology

Compared with the string-based model, the tree-based model uses a syntax tree as input. The advantages are: fast decoding speed, concise model, and no need for binarization; however, this model has a flaw: only a single syntax tree is used to guide translation , since syntax-based models are sensitive to parsing performance, causing parsing errors to introduce false translations

A simple method is to use the N-best tree, decode each tree, and finally output the translation result with the highest probability, but this method has a limited search space and cannot share the same nodes of different trees, resulting in repeated decoding of many nodes , which wastes both space and time

At the same time, since this tree-based statistical machine translation system only uses a single syntax tree to guide translation, it often has the problem of high translation error rate

Images