Statistical machine translation method with error self-diagnosis and self-correction functions

Abstract

The invention discloses a statistical machine translation method with error self-diagnosis and self-correction functions. The method comprises the following steps: firstly, defining the translation error type, training an error classifier, carrying out translation error classifying on a test set, then mapping the translation error from a target language terminal to a source language terminal and constructing a repeated word graph network, optimizing the repeated word graph network of the source language, and finally, carrying out word graph decoding to obtain a self-correction result. Compared with the current statistical machine translation method, according to the statistical machine translation method with error self-diagnosis and self-correction functions, the translation error rate is effectively lowered, and the translation performance is improved.

Description

technical field [0001] The invention belongs to the technical field of statistical machine translation methods, in particular to a statistical machine translation method with error self-diagnosis and self-correction functions. Background technique [0002] Software localization refers to the process of integrating information related to specific regional settings and information translation when software is transplanted in regions and countries with different cultural and language backgrounds, so as to adapt to local culture and usage habits. Translation plays a vital role in the localization process, and the adaptation to local culture and language directly affects the promotion of the software in that region or country. In the software localization industry, the traditional method is to first use the translation memory (Translation Memory, TM) to search and output the translation examples of the software interface, terms, manuals or technical documents according to the fuz...

AI Technical Summary

Problems solved by technology

[0004] However, the current statistical machine translation systems for industrial applications still cannot really meet the translation requirements due to translation accuracy, so most of them are used as auxiliary tools for human translation, and cannot independently provide translation results with high confidence, mainly in two aspects :

[0005] 1. Insufficient ability to predict translation errors: it is difficult to accurately diagnose and predict potential translation errors in translation results, and this function is important for post-editors to search, judge and correct the errors with high efficiency and low consumption significance

The confidence estimation accuracy of the current method needs to be further improved

[0006] 2. Poor self-correction ability for translation errors: For translation errors automatically diagnosed by the system, there are currently two solutions that can provide self-correction functions

However, none of these methods worked as expected.

Images