CN-119204031-B - Translation text quality evaluation method, device, computer equipment and storage medium

CN119204031BCN 119204031 BCN119204031 BCN 119204031BCN-119204031-B

Abstract

The embodiment of the application provides a translation text quality assessment method, a translation text quality assessment device, computer equipment and a storage medium. The method comprises the steps of obtaining a translation source text and a reference translation of a translation to be evaluated, performing quality evaluation on the reference translation based on the translation source text to obtain reference quality of the reference translation, performing quality scoring on the translation to be evaluated through the translation source text to obtain a first quality score, performing quality scoring on the translation to be evaluated through the reference translation to obtain a second quality score, adjusting weights of the first quality score according to the reference quality of the reference translation to obtain a first target quality score, and adjusting weights of the second quality score to obtain a second target quality score, wherein the weights of the first quality score decrease along with the increase of the reference quality, and obtaining a quality evaluation result of the translation to be evaluated based on the first target quality score and the second target quality score. Therefore, the accuracy of quality evaluation of the translation text can be effectively improved.

Inventors

LOU LIANZHANG
XIANG YANG
HOU YONGSHUAI
DENG JIANHAO
PAN YOUCHENG

Assignees

鹏城实验室

Dates

Publication Date: 20260505
Application Date: 20240809

Claims (8)

1. A method for evaluating quality of translated text, the method comprising: Acquiring a translation source text and a reference translation of a translation to be evaluated, and performing quality evaluation on the reference translation based on the translation source text to obtain reference quality of the reference translation; Extracting a first target feature from the translation source text, wherein the first target feature comprises at least one of grammatical features, sentence features and semantic features, extracting a second target feature from the translation to be evaluated, wherein the feature attribute of the second target feature corresponds to the feature attribute of the first target feature, calculating an alignment degree index of the first target feature and the corresponding second target feature, and determining a first quality score of the translation to be evaluated based on the alignment degree index; performing quality scoring on the translation to be evaluated through the reference translation to obtain a second quality score; Obtaining a first weight coefficient and a second weight coefficient, multiplying the first weight coefficient by the reference quality to obtain a first intermediate coefficient, obtaining a target weight of the translation source text based on the sum of the first intermediate coefficient and the second weight coefficient, obtaining a first target quality score according to the product of the target weight and the first quality score, and adjusting the second quality score to obtain a second target quality score, wherein the weight of the first quality score is reduced along with the increase of the reference quality; And obtaining a quality evaluation result of the translation to be evaluated based on the first target quality score and the second target quality score.
2. The method for evaluating the quality of a translated text according to claim 1, wherein the adjusting the weight of the second quality score to obtain a second target quality score includes: Obtaining the complement of the target weight; and obtaining a second target quality fraction according to the product of the complement of the target weight and the second quality fraction.
3. The method for evaluating the quality of a translated text according to claim 1, wherein the performing the quality evaluation on the reference translation based on the translation source text to obtain the reference quality of the reference translation includes: Acquiring a plurality of translation source texts from a translation source text set, and acquiring a plurality of reference translations from a reference translation set; Consistency comparison is carried out on each translation source text and the corresponding reference translation, and consistency scores used for representing the quality of the reference translation are obtained; and obtaining the sum of a plurality of the consistency scores of the plurality of the reference translations, and obtaining the reference quality of the reference translations through the ratio of the sum of the plurality of the consistency scores to the number of the reference translations.
4. The method for evaluating the quality of a translated text according to claim 1, wherein the evaluating the quality of the reference translation based on the translation source text, to obtain the reference quality of the reference translation, further comprises: acquiring a plurality of translation source texts from a translation source text set, and acquiring a plurality of reference translations from a reference translation set; For each reference translation, carrying out consistency comparison on the reference translation and the corresponding translation source text to obtain consistency scores for representing the quality of different reference translations; And taking the consistency score as the reference quality of the corresponding reference translation.
5. The method for evaluating the quality of a translated text according to claim 1, wherein the scoring the quality of the translation to be evaluated by the reference translation to obtain a second quality score comprises: vector encoding is carried out on the reference translation, and a first vector encoding is obtained; vector encoding is carried out on the translation to be evaluated, and a second vector encoding is obtained; And calculating the semantic similarity of the first vector code and the corresponding second vector code, and determining a second quality score of the translation to be evaluated based on the semantic similarity.
6. A translation text quality evaluation apparatus, the apparatus comprising: The evaluation module is used for acquiring a translation source text and a reference translation text of the translation to be evaluated, and carrying out quality evaluation on the reference translation text based on the translation source text to obtain reference quality of the reference translation; A first scoring module, configured to extract a first target feature from the translation source text, where the first target feature includes at least one of a grammatical feature, a sentence feature, and a semantic feature, extract a second target feature from the translation to be evaluated, where a feature attribute of the second target feature corresponds to a feature attribute of the first target feature, calculate an alignment index of the first target feature and the corresponding second target feature, and determine a first quality score of the translation to be evaluated based on the alignment index; the second scoring module is used for scoring the quality of the translation to be evaluated through the reference translation to obtain a second quality score; The adjusting module is used for obtaining a first weight coefficient and a second weight coefficient, multiplying the first weight coefficient by the reference quality to obtain a first intermediate coefficient, obtaining a target weight of the translation source text based on the sum of the first intermediate coefficient and the second weight coefficient, obtaining a first target quality score according to the product of the target weight and the first quality score, and adjusting the second quality score to obtain a second target quality score, wherein the weight of the first quality score is reduced along with the increase of the reference quality; and the acquisition module is used for acquiring the quality evaluation result of the translation to be evaluated based on the first target quality score and the second target quality score.
7. A computer device, characterized in that it comprises a memory storing a computer program and a processor implementing the translated text quality evaluation method according to any one of claims 1 to 5 when executing the computer program.
8. A computer-readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the translated text quality evaluation method of any one of claims 1 to 5.

Description

Translation text quality evaluation method, device, computer equipment and storage medium Technical Field The present application relates to the field of natural language processing technologies, and in particular, to a method and apparatus for evaluating quality of translated text, a computer device, and a storage medium. Background Machine translation refers to target language text generated by translation of a computer program, and can be automatically converted by the computer program based on source language text through an algorithm. In order to ensure accuracy and reliability of translation so as to provide high quality translation services in various application scenarios, quality assessment of machine translations is often required. In the related art, quality evaluation of a machine translation may be performed by comparing a reference translation with the machine translation. However, when the quality of the reference translation is uneven and the machine faces the wrong reference translation, the wrong evaluation is given to the current machine translation, and the accuracy of quality evaluation to the machine translation is reduced. Disclosure of Invention The embodiment of the application mainly aims to provide a translation text quality assessment method, a device, computer equipment and a storage medium, which can effectively improve the accuracy of quality assessment of translation text. To achieve the above object, a first aspect of an embodiment of the present application provides a method for evaluating quality of translated text, including: Acquiring a translation source text and a reference translation of a translation to be evaluated, and performing quality evaluation on the reference translation based on the translation source text to obtain reference quality of the reference translation; Performing quality scoring on the translation to be evaluated through the translation source text to obtain a first quality score; performing quality scoring on the translation to be evaluated through the reference translation to obtain a second quality score; According to the reference quality of the reference translation, carrying out weight adjustment on the first quality score to obtain a first target quality score, and carrying out weight adjustment on the second quality score to obtain a second target quality score, wherein the weight of the first quality score is reduced along with the increase of the reference quality; And obtaining a quality evaluation result of the translation to be evaluated based on the first target quality score and the second target quality score. Accordingly, a second aspect of an embodiment of the present application proposes a translated text quality evaluation apparatus, the apparatus comprising: The evaluation module is used for acquiring a translation source text and a reference translation text of the translation to be evaluated, and carrying out quality evaluation on the reference translation text based on the translation source text to obtain reference quality of the reference translation; the first scoring module is used for scoring the quality of the translation to be evaluated through the translation source text to obtain a first quality score; the second scoring module is used for scoring the quality of the translation to be evaluated through the reference translation to obtain a second quality score; The adjusting module is used for adjusting the weight of the first quality score according to the reference quality of the reference translation to obtain a first target quality score, and adjusting the weight of the second quality score to obtain a second target quality score, wherein the weight of the first quality score is reduced along with the increase of the reference quality; and the acquisition module is used for acquiring the quality evaluation result of the translation to be evaluated based on the first target quality score and the second target quality score. In some embodiments, the adjustment module is further configured to: determining a target weight of the translation source text according to the reference quality of the reference translation; and obtaining a first target quality fraction according to the product of the target weight and the first quality fraction. In some embodiments, the adjustment module is further configured to: Obtaining the complement of the target weight; and obtaining a second target quality fraction according to the product of the complement of the target weight and the second quality fraction. In some embodiments, the adjustment module is further configured to: acquiring a first weight coefficient and a second weight coefficient; Multiplying the reference quality by the first weight coefficient to obtain a first intermediate coefficient; and obtaining the target weight of the translation source text based on the sum of the first intermediate coefficient and the second weight coefficient. In some embodiments, the eval