Translation text quality evaluation method and device, computer device and storage medium

Abstract

This application provides a method, apparatus, computer device, and storage medium for evaluating the quality of translated text. The method includes: acquiring a source text and a reference translation of the text to be evaluated; evaluating the quality of the reference translation based on the source text to obtain a reference quality of the reference translation; assigning a quality score to the text to be evaluated using the source text to obtain a first quality score; assigning a quality score to the text to be evaluated using the reference translation to obtain a second quality score; 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 decreases as the reference quality increases; and obtaining the quality evaluation result of the text to be evaluated based on the first target quality score and the second target quality score. This effectively improves the accuracy of quality evaluation of translated text.

Description

Technical Field

[0001] This application relates to the field of natural language processing technology, and in particular to a method, apparatus, computer device, and storage medium for evaluating the quality of translated text. Background Technology

[0002] Machine translation refers to target language text generated through computer programs. It is typically produced automatically by a computer program based on source language text, using algorithms. To ensure the accuracy and reliability of the translation and to provide high-quality translation services in various application scenarios, machine translations usually require quality assessment.

[0003] In related technologies, the quality of machine translations can be evaluated by comparing reference translations with machine translations. However, the quality of reference translations varies, and when faced with incorrect reference translations, the machine may give an incorrect evaluation of its current translation, reducing the accuracy of the quality evaluation. Summary of the Invention

[0004] The main objective of this application is to provide a method, apparatus, computer device, and storage medium for evaluating the quality of translated text, which can effectively improve the accuracy of evaluating the quality of translated text.

[0005] To achieve the above objectives, a first aspect of this application proposes a method for assessing the quality of translated text, the method comprising:

[0006] Obtain the source text and reference translation of the translation to be evaluated, and evaluate the quality of the reference translation based on the source text to obtain the reference quality of the reference translation;

[0007] The quality of the translation to be evaluated is scored using the source text to obtain a first quality score.

[0008] A second quality score is obtained by scoring the quality of the translation to be evaluated using the reference translation;

[0009] Based on the reference quality of the reference translation, the weight of the first quality score is adjusted to obtain a first target quality score, and the weight of the second quality score is adjusted to obtain a second target quality score; wherein, the weight of the first quality score decreases as the reference quality increases;

[0010] Based on the first target quality score and the second target quality score, the quality assessment result of the translation to be evaluated is obtained.

[0011] Accordingly, a second aspect of this application provides a translated text quality assessment apparatus, the apparatus comprising:

[0012] The evaluation module is used to obtain the source text and reference translation of the translation to be evaluated, and to evaluate the quality of the reference translation based on the source text to obtain the reference quality of the reference translation.

[0013] The first scoring module is used to score the quality of the translation to be evaluated based on the source text and obtain a first quality score.

[0014] The second scoring module is used to score the quality of the translation to be evaluated based on the reference translation, and obtain a second quality score.

[0015] An adjustment module is used to adjust the weight of the first quality score according to the reference quality of the reference translation to obtain a first target quality score, and to adjust the weight of the second quality score to obtain a second target quality score; wherein the weight of the first quality score decreases as the reference quality increases;

[0016] The acquisition module is used to obtain the quality assessment result of the translation to be evaluated based on the first target quality score and the second target quality score.

[0017] In some embodiments, the adjustment module is further configured to:

[0018] The target weight of the source text to be translated is determined based on the reference quality of the reference translation.

[0019] The first target quality score is obtained by multiplying the target weight by the first quality score.

[0020] In some embodiments, the adjustment module is further configured to:

[0021] Obtain the complement of the target weight;

[0022] The second target quality score is obtained by multiplying the complement of the target weight by the second quality score.

[0023] In some embodiments, the adjustment module is further configured to:

[0024] Obtain the first weighting coefficient and the second weighting coefficient;

[0025] The first intermediate coefficient is obtained by multiplying the reference quality and the first weighting coefficient.

[0026] The target weight of the translated source text is obtained based on the sum of the first intermediate coefficient and the second weight coefficient.

[0027] In some implementations, the evaluation module is further configured to:

[0028] Obtain multiple source texts from the source text set and multiple reference translations from the reference translation set;

[0029] Each source text is compared with the corresponding reference translation to obtain a consistency score that characterizes the quality of the reference translation.

[0030] The sum of the consistency scores of multiple reference translations is obtained, and the reference quality of the reference translation is obtained by the ratio of the sum of the multiple consistency scores to the number of reference translations.

[0031] In some implementations, the evaluation module is further configured to:

[0032] Obtain multiple source texts from the source text set and multiple reference translations from the reference translation set;

[0033] For each of the aforementioned reference translations, the reference translation is compared with the corresponding source text to obtain a consistency score that characterizes the quality of the different reference translations.

[0034] The consistency score is used as the reference quality of the corresponding reference translation.

[0035] In some implementations, the evaluation module is further configured to:

[0036] Extract a first target feature from the source text being translated, wherein the first target feature includes at least one of grammatical features, sentence features, and semantic features;

[0037] Extract a second target feature from the translation to be evaluated, wherein the feature attributes of the second target feature correspond to the feature attributes of the first target feature;

[0038] Calculate the alignment index between the first target feature and the corresponding second target feature, and determine the first quality score of the translation to be evaluated based on the alignment index.

[0039] In some implementations, the second scoring module is further configured to:

[0040] The reference translation is vector-encoded to obtain the first vector code;

[0041] The translation to be evaluated is vector-encoded to obtain a second vector code;

[0042] Calculate the semantic similarity between the first vector code and the corresponding second vector code, and determine the second quality score of the translation to be evaluated based on the semantic similarity.

[0043] To achieve the above objectives, a third aspect of this application provides a computer device, the computer device including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the translation text quality assessment method according to any one of the embodiments of the first aspect of this application.

[0044] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the translation text quality assessment method described in any one of the embodiments of the first aspect of the present application.

[0045] This application embodiment obtains the source text and reference translation of the translation to be evaluated, and performs a quality assessment of the reference translation based on the source text to obtain a reference quality of the reference translation; scores the translation to be evaluated using the source text to obtain a first quality score; scores the translation to be evaluated using the reference translation to obtain a second quality score; adjusts the weight of the first quality score according to the reference quality of the reference translation to obtain a first target quality score, and adjusts the weight of the second quality score to obtain a second target quality score; wherein the weight of the first quality score decreases as the reference quality increases; and obtains the quality assessment result of the translation to be evaluated based on the first target quality score and the second target quality score. Therefore, by considering the different qualities of different reference translations, the reference quality of the corresponding reference translation can be obtained. This allows for adjustments to the first quality score of the source text and the second quality score of the reference translation based on the reference quality. Specifically, when the reference quality of the reference translation is high, the weight of the first quality score is lower; conversely, when the reference quality of the reference translation is low, the weight of the first quality score is higher. This dynamic adjustment of the contribution of the reference quality to the quality assessment of the translation being evaluated enables accurate quality assessment of the translated text under different reference translation qualities. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of the architecture of the translation text quality assessment system provided in the embodiments of this application;

[0047] Figure 2 This is a flowchart of the translation text quality assessment method provided in the embodiments of this application;

[0048] Figure 3 This is a flowchart illustrating the overall process of the translated text quality assessment method provided in this application embodiment;

[0049] Figure 4 This is a schematic diagram of the functional modules of the translation text quality assessment device provided in the embodiments of this application;

[0050] Figure 5 This is a schematic diagram of the hardware structure of the computer device provided in the embodiments of this application. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0052] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0054] Machine translation refers to target language text generated through computer programs. It is typically produced automatically by a computer program based on source language text, using algorithms. To ensure the accuracy and reliability of the translation and to provide high-quality translation services in various application scenarios, machine translations usually require quality assessment.

[0055] In related technologies, the quality of machine translations can be evaluated by comparing reference translations with machine translations. However, the quality of reference translations varies, and when faced with incorrect reference translations, the machine may give an incorrect evaluation of its current translation, reducing the accuracy of the quality evaluation.

[0056] Based on this, embodiments of this application provide a method, apparatus, computer device, and storage medium for evaluating the quality of translated text, which can effectively improve the accuracy of quality evaluation of machine translations.

[0057] The translation text quality assessment method, apparatus, computer equipment, and storage medium provided in this application are specifically described through the following embodiments. First, the translation text quality assessment system in this application is described.

[0058] Please refer to Figure 1 In some embodiments, the translation text quality assessment system provided in this application includes a terminal 11 and a server 12.

[0059] For example, terminal 11 can be a mobile terminal device or a non-mobile terminal device. Specifically, terminal 11 can be a smartphone, tablet, laptop, PDA, in-vehicle terminal device, wearable device, personal digital assistant, or other mobile terminal device, or a personal computer or other non-mobile terminal device. Terminal 11 provides a convenient user interface and data upload portal. For example, when terminal 11 is a smartphone or tablet, the application on the smartphone or tablet allows users to easily upload documents or text fragments for instant translation quality assessment.

[0060] For example, server 12 can be a high-performance server cluster, cloud server, or distributed computing system, etc. Server 12 can centrally process complex algorithms and computational tasks to ensure efficient system operation. For instance, when server 12 is a high-performance server cluster, it can handle a large number of translation quality assessment requests from terminal 11 and perform complex computational tasks, such as calculating the reference quality of the reference translation, scoring the quality of the translation to be evaluated using the source text to obtain a first quality score, scoring the quality of the translation to be evaluated using the reference translation to obtain a second quality score, and adjusting the weights of the first and second quality scores, etc.

[0061] The method for assessing the quality of translated text in this application can be illustrated through the following examples.

[0062] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent will be obtained first. Furthermore, the collection, use, and processing of this data will comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user will be obtained through pop-ups or redirects to confirmation pages. Only after obtaining the user's separate permission or consent will the necessary user-related data for the normal operation of the embodiments of this application be obtained.

[0063] In this embodiment, the description will focus on the translation text quality assessment device, which can be integrated into a computer device. See also Figure 2 , Figure 2 This is a flowchart illustrating the steps of the translation text quality assessment method provided in this application embodiment. Taking the integration of the translation text quality assessment device into a terminal or server as an example, the specific process when the processor on the terminal or server executes the program instructions corresponding to the translation text quality assessment method is as follows:

[0064] Step 101: Obtain the source text and reference translation of the translation to be evaluated, and conduct a quality assessment of the reference translation based on the source text to obtain the reference quality of the reference translation.

[0065] In some implementations, since high-quality reference translations can better reflect the true level of the translation to be evaluated, while low-quality reference translations may lead to evaluation bias, in order to ensure that the evaluation of the translation to be evaluated is not overly reliant on the reference translation when the quality of the reference translation is poor, the quality of the reference translation can be quantified to obtain reference quality, so that the quality of the translation to be evaluated can be accurately and fairly evaluated subsequently.

[0066] The translation to be evaluated can be a machine translation system output that needs to have its quality assessed.

[0067] The source text can be the original, untranslated text, which serves as the input for the machine translation task and is one of the bases for comparing the translation quality with the target translation. Inputting the source text into the machine translation system yields the target translation.

[0068] The reference translation can be obtained manually from the source text, or it can be obtained through crowdsourcing or other means. The reference translation can be used as one of the benchmarks for evaluating the quality of the translation to be evaluated.

[0069] Among them, reference quality can be a measure of the accuracy of the reference translation relative to the source text, used to characterize the translation quality of the reference translation itself.

[0070] For example, the translation to be evaluated can be obtained by a statistical machine translation system based on the source text, or by an online translation service such as Google Translate or Baidu Translate. As long as the translation needs to be evaluated, it can be used as the translation to be evaluated. This application does not impose specific restrictions on the source of the translation to be evaluated.

[0071] For example, the quality of the reference translation can be evaluated based on the source text to obtain the reference quality of the reference translation. Specifically, a trained SRC-only index model can be obtained, and the set of source texts corresponding to all translations to be evaluated can be obtained, denoted as S = {s1, s2, ..., sn}, where si represents the i-th source text. The set of reference translations corresponding to the source text set can also be obtained, denoted as R = {r1, r2, ..., rn}, where ri is the reference translation of si, and each translation to be evaluated has a corresponding ri and si.

[0072] For example, the SRC-on-ly index model can be a bidirectional encoder based on deep learning, or a statistical model based on feature matching, etc.

[0073] Furthermore, the SRC-on-ly index model can be used to evaluate each source text si and its corresponding reference translation ri, to determine the extent to which the reference translation ri retains the information content of the source text si, and obtain the consistency score of each reference translation ri.

[0074] For example, the source text and the corresponding reference translation can first be encoded using high-dimensional vectors, and then the semantic similarity between the source text and the reference translation can be calculated. The degree of semantic similarity is equivalent to the reference quality of the reference translation. Alternatively, the cross-linguistic semantic alignment between the source text and the reference translation can be calculated based on the source text, and this semantic alignment represents the reference quality of the reference translation. Alternatively, professional translators can directly score the reference translation to obtain its reference quality. Specifically, the appropriate method for calculating reference quality can be selected according to the actual situation; this application does not impose specific limitations on this.

[0075] In some implementations, the average consistency score of all reference translations in the overall reference translation set can be calculated as the reference quality of each reference translation. Alternatively, the consistency score of each reference translation can be calculated one by one as the reference quality of each reference translation. Or, the mode of the consistency scores of all reference translations in the reference translation set can be used as the reference quality of each reference translation. This application does not impose specific limitations on this.

[0076] For example, when the reference quality is a number other than 0 to 1, it can be converted to a reference quality between 0 and 1. For instance, if the output is a distance metric, a smaller distance between the source text and the reference translation indicates a higher degree of matching, and this can be converted to a reference quality through normalization.

[0077] In some implementations, to improve the efficiency of evaluating the translation under evaluation while maintaining accuracy, the average quality score of all reference translations can be used as the quality score of all reference translations. This reduces computational resource consumption and makes the entire evaluation process more efficient. For example, step 101, "evaluating the quality of the reference translations based on the source text to obtain the reference quality of the reference translations," may include:

[0078] (101.a1) Obtain multiple source texts from the source text set and multiple reference translations from the reference translation set;

[0079] (101.a2) Compare the consistency of each source text with the corresponding reference translation to obtain a consistency score that characterizes the quality of the reference translation;

[0080] (101.a3) Obtain the sum of multiple consistency scores of multiple reference translations, and obtain the reference quality of the reference translations by the ratio of the sum of multiple consistency scores to the number of reference translations.

[0081] The source text set can be a collection of source texts corresponding to the translation set to be evaluated. Each source text can be used as input to the machine translation system to obtain the corresponding translation to be evaluated.

[0082] The reference translation set can correspond to the source text set. The reference translation set contains standard translation results for each source text and is used as one of the benchmarks for evaluating the quality of the translation to be evaluated.

[0083] The consistency score is a quantitative indicator that measures the degree of matching between the source and reference translations in terms of content, meaning, and expression, used to assess the reference quality of the translation. The consistency score can be calculated during the evaluation process, based on methods such as semantic similarity, cross-linguistic semantic alignment, or manual scoring. The consistency score can be used to characterize the semantic matching degree between the source and reference translations, for example, by calculating semantic similarity.

[0084] For example, in an English-Chinese translation scenario, there is a set of source texts S and a corresponding set of reference translations R. The source text set S contains 5 English sentences, and the reference translation set R contains the standard translations corresponding to the 5 English sentences. It is understandable that there are also 5 translations to be evaluated for each of the 5 English sentences.

[0085] Furthermore, the consistency score between each pair of source texts and their corresponding reference translations can be calculated using the SRC-on-ly index model, and this consistency score can be used as the basis for evaluating the quality of the source texts. For example, the consistency score is 0.92 for s1 and r1; 0.88 for s2 and r2; 0.95 for s3 and r3; 0.90 for s4 and r4; and 0.85 for s5 and r5.

[0086] Furthermore, with 5 reference translations, the sum of the consistency scores is 0.92 + 0.88 + 0.95 + 0.90 + 0.85 = 4.50. Therefore, the ratio of the sum of the consistency scores to the number of reference translations is 0.9. Thus, the reference quality of the reference translations is 0.9, and this 0.9 can be used as the reference quality for each subsequent translation to be evaluated.

[0087] By quantifying the reference quality of the reference translations, we can quantify the degree of matching between the reference translations and the source text, providing a more objective and quantitative basis for the quality assessment of the reference translations. Furthermore, by determining the reference quality of the reference translations through the ratio of the sum of multiple consistency scores to the number of reference translations, we can focus on the overall translation indicators of the reference translations, improving the efficiency and accuracy of the assessment of the translations being evaluated.

[0088] In some implementations, to more accurately reflect the quality differences of individual reference translations and avoid situations where excellent translations are dragged down by inferior ones, the reference quality of each reference translation can be calculated individually to adjust the evaluation strategy accordingly. This is particularly effective when automatically adjusting the target weight of the source text, as it more precisely reflects the personalized impact of the reference translation quality. Therefore, in addition to obtaining the reference quality of a reference translation by the ratio of the sum of multiple consistency scores to the number of reference translations, this application also proposes a scheme to calculate the reference quality of each reference translation individually and adjust the target weight of the corresponding source text accordingly. For example, step 101, "evaluating the quality of the reference translation based on the source text to obtain the reference quality of the reference translation," may further include:

[0089] (101.b1) Obtain multiple source texts from the source text set and multiple reference translations from the reference translation set;

[0090] (101.b2) For each reference translation, the reference translation is compared with the corresponding source text to obtain a consistency score used to characterize the quality of different reference translations.

[0091] (101.b3) The consistency score is used as a reference quality for the corresponding reference translation.

[0092] For example, in an English-Chinese translation scenario, there is a set of source texts S and a corresponding set of reference translations R. The source text set S contains three English sentences, and the reference translation set R contains the standard translations corresponding to the three English sentences. It is understandable that there are also three translations to be evaluated for each of the three English sentences.

[0093] For example, further, the consistency score between each pair of source texts and their corresponding reference translations can be calculated using the SRC-on-ly index model, and this consistency score can be used as the basis for evaluating the quality of the source texts. For instance, the consistency score is 0.92 for s1 and r1; 0.95 for s2 and r2; and 0.90 for s3 and r3.

[0094] In other words, 0.92 can be used as the reference mass for r1, 0.95 as the reference mass for r2, and 0.90 as the reference mass for r3.

[0095] The corresponding set of translations to be evaluated, T, has t1 corresponding to r1 and s1, t2 corresponding to r2 and s2, and t3 corresponding to r3 and s3. The target weight of the first quality score corresponding to s1 is determined based on the reference quality of r1 (0.92), the target weight of the first quality score corresponding to s2 is determined based on the reference quality of r2 (0.95), and the target weight of the first quality score corresponding to s3 is determined based on the reference quality of r3 (0.90).

[0096] By determining the reference quality of each reference translation individually, we can more precisely capture the quality differences of each reference translation, thereby improving the accuracy of the evaluation of the translation to be evaluated.

[0097] Step 102: The quality of the translation to be evaluated is scored by translating the source text to obtain the first quality score.

[0098] In some implementations, in order to assess whether the translation under evaluation accurately conveys the meaning of the source text, the source text can be used to score the quality of the translation under evaluation to determine the fidelity and accuracy of the translation under evaluation.

[0099] The first quality score can be the SRC-only index of the translation to be evaluated, which is the score given solely based on the comparison between the source text and the translation to be evaluated during the evaluation process.

[0100] For example, the fidelity and accuracy of the translation can be quantified by comparing the correspondence between the source text and the translation to be evaluated in terms of grammatical features, sentence features, and semantic features, and using the SRC-only index model. This model considers factors such as lexical matching, grammatical structure consistency, and semantic preservation, and combines these factors into a first quality score between 0 and 1. For instance, if the SRC-only index model considers the translation to be very close to the source text in all key aspects, it might assign a high first quality score, such as 0.95.

[0101] In some implementations, the SRC-on-ly index model can also score the quality of the source text and the translation to be evaluated based on other factors. Besides the SRC-on-ly index model, other methods can be used to determine the first quality score, such as attention-based sequence-to-sequence models, BERT-based scoring models, or statistical methods. This application does not impose specific limitations on these methods.

[0102] In some implementations, to quantify the accuracy, fluency, and fidelity of the translation to be evaluated, an alignment index can be determined by comparing a first target feature in the source text with a second target feature in the translation to be evaluated. A first quality score for the translation to be evaluated is then determined based on this alignment index, thereby improving the accuracy of the evaluation. For example, step 102 may include:

[0103] (102.1) Extract a first target feature from the source text of the translation, wherein the first target feature includes at least one of grammatical features, sentence features and semantic features;

[0104] (102.2) Extract the second target feature from the translation to be evaluated, wherein the feature attributes of the second target feature correspond to the feature attributes of the first target feature;

[0105] (102.3) Calculate the alignment index of the first target feature and the corresponding second target feature, and determine the first quality score of the translation to be evaluated based on the alignment index.

[0106] The first target feature can be a key attribute extracted from the source text, encompassing at least one of grammatical features, sentence features, and semantic features. For example, the first target feature can be a grammatical feature, or it can be a combination of grammatical and sentence features, etc.

[0107] Among them, grammatical features can refer to the features of sentence structure, including parts of speech, collocation of sentence components, tense, voice and other features that represent grammatical norms.

[0108] Among these, statement features can include statement length, complexity, and coherence.

[0109] Among them, semantic features can be the expression of concepts, implicit meanings, cultural characteristics, etc.

[0110] The second target feature can be a feature extracted from the translation to be evaluated. The second target feature corresponds to the first target feature. For example, if the first target feature is the grammatical feature of the source text, then the second target feature is the grammatical feature of the translation to be evaluated. Or, if the first target feature is the grammatical feature, sentence feature, and semantic feature of the source text, then the second target feature is the grammatical feature, sentence feature, and semantic feature of the translation to be evaluated.

[0111] The alignment index is a quantitative indicator of the alignment between the first target feature of the source text and the second target feature of the translation to be evaluated. The alignment index is between 0 and 1. The closer the value is to 1, the higher the feature alignment, that is, the better the quality of the translation to be evaluated.

[0112] In some implementations, the first target feature and the second target feature are not limited to grammatical features, sentence features and semantic features, but can also be other features, as long as they do not deviate from the concept of this application and conform to the actual translation text quality assessment scenario. This application embodiment does not impose specific limitations on this.

[0113] For example, in an English-Chinese translation scenario (or other language translation scenarios), specific grammatical structures (such as passive voice), sentence complexity (ratio of long to short sentences), and specific semantic units (translation accuracy of idioms or proper nouns) from the source text are extracted as the first target features, and corresponding second target features are determined from the translation to be evaluated. Then, using predefined rules or machine learning models (such as the SRC-only index model), the alignment index of these features is calculated, such as by calculating lexical overlap, syntactic similarity, and cosine similarity of semantic vectors to determine the corresponding alignment index.

[0114] In some implementations, when there are multiple first target features and second target features selected, the alignment index of the first target feature and the corresponding second target feature can be calculated one by one and averaged to obtain the final first quality score. Alternatively, a comprehensive alignment index can be directly given as the first quality score through the SRC-only index model.

[0115] By extracting multi-dimensional target features such as grammatical features, sentence features, and semantic features from the source text and the translation to be evaluated, and comparing them one by one, the quality of the translation of the translation to be evaluated can be comprehensively and accurately assessed.

[0116] Step 103: Use the reference translation to score the quality of the translation to be evaluated, and obtain the second quality score.

[0117] In some implementations, in order to improve the comprehensiveness and accuracy of the evaluation of the translation to be evaluated, in addition to scoring the quality of the translation to be evaluated using the source text, the quality of the translation to be evaluated can also be scored using the reference translation, making the quality evaluation results more comprehensive and objective.

[0118] The second quality score can be the REF-only index of the translation to be evaluated, which is the score given only based on the comparison between the reference translation and the translation to be evaluated during the evaluation process.

[0119] For example, the reference translation and the corresponding translation to be evaluated can be vector-encoded separately, and the semantic similarity between the reference translation and the translation to be evaluated can be calculated based on the vector encoding. The semantic similarity can then be used as the second quality score of the translation to be evaluated.

[0120] In some implementations, in addition to the REF-on-ly metric, a second quality score can be determined by assessing the number of editing operations (such as insertions, deletions, and replacements) required to improve the translation to match the reference translation. Other assessment techniques can also be used to score the quality of the translation to be evaluated against the reference translation to obtain a second quality score; this application does not impose specific limitations on these techniques.

[0121] By using reference translations to score the quality of the translation to be evaluated, the degree of fit between the translation to be evaluated and the standard reference translation can be assessed, thus enhancing the objectivity and accuracy of the evaluation.

[0122] In some implementations, to more systematically evaluate translation quality, a second quality score can be determined by calculating the semantic similarity between the reference translation and the translation to be evaluated, thereby improving the accuracy of the evaluation of the translation to be evaluated. For example, step 103 may include:

[0123] (103.1) The reference translation is vector-encoded to obtain the first vector code;

[0124] (103.2) The translation to be evaluated is vector-encoded to obtain the second vector code;

[0125] (103.3) Calculate the semantic similarity between the first vector code and the corresponding second vector code, and determine the second quality score of the translation to be evaluated based on the semantic similarity.

[0126] The first vector code can be a set of numerical vectors obtained by converting the reference translation. The first vector code is used to represent the semantic features of the reference translation, and each word or entire sentence in the reference translation has a corresponding first vector code in a multi-dimensional space.

[0127] The second vector code can be a set of numerical vectors obtained by converting the translation to be evaluated. The first vector code is used to characterize the semantic features of the translation to be evaluated, and each word or the entire sentence of the translation to be evaluated has a corresponding second vector code in a multi-dimensional space.

[0128] Semantic similarity can be a quantitative indicator that measures the degree of semantic similarity between two first vector codes and second vector codes.

[0129] Understandably, in order to ensure that the translation to be evaluated has a comparable representation to the reference translation in the same vector space, the same vector space model and method can be used to encode both the reference translation and the translation to be evaluated, so as to provide a basis for subsequent semantic similarity calculation.

[0130] For example, common methods for calculating the semantic similarity between the first and second vector codes include cosine similarity, Euclidean distance, or more complex semantic matching models, such as the REF-on-ly index model. By comparing the similarity between the first and second vector codes, the degree of closeness between the translation to be evaluated and the reference translation in terms of preserving the original meaning, fluency, and contextual adaptability can be assessed. High semantic similarity means that the translation to be evaluated and the reference translation have a high degree of semantic matching, usually indicating high translation quality. Conversely, low similarity may indicate that the translation to be evaluated has problems such as inaccuracy or unnatural expression.

[0131] It is understandable that both the first and second mass scores are between 0 and 1. If they are not within this range, they can be normalized.

[0132] By converting the reference translation and the translation to be evaluated into vector codes and then calculating the semantic similarity between them, a more objective and quantitative evaluation method can be provided for evaluating the translation to be evaluated, thereby improving the accuracy of the quality evaluation results.

[0133] Step 104: Adjust the weight of the first quality score according to the reference quality of the reference translation to obtain the first target quality score, and adjust the weight of the second quality score to obtain the second target quality score; wherein, the weight of the first quality score decreases as the reference quality increases.

[0134] In some implementations, when the quality of the reference translation is high, its reliability as an evaluation benchmark is stronger. In this case, the weight of the first quality score (based on the quality assessment score of the translated source text) is reduced to ensure that the quality assessment results are closer to the actual perception of translation quality. Conversely, if the quality of the reference translation is low, the weight of the first quality score can be increased to maintain the objectivity and accuracy of the assessment.

[0135] The first target quality score can be the result of adjusting the first quality score obtained based on the translation source text after taking into account the impact of the quality of the reference translation.

[0136] The second target quality score can be an adjustment to the second quality score obtained from the evaluation of the reference translation, after also considering the influence of the reference translation quality. Its adjustment logic is similar to that of the first target quality score, but it uses the complement of the target weight to multiply the second quality score. This means that when the reference translation quality is high, the weight of the second quality score increases, placing greater emphasis on the evaluation criteria of the reference translation.

[0137] For example, the target weight of the source text to be translated can be determined based on the reference quality of the reference translation. Then, the target weight is multiplied by the original first quality score and second quality score respectively to obtain a new first target quality score and second target quality score that reflect the influence of the quality of the reference translation.

[0138] By applying reference quality to both the first and second quality scores simultaneously, a more accurate and comprehensive assessment of the translated text quality can be achieved, improving the practicality and flexibility of the assessment system.

[0139] In some implementations, since the quality of the reference translation directly determines whether it can well represent the expected translation effect, a target weight can be determined based on the reference quality of the reference translation, and the first quality score can be adjusted using the target weight to more reasonably evaluate the quality of the translation to be evaluated based on reference translations of different qualities. For example, step 104, "adjusting the weight of the first quality score according to the reference quality of the reference translation to obtain a first target quality score," may include:

[0140] (104.a1) Determine the target weight of the source text to be translated based on the reference quality of the reference translation;

[0141] (104.a2) The first target quality score is obtained by multiplying the target weight by the first quality score.

[0142] The target weight can be an adjustment factor determined based on the reference quality of the reference translation. The target weight can be used to reflect the reliability of the reference translation and the degree of importance it should be given when evaluating the translation to be evaluated.

[0143] Specifically, when the quality of the reference translation is high, the target weight tends to be smaller. This means that in the overall evaluation, the influence of the first quality score directly based on the translated source text will be relatively reduced, and more trust will be placed on the guidance of the reference translation. Conversely, if the quality of the reference translation is low, the target weight may be larger to maintain the emphasis on the evaluation of fidelity to the translated source text and ensure the objectivity of the evaluation.

[0144] For example, the target weight w can be calculated using the following formula:

[0145] w=α*q+β

[0146] Where q represents the reference quality of the reference translation, α is the first weighting coefficient, β is the second weighting coefficient, and α and β are linear coefficients and bias, respectively. α and β can be set according to the indicators used when calculating the first quality score and the second quality score, such as the SRC-on-ly indicator when calculating the first quality score and the REF-on-ly indicator when calculating the second quality score, or α and β can be set according to the actual situation.

[0147] For example, the first target quality score, score1, can be calculated using the following formula:

[0148] score1=w*s1

[0149] Where w represents the target weight and s1 represents the first quality score.

[0150] The first target quality score is determined by multiplying the target weight by the first quality score. This ensures the rationality and accuracy of the evaluation results regardless of whether the reference translation is of high or low quality, which is beneficial for obtaining accurate quality evaluation results of the translation to be evaluated in the future.

[0151] In some implementations, since the quality of the reference translation directly determines whether it can well represent the expected translation effect, a target weight can be determined based on the reference quality of the reference translation, and the second quality score can be adjusted using the complement of the target weight to more reasonably evaluate the quality of the translation to be evaluated based on reference translations of different qualities. For example, step 104, "adjusting the weight of the second quality score to obtain the second target quality score," may include:

[0152] (104.b1) Obtain the complement of the target weight;

[0153] (104.b2) The second target quality score is obtained by multiplying the complement of the target weight by the second quality score.

[0154] The complement of the target weight can be a supplementary adjustment value made to the target weight when calculating the second quality score in order to balance the adjustment of the first quality score. The sum of the target weight and the complement of the target weight is 1.

[0155] Understandably, setting a complement to the weights of a certain table ensures that the adjustments to the first and second quality scores complement each other during the overall evaluation process, maintaining the balance and rationality of the quality assessment. When the target weight decreases due to the high quality of the reference translation, its complement is used to increase the weight of the second quality score, strengthening the impact of the reference translation quality on the evaluation, and vice versa.

[0156] For example, the complement of the target weight can be determined by the difference between 1 and the target weight.

[0157] For example, the second target quality score, score2, can be calculated using the following formula:

[0158] score2=(1-w)*s2

[0159] Where w represents the target weight, (1-w) represents the complement of the target weight, and s2 represents the second quality score.

[0160] By multiplying the complement of the target weight with the second quality score, the second target quality score is determined. This ensures the rationality and accuracy of the evaluation results regardless of whether the reference translation is of high or low quality, which is beneficial for obtaining accurate quality evaluation results of the translation to be evaluated in the future.

[0161] In some implementations, to adaptively adjust the evaluation strategy across reference translations of different quality levels, the target weight is reduced when the quality of the reference translation is high and increased when the quality of the reference translation is low, to ensure the objectivity and accuracy of the evaluation. For example, (104.a1) may include:

[0162] (104.a1.1) Obtain the first weight coefficient and the second weight coefficient;

[0163] (104.a1.2) The first intermediate coefficient is obtained by multiplying the reference mass and the first weighting coefficient;

[0164] (104.a1.3) The target weight of the source text to be translated is obtained based on the sum of the first intermediate coefficient and the second weight coefficient.

[0165] The first weighting coefficient can be a pre-defined linear coefficient.

[0166] The second weighting coefficient can be a pre-set bias.

[0167] The first intermediate coefficient can be a temporary coefficient generated during the calculation process.

[0168] For example, the target weight w can be calculated using the following formula:

[0169] w=α*q+β

[0170] Where q represents the reference quality of the reference translation, α*q is the first intermediate coefficient, α is the first weight coefficient, β is the second weight coefficient, and α and β are linear coefficients and bias, respectively. α and β can be set according to the indicators used when calculating the first quality score and the second quality score, such as the SRC-on-ly indicator when calculating the first quality score and the REF-on-ly indicator when calculating the second quality score, or α and β can be set according to the actual situation.

[0171] By determining the target weight corresponding to the source text to be translated, that is, the target weight for adjusting the first quality score, the contribution of the first quality score and the second quality score can be adjusted. This enables a more efficient, accurate, and adaptable assessment mechanism in translation quality evaluation, ensuring the flexibility and effectiveness of translation quality control.

[0172] Step 105: Based on the first target quality score and the second target quality score, obtain the quality assessment result of the translation to be evaluated.

[0173] In some implementations, in order to comprehensively evaluate the translation to be evaluated, two different perspectives on translation quality assessment can be considered together, that is, the first target quality score and the second target quality score are added together, so as to provide a more objective and adaptable comprehensive quality assessment result for the translation to be evaluated.

[0174] The quality assessment result can be a numerical value or grade obtained by comprehensively calculating the first target quality score and the second target quality score, which is used to comprehensively and objectively reflect the translation quality of the translation to be evaluated.

[0175] For example, when the quality assessment result is a numerical value, the quality assessment result can be represented by a score. Specifically, the calculation method for the quality assessment result is as follows:

[0176] score=w*s1+(1-w)*s2

[0177] Where w is the target weight, s1 represents the first quality score, and s2 represents the second quality score.

[0178] Alternatively, score = score1 + score2, where score1 represents the first target quality score and score2 represents the second target quality score.

[0179] In some implementations, the first target quality score and the second target quality score can be added together to obtain the quality assessment result of the translation to be evaluated; alternatively, the quality assessment level of the translation to be evaluated, such as high quality, can be determined based on the value obtained by adding the first target quality score and the second target quality score. The embodiments of this application do not limit the form of the quality assessment result.

[0180] This application embodiment obtains the source text and reference translation of the translation to be evaluated, and performs a quality assessment of the reference translation based on the source text to obtain a reference quality of the reference translation; scores the translation to be evaluated using the source text to obtain a first quality score; scores the translation to be evaluated using the reference translation to obtain a second quality score; adjusts the weight of the first quality score according to the reference quality of the reference translation to obtain a first target quality score, and adjusts the weight of the second quality score to obtain a second target quality score; wherein the weight of the first quality score decreases as the reference quality increases; and obtains the quality assessment result of the translation to be evaluated based on the first target quality score and the second target quality score. In this way, by considering the different qualities of different reference translations, the reference quality of the corresponding reference translation can be obtained. Then, the first quality score corresponding to the source text and the second quality score corresponding to the reference translation can be adjusted based on the reference quality. This achieves the following: when the reference quality of the reference translation is high, the weight of the first quality score is low; when the reference quality of the reference translation is low, the weight of the first quality score is high. This dynamically adjusts the contribution of the reference quality to the quality assessment of the translation to be evaluated. This application can achieve accurate quality assessment of the translation to be evaluated under reference translations of different qualities.

[0181] Please see Figure 3 This application combines Figure 3 This paper introduces the overall process of translation text quality assessment methods. For example, firstly, the reference quality of the source translation can be calculated, and the SRC-only index is selected to score the quality of the translation to be assessed using the source text, resulting in a first quality score. Next, the REF-only index is selected to calculate the quality score of the translation to be assessed using the source translation, resulting in a second quality score. Further, based on the reference quality of the source translation, the target weight of the first quality score can be adjusted, and the first quality score can be calculated to obtain a first target quality score. Similarly, the second quality score can be calculated to obtain a second target quality score. Finally, the first and second target quality scores are added together to obtain the quality assessment result of the translation to be assessed. Alternatively, the quality assessment level of the translation to be assessed can be determined based on the value obtained by adding the first and second target quality scores.

[0182] This application embodiment obtains the source text and reference translation of the translation to be evaluated, and performs a quality assessment of the reference translation based on the source text to obtain a reference quality of the reference translation; scores the translation to be evaluated using the source text to obtain a first quality score; scores the translation to be evaluated using the reference translation to obtain a second quality score; adjusts the weight of the first quality score according to the reference quality of the reference translation to obtain a first target quality score, and adjusts the weight of the second quality score to obtain a second target quality score; wherein the weight of the first quality score decreases as the reference quality increases; and obtains the quality assessment result of the translation to be evaluated based on the first target quality score and the second target quality score. In this way, by considering the different qualities of different reference translations, the reference quality of the corresponding reference translation can be obtained. Then, the first quality score corresponding to the source text and the second quality score corresponding to the reference translation can be adjusted based on the reference quality. This achieves the following: when the reference quality of the reference translation is high, the weight of the first quality score is low; when the reference quality of the reference translation is low, the weight of the first quality score is high. This dynamically adjusts the contribution of the reference quality to the quality assessment of the translation to be evaluated. This application can achieve accurate quality assessment of the translation to be evaluated under reference translations of different qualities.

[0183] Please see Figure 4 This application also provides a translation text quality assessment device, which can implement the above-described translation text quality assessment method. The translation text quality assessment device includes:

[0184] Evaluation module 41 is used to obtain the source text and reference translation of the translation to be evaluated, and to evaluate the quality of the reference translation based on the source text to obtain the reference quality of the reference translation.

[0185] The first scoring module 42 is used to score the quality of the translation to be evaluated based on the source text and obtain the first quality score.

[0186] The second scoring module 43 is used to score the quality of the translation to be evaluated by referring to the translation and obtain a second quality score.

[0187] The adjustment module 44 is used to adjust the weight of the first quality score according to the reference quality of the reference translation to obtain a first target quality score, and to adjust the weight of the second quality score to obtain a second target quality score; wherein, the weight of the first quality score decreases as the reference quality increases;

[0188] The acquisition module 45 is used to obtain the quality assessment result of the translation to be evaluated based on the first target quality score and the second target quality score.

[0189] The specific implementation of this translation text quality assessment device is basically the same as the specific embodiment of the translation text quality assessment method described above, and will not be repeated here. Subject to meeting the requirements of the embodiments of this application, the translation text quality assessment device may also be equipped with other functional modules to implement the translation text quality assessment method in the above embodiments.

[0190] This application also provides a computer device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method for assessing the quality of translated text. This computer device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0191] Please see Figure 5 , Figure 5 The hardware structure of a computer device according to another embodiment is illustrated. The computer device includes:

[0192] The processor 51 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0193] The memory 52 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 52 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 52 and is called and executed by the processor 51 using the translated text quality assessment method of the embodiments of this application.

[0194] Input / output interface 53 is used to implement information input and output;

[0195] The communication interface 54 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0196] Bus 55 transmits information between various components of the device (e.g., processor 51, memory 52, input / output interface 53, and communication interface 54);

[0197] The processor 51, memory 52, input / output interface 53, and communication interface 54 are connected to each other within the device via bus 55.

[0198] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for assessing the quality of translated text.

[0199] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0200] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0201] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0202] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0203] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0204] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0205] It should be understood that in this application, "at least one" and "several" refer to one or more, and "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0206] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0207] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0208] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0209] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0210] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for assessing the quality of translated texts, characterized in that, The method includes: Obtain the source text and reference translation of the translation to be evaluated, and evaluate the quality of the reference translation based on the source text to obtain the reference quality of the reference translation; A first target feature is extracted from the source text, wherein the first target feature includes at least one of grammatical features, sentence features, and semantic features. A second target feature is extracted from the translation to be evaluated, wherein the feature attributes of the second target feature correspond to the feature attributes of the first target feature. An alignment index between the first target feature and the corresponding second target feature is calculated, and a first quality score of the translation to be evaluated is determined based on the alignment index. A second quality score is obtained by scoring the quality of the translation to be evaluated using the reference translation; A first weighting coefficient and a second weighting coefficient are obtained. A first intermediate coefficient is obtained by multiplying the reference quality and the first weighting coefficient. The target weight of the translated source text is obtained based on the sum of the first intermediate coefficient and the second weighting coefficient. A first target quality score is obtained by multiplying the target weight and the first quality score. The weight of the second quality score is adjusted to obtain a second target quality score. The weight of the first quality score decreases as the reference quality increases. Based on the first target quality score and the second target quality score, the quality assessment result of the translation to be evaluated is obtained.

2. The method for assessing the quality of translated texts according to claim 1, characterized in that, The step of adjusting the weights of the second quality score to obtain the second target quality score includes: Obtain the complement of the target weight; The second target quality score is obtained by multiplying the complement of the target weight by the second quality score.

3. The method for assessing the quality of translated texts according to claim 1, characterized in that, The process of evaluating the quality of the reference translation based on the source text to obtain the reference quality of the reference translation includes: Obtain multiple source texts from the source text set and multiple reference translations from the reference translation set; Each source text is compared with the corresponding reference translation to obtain a consistency score that characterizes the quality of the reference translation. The sum of the consistency scores of multiple reference translations is obtained, and the reference quality of the reference translation is obtained by the ratio of the sum of the multiple consistency scores to the number of reference translations.

4. The method for assessing the quality of translated texts according to claim 1, characterized in that, The process of evaluating the quality of the reference translation based on the source text to obtain the reference quality of the reference translation also includes: Obtain multiple source texts from the source text set and multiple reference translations from the reference translation set; For each of the aforementioned reference translations, the reference translation is compared with the corresponding source text to obtain a consistency score that characterizes the quality of the different reference translations. The consistency score is used as the reference quality of the corresponding reference translation.

5. The method for assessing the quality of translated texts according to claim 1, characterized in that, The step of scoring the quality of the translation to be evaluated using the reference translation to obtain a second quality score includes: The reference translation is vector-encoded to obtain the first vector code; The translation to be evaluated is vector-encoded to obtain a second vector code; Calculate the semantic similarity between the first vector code and the corresponding second vector code, and determine the second quality score of the translation to be evaluated based on the semantic similarity.

6. A device for assessing the quality of translated texts, characterized in that, The device includes: The evaluation module is used to obtain the source text and reference translation of the translation to be evaluated, and to evaluate the quality of the reference translation based on the source text to obtain the reference quality of the reference translation. The first scoring module is used to extract a first target feature from the source text of the translation, wherein the first target feature includes at least one of grammatical features, sentence features and semantic features; extract a second target feature from the translation to be evaluated, wherein the feature attributes of the second target feature correspond to the feature attributes of the first target feature; calculate the alignment index between 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 to score the quality of the translation to be evaluated based on the reference translation, and obtain a second quality score. An adjustment module is used to obtain a first weighting coefficient and a second weighting coefficient, multiply the reference quality by the first weighting coefficient to obtain a first intermediate coefficient, obtain the target weight of the translated source text based on the sum of the first intermediate coefficient and the second weighting coefficient, obtain a first target quality score by multiplying the target weight by the first quality score, and adjust the weight of the second quality score to obtain a second target quality score; wherein, the weight of the first quality score decreases as the reference quality increases; The acquisition module is used to obtain the quality assessment 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, The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the translation text quality assessment method according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the translation text quality assessment method according to any one of claims 1 to 5.

Images