Data processing method and device, electronic equipment and medium

Abstract

The invention provides a data processing method and device, electronic equipment and a medium, and belongs to the technical field of information. The method comprises the following steps: obtaining a result of performing target language translation on a source language text sample through at least two translation service providers to obtain at least two first translation results; according to the at least two first translation results, evaluation values of the at least two translation service providers in each dimension in multiple dimensions are determined, and the multiple dimensions comprise at least two items of consistency, cultural adaptability, fidelity, understandability, real-time performance, accuracy and price; determining a comprehensive evaluation value of the at least two translation service providers according to the evaluation values of the at least two translation service providers in each dimension; and selecting a target translation service provider from the at least two translation service providers to translate the source language text according to the comprehensive evaluation values of the at least two translation service providers. The method has relatively high selection efficiency and meets interpretation requirements of a real-time use scene.

Description

Technical Field

[0001] The present disclosure belongs to the field of information technology, and particularly relates to a data processing method, apparatus, electronic device, and medium. Background Art

[0002] Currently, electronic devices such as smart speakers can access different simultaneous interpretation service providers to provide translation services. However, the translation quality of each service provider varies in different fields. For example, service provider A performs better in the e-commerce field, while service provider B is more accurate in the software field. In the prior art, users need to manually switch the simultaneous interpretation service provider in the device APP. However, ordinary users are difficult to timely understand the translation performance of each service provider in different fields, resulting in low selection efficiency and difficulty in meeting the translation requirements of real-time usage scenarios. Summary of the Invention

[0003] The objective of the embodiments of the present disclosure is to provide a data processing method, apparatus, electronic device, and medium, which can solve the problems of low selection efficiency of translation service providers in the prior art and difficulty in meeting the translation requirements of real-time usage scenarios.

[0004] In a first aspect, the embodiments of the present disclosure provide a data processing method, which includes:

[0005] Obtain the results of translating a source language text sample into a target language by at least two translation service providers to obtain at least two first translation results;

[0006] According to the at least two first translation results, respectively determine the evaluation values of the at least two translation service providers in each dimension among multiple dimensions, where the multiple dimensions include at least two of the following: consistency, cultural adaptability, fidelity, comprehensibility, real-time performance, accuracy, price. The consistency is used to indicate the similarity of the translation results of the at least two translation service providers. The cultural adaptability is used to indicate the translation accuracy of the translation service provider for culturally characteristic vocabulary. The fidelity is used to indicate the matching degree of the translation results of the translation service provider with the source language text. The comprehensibility includes at least one of grammatical accuracy, readability, fluency, and ambiguity;

[0007] According to the evaluation values of the at least two translation service providers in each dimension, determine the comprehensive evaluation values of the at least two translation service providers;

[0008] According to the comprehensive evaluation values of the at least two translation service providers, select a target translation service provider from the at least two translation service providers;

[0009] Translate the source language text through the target translation service provider.

[0010] In a second aspect, embodiments of the present disclosure provide a data processing apparatus, including:

[0011] A first acquisition module, configured to acquire results of target language translation of a source language text sample by at least two translation service providers, so as to obtain at least two first translation results;

[0012] A first determination module, configured to respectively determine evaluation values of the at least two translation service providers in each dimension among multiple dimensions according to the at least two first translation results, where the multiple dimensions include at least two of the following: consistency, cultural adaptability, fidelity, understandability, real-time performance, accuracy, and price. The consistency is used to indicate the similarity of the translation results of the at least two translation service providers. The cultural adaptability is used to indicate the translation accuracy of the translation service provider for culturally characteristic vocabulary. The fidelity is used to indicate the matching degree of the translation results of the translation service provider with the source language text. The understandability includes at least one of grammatical accuracy, readability, fluency, and ambiguity;

[0013] A second determination module, configured to determine a comprehensive evaluation value of the at least two translation service providers according to the evaluation values of the at least two translation service providers in the respective dimensions;

[0014] A first selection module, configured to select a target translation service provider from the at least two translation service providers according to the comprehensive evaluation values of the at least two translation service providers;

[0015] A processing module, configured to translate the source language text through the target translation service provider.

[0016] In a third aspect, embodiments of the present disclosure provide an electronic device, where the electronic device includes a processor and a memory, and the memory stores a program or instruction that can run on the processor. When the program or instruction is executed by the processor, the steps of the method described in the first aspect are implemented.

[0017] In a fourth aspect, embodiments of the present disclosure provide a readable storage medium, where a program or instruction is stored on the readable storage medium. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0018] In a fifth aspect, embodiments of the present disclosure provide a chip, where the chip includes a processor and a communication interface, the communication interface is coupled to the processor, and the processor is configured to run a program or instruction to implement the method described in the first aspect.

[0019] Sixthly, an embodiment of the present disclosure provides a computer program product. The program product is stored in a storage medium and is executed by at least one processor to implement the method described in the first aspect.

[0020] In an embodiment of the present disclosure, obtain the results of target language translations of a source language text sample by at least two translation service providers to obtain at least two first translation results; according to the at least two first translation results, respectively determine the evaluation values of the at least two translation service providers in each dimension among multiple dimensions, where the multiple dimensions include at least two of the following: consistency, cultural adaptability, fidelity, comprehensibility, real-time performance, accuracy, price, the consistency is used to indicate the similarity of the translation results of the at least two translation service providers, the cultural adaptability is used to indicate the translation accuracy of the cultural feature vocabulary by the translation service provider, the fidelity is used to indicate the matching degree of the translation results of the translation service provider and the source language text, and the comprehensibility includes at least one of grammatical accuracy, readability, fluency, and ambiguity; according to the evaluation values of the at least two translation service providers in each dimension, determine the comprehensive evaluation values of the at least two translation service providers; according to the comprehensive evaluation values of the at least two translation service providers, select a target translation service provider from the at least two translation service providers; translate the source language text through the target translation service provider. In this way, by evaluating the translation services of each translation service provider from multiple dimensions and selecting the target translation service provider by synthesizing the multi-dimensional evaluation values, it is possible to ensure that a translation service provider with good translation service quality or high cost performance is selected to translate the current source language text, ensure that the translation quality meets the translation requirements of the current scenario, and have a high selection efficiency, and can adapt to the interpretation requirements of real-time usage scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a flowchart of a data processing method provided by an embodiment of the present disclosure;

[0022] Figure 2 is a structural diagram of a data processing device provided by an embodiment of the present disclosure;

[0023] Figure 3 is a structural diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0024] Next, the technical solutions in the embodiments of the present disclosure will be clearly described in conjunction with the accompanying drawings in the embodiments of the present disclosure. Obviously, the described embodiments are some, but not all, of the embodiments of the present disclosure. Based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art belong to the scope of protection of the present disclosure.

[0025] The terms "first", "second", etc. in the description and claims of the present disclosure are used to distinguish similar objects, rather than to describe a specific order or sequence. It should be understood that such terms can be interchanged under appropriate circumstances so that the embodiments of the present disclosure can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally of the same category, and the number of objects is not limited. For example, the first object can be one or multiple. In addition, "and / or" in the description and claims means at least one of the connected objects, and the character " / " generally indicates an "or" relationship between the associated objects before and after.

[0026] To make the embodiments of the present disclosure clearer, the following is an introduction to the relevant technical knowledge involved in the embodiments of the present disclosure:

[0027] The existing translation services have the following defects:

[0028] 1) The quality of simultaneous interpretation services is inconsistent

[0029] Electronic devices such as smart speakers, smart headphones, and smart watches can access translation services provided by different simultaneous interpretation suppliers. However, the translation quality of each supplier varies in different fields. For example, translation service provider A performs better in the e-commerce field, while translation service provider B is more accurate in the windows software field. This results in users being unable to continuously obtain a high-quality translation experience in diverse meeting scenarios.

[0030] 2) Lack of an intelligent supplier selection mechanism

[0031] Users need to manually switch simultaneous interpretation suppliers in the mobile APP. However, ordinary users can hardly understand the translation performance of each supplier in different fields in a timely manner, and the manual selection is inefficient and cannot meet the needs of real-time meetings.

[0032] 3) It is difficult to achieve a balance between translation quality and price

[0033] The charging standards of each supplier are different. When users pursue high-quality translation, they may face high costs. There is a lack of a mechanism that can comprehensively consider translation quality and price to help users select the service with the highest cost performance.

[0034] In response to the above technical problems, the present disclosure proposes a brand-new method for selecting a translation service provider, aiming to solve the problems in the prior art from the following aspects:

[0035] 1) Automatically identify the industry field of the speech content

[0036] Analyze the speech content in real time through speech recognition and natural language processing technologies, determine its industry, and provide a basis for selecting the most suitable simultaneous interpretation service provider.

[0037] 2) Establish a standardized sample testing mechanism

[0038] Construct and regularly update a test sample library containing typical expressions in various industries, and use these samples to evaluate the translation quality of each service provider to improve the accuracy and fairness of the evaluation.

[0039] 3) Compare the translation results of multiple service providers in real time

[0040] Send the speech samples to multiple service providers simultaneously to obtain the translation results. Through comparison and evaluation, objectively score based on the principle of the minority obeying the majority, and identify the service provider with the best translation quality.

[0041] 4) Comprehensively evaluate the translation quality from multiple dimensions

[0042] Comprehensively evaluate the translation results from multiple dimensions such as cultural adaptability, fidelity, comprehensibility, real-time performance, accuracy, and price to ensure the reliability of the selection result.

[0043] 5) Intelligently select the optimal service provider

[0044] Based on the comprehensive score and price factors, calculate the cost performance of each service provider, and automatically select the simultaneous interpretation service with the best quality or the highest cost performance for the user.

[0045] The embodiments of the present disclosure relate to the fields of artificial intelligence and natural language processing, and aim to solve the problem of inconsistent quality of existing simultaneous interpretation services in different fields. Collect the speech of the speaker through electronic devices such as smart speakers, smart headphones, and smart watches, use AI technology to automatically identify the industry to which the content belongs, and select the most suitable service from multiple simultaneous interpretation service providers. The key technologies include automatic industry identification, sample-based translation quality evaluation, multi-dimensional comprehensive scoring, and intelligent selection of service providers. This solution improves the accuracy of simultaneous interpretation and the user experience.

[0046] The following will combine the accompanying drawings and specifically illustrate the data processing method provided by the embodiments of the present disclosure through specific embodiments and their application scenarios.

[0047] Please refer to Figure 1 , Figure 1 which is the flowchart of the data processing method provided by the embodiments of the present disclosure. As Figure 1 shown, the method includes the following steps:

[0048] Step 101, obtain the results of translating the source language text sample into the target language by at least two translation service providers, and obtain at least two first translation results.

[0049] The above-mentioned at least two translation service providers may refer to the connected suppliers providing translation services. There are differences in the translation service quality provided by different translation service providers, with different focuses in different fields. For example, the translation accuracy of translation service provider A in the e-commerce field is better than that of translation service provider B, while the translation accuracy of translation service provider B in the field of communication technology is better than that of translation service provider A. In addition to the field advantages, there may also be differences in the translation quality of different translation service providers. Therefore, in order to select a translation service provider that suits the current usage scenario and has good translation service performance, the present disclosure proposes to evaluate different translation service providers from multiple dimensions and select the optimal translation service provider based on the comprehensive evaluation results of multiple dimensions.

[0050] In the embodiments of the present application, the source language text can be understood as the language text before translation. Specifically, a segment of speech input by a user can be collected through a microphone, and then using speech recognition technology, the collected speech signal can be converted into text. The above-mentioned target language is the language to be translated. Exemplarily, if the source language is Chinese and the target language is English, it is necessary to translate the current Chinese into English through a translation service provider.

[0051] The above-mentioned source language text sample can be understood as a segment or an example text excerpted from the source language text, serving as a sample for testing the translation performance of each translation service provider.

[0052] In this step, the source language text sample can be sent to the at least two translation service providers to obtain the results of the at least two translation service providers translating the source language text sample into the target language. Each translation result is the first translation result. It should be noted here that since it is only necessary to evaluate each translation service provider first in order to select a target translation service provider with better performance, in order to save translation and evaluation resources, a segment or an example text excerpted from the source language text can be used as the source language text sample and sent to the at least two translation service providers. When evaluating, the at least two translation service providers only need to translate the source language text sample. The present disclosure can also perform real-time translation on the source language text. Exemplarily, a segment of the source language text generated in real time can be intercepted as the source language text sample. After evaluating and screening out the target translation service provider, the translation service provided by the target translation service provider can be accessed to perform real-time translation on the source language text generated in real time. In addition, the present disclosure can also adjust the translation service provider in real time following the output of the source language text to be translated. For example, in some multi-party communication meetings, information such as the industry corresponding to the source language text can be recognized in real time, and based on the recognition results, each translation service provider can be re-evaluated in the background, and the currently optimal translation service provider can be recommended.

[0053] Step 102: Determine the evaluation values of each dimension among multiple dimensions for the at least two translation service providers according to the at least two first translation results. The multiple dimensions include at least two of the following: consistency, cultural adaptability, fidelity, comprehensibility, real-time performance, accuracy, and price. The consistency is used to indicate the similarity of the translation results of the at least two translation service providers. The cultural adaptability is used to indicate the translation accuracy of the cultural feature words by the translation service provider. The fidelity is used to indicate the matching degree between the translation result of the translation service provider and the source language text. The comprehensibility includes at least one of grammatical accuracy, readability, fluency, and ambiguity.

[0054] In the embodiments of the present disclosure, a multi-dimensional comprehensive evaluation can be performed on each translation service provider based on the results of the target language translation of the source language text sample by each translation service provider. For example, the translation quality of each translation service provider can be evaluated from dimensions such as the consistency of the translation results of the translation service provider, the accuracy of the translation of cultural feature words, that is, cultural adaptability, the matching degree between the translation result and the source language text, that is, fidelity, the readability, grammatical accuracy, fluency, ambiguity, real-time performance, accuracy, etc. of the translation result. The cost performance of each translation service provider can also be evaluated in combination with the price dimension.

[0055] Exemplarily, the similarity of the first translation results of each translation service provider can be compared to determine the consistency of the translation results of each translation service provider; check whether the cultural feature words in the source language culture are accurately translated in the first translation results of each translation service provider to obtain the cultural adaptability of each translation service provider; compare the first translation results of each translation service provider with the source language text sample to determine whether each keyword in the source language text sample is included and whether the semantics match to obtain the fidelity of each translation service provider; calculate the readability and fluency of the first translation results of each translation service provider, check for grammar errors, and detect whether there is ambiguity, so as to obtain the comprehensibility of each translation service provider; detect the response time of each translation service provider from receiving the request to returning the translation result to determine the real-time performance of each translation service provider; evaluate the accuracy of the first translation results of each translation service provider, check for translation errors, and determine the accuracy of each translation service provider; obtain the charging standard of each translation service provider to determine the price of each translation service provider.

[0056] Then, the evaluation results of each translation service provider in each dimension can be converted into evaluation values. For example, the higher the consistency, the higher the consistency evaluation value, and the evaluation value can be represented by indicators such as scores and percentages.

[0057] Optionally, when the multiple dimensions include consistency and the number of the at least two translation service providers is greater than 2, step 102 includes:

[0058] Performing clustering calculation or similarity calculation on the at least two first translation results, determining that the consistency evaluation value of the first translation service provider is a first value, and determining that the consistency evaluation value of the second translation service provider is a second value, where the first translation service provider is the translation service provider with consistent first translation results among the at least two translation service providers, the second translation service provider is the translation service provider other than the target translation service provider among the at least two translation service providers, and the first value is greater than the second value.

[0059] In some embodiments, when evaluating the translation quality of each translation service provider from the dimension of consistency, the first translation results of each translation service provider can be compared for consistency. Specifically, text clustering or similarity calculation can be used to find the target translation service provider that is consistent with most of the first translation results, and the score of the target translation service provider that is consistent with most of the first translation results, that is, the consistency evaluation value, can be higher, while the score of the translation service provider that is inconsistent with most of the first translation results can be lower. For example, the score of the target translation service provider that is consistent with most of the first translation results is 90 points, and the score of the translation service provider that is inconsistent with most of the first translation results is 70 points. Among them, being consistent with most of the first translation results can mean that the similarity with most of the first translation results is greater than a certain value, or belonging to the same cluster as most of the first translation results.

[0060] In this way, the translation accuracy of each translation service provider can be evaluated from the dimension of consistency, which helps to exclude translation service providers with poor translation results.

[0061] Optionally, when the multiple dimensions include cultural adaptability, step 102 includes:

[0062] Performing word segmentation and part-of-speech tagging on the source language text sample, and identifying cultural feature words in the source language text sample according to the results of word segmentation and part-of-speech tagging of the source language text sample;

[0063] Performing word segmentation and part-of-speech tagging on each of the at least two first translation results, and checking whether each of the first translation results contains target language cultural feature words corresponding to the cultural feature words in the source language text sample according to the results of word segmentation and part-of-speech tagging of each of the first translation results, to obtain a first check result;

[0064] Calculating the semantic similarity between the source language text sample and each of the first translation results in the dimension of cultural features;

[0065] Determine the cultural adaptability evaluation values of the translation service providers corresponding to the first translation results according to the first inspection results and semantic similarities of the first translation results.

[0066] In some embodiments, the cultural adaptability of the first translation results of each translation service provider can be evaluated. Specifically, a cultural feature word library can be pre-constructed, including:

[0067] Collect cultural feature vocabulary: Collect cultural feature vocabulary and phrases including idioms, common sayings, slang, allusions, etc., and establish a corresponding word library between the source language (such as Chinese) and the target language (such as English).

[0068] Corpus establishment: Use a large parallel corpus to extract sentence pairs containing cultural features to form a cultural corpus.

[0069] Then, cultural elements in the source language text sample can be identified, including:

[0070] Text analysis: Segment and label the parts of speech of the source language text sample to identify the cultural feature vocabulary in it.

[0071] Mark cultural elements: Mark the identified cultural feature vocabulary in the source language text sample for subsequent processing.

[0072] Then, cultural matching is performed on each first translation result, including:

[0073] Preprocessing of translation results: Segment and label the parts of speech of the first translation results returned by each translation service provider.

[0074] Cultural element matching: Check whether each first translation result contains the target language cultural expressions corresponding to the cultural feature vocabulary marked in the source language text sample. If the first translation result can accurately use the equivalent cultural expressions in the target language, a higher cultural adaptability score is given to its translation service provider.

[0075] It is also possible to calculate the semantic similarity between each first translation result and the source language text sample in the cultural feature dimension:

[0076] Semantic representation: Use a pre-trained language model, such as BERT, RoBERTa, etc., to encode the source language text sample and each first translation result into vector representations.

[0077] Similarity calculation: Calculate the semantic similarity between the source language text sample and the translation result in the cultural feature dimension, and specifically, measurement methods such as cosine similarity can be used.

[0078] Finally, based on the cultural element matching results and similarity calculation results, score each first translation result. The scoring rules can be as follows: set scoring criteria. For example, if the translation result contains the correct cultural corresponding expressions and has a high semantic similarity, it gets 100 points; if it is a literal translation but has a reasonable explanation, it gets 70 points; if there are translation errors or cultural elements are omitted, it gets 50 points or lower. When integrating into the comprehensive evaluation value calculation, the cultural adaptability score can also be weighted according to the importance of cultural elements in the sentence. For example, if the importance is high, a higher weight is set for the cultural adaptability score.

[0079] Through this implementation method, it is possible to evaluate the translation quality of each translation service provider in terms of cultural adaptability, and ensure that a translation service provider who can accurately translate cultural feature words can be selected subsequently.

[0080] Optionally, when the multiple dimensions include fidelity, step 102 includes:

[0081] Extract key sentences and phrases from the source language text sample, where the key sentences and phrases include at least one of key entities, keywords, and key phrases;

[0082] Check whether each first translation result in the at least two first translation results correctly translates the key sentences and phrases, and obtain a second check result;

[0083] Translate each first translation result back to the source language, and compare the source language translation results of each first translation result with the source language text sample for consistency to obtain a consistency comparison result;

[0084] Determine the fidelity evaluation values of the translation service providers corresponding to each first translation result according to the second check result and the consistency comparison result of each first translation result.

[0085] In some embodiments, the first translation results of each translation service provider can be evaluated for fidelity. Specifically, when implementing, key content extraction can be performed on the source language text sample, including:

[0086] Important information recognition: Use Named Entity Recognition (NER) technology to extract key entities such as person names, place names, organization names, and proper nouns from the source language text sample.

[0087] Keyword extraction: Adopt algorithms such as Term Frequency - Inverse Document Frequency (TF-IDF) and TextRank to extract keywords and key phrases from the source language text sample.

[0088] Then, compare each first translation result, including:

[0089] Entity and keyword matching: Check whether the key entities, keywords, and key phrases are correctly translated and retained in each first translation result.

[0090] Translation integrity check: Ensure that important information is not omitted and non-existent information is not added in the translation result.

[0091] Consistency detection can also be performed on each first translation result, including:

[0092] Semantic matching: Use machine translation quality evaluation models, such as Bilingual Evaluation Understudy (BLEU), Translation Error Rate (TER), Metric for Evaluation of Translation with Explicit Ordering (METEOR), etc., to evaluate the consistency between the source language text sample and the first translation result.

[0093] Bidirectional translation verification: Translate the first translation result back to the source language and compare it with the source language text sample to check for consistency.

[0094] Finally, score each first translation result according to the translation integrity check result and the consistency comparison result. The scoring rules can include two aspects: matching score, that is, calculate the fidelity score according to the matching degree and consistency of the key content; error penalty: for important information that is mistranslated or omitted, such as mistranslating or omitting key entities, keywords, key phrases and other important information, give corresponding deductions.

[0095] Through this implementation method, the fidelity of the translations provided by each translation service provider can be evaluated, ensuring that a translation service provider with a higher translation fidelity can be selected subsequently.

[0096] Optionally, when the multiple dimensions include comprehensibility, step 102 includes:

[0097] Use the grammar analyzer of the target language to perform grammar analysis on each first translation result among the at least two first translation results, and identify and record the number of grammar errors in each first translation result;

[0098] Adopt a readability algorithm to calculate the readability score of each first translation result;

[0099] Calculate the fluency of each first translation result;

[0100] Perform ambiguity detection on each of the first translation results to obtain an ambiguity detection result. The ambiguity detection includes at least one of the following: analyzing whether there is ambiguity in each of the first translation results, and checking whether the reference of pronouns in each of the first translation results is clear;

[0101] Determine the comprehensibility evaluation values of the respective translation service providers corresponding to each of the first translation results based on the number of grammar errors, the readability scores, fluency, and ambiguity detection results in each of the first translation results.

[0102] In some embodiments, the comprehensibility of the first translation results of each translation service provider can be evaluated. Specifically, it can include the following aspects:

[0103] 1) Grammar checking:

[0104] The grammar analyzer analyzes grammar errors: Use a grammar analyzer of the target language, such as SyntaxNet, spaCy, etc., to perform syntactic analysis on each of the first translation results to identify grammar errors;

[0105] Error type statistics: Record the types and quantities of grammar errors in each of the first translation results.

[0106] 2) Readability measure:

[0107] Readability formula scoring: Adopt readability formulas such as the Flesch-Kincaid formula and the Gunning Fog Index to calculate the readability scores of each of the first translation results;

[0108] Sentence length check: Statistically analyze the average length of sentences. Sentences that are too long or too short may affect the degree of understanding.

[0109] 3) Fluency evaluation:

[0110] Language model scoring: Use a pre-trained language model, such as Generative Pre-trained Transformer (GPT), BERT, etc., to estimate the fluency of each of the first translation results. A higher fluency indicates a better fit to the target language expression habits.

[0111] Perplexity calculation: Calculate the perplexity of each of the first translation results. The lower the value, the higher the language fluency.

[0112] 4) Ambiguity detection:

[0113] Context consistency: Analyze whether there is ambiguity or self-contradiction in each of the first translation results.

[0114] Anaphora resolution: Check whether the reference of pronouns in each first translation result is clear and correct.

[0115] Finally, based on the above detection results, the comprehensibility of each first translation result can be scored. For example, error deductions can be made according to the severity and quantity of grammar errors, ambiguity, unclear reference, etc., and points can also be added for fluency. Higher scores can be given to translation results with high fluency and good readability.

[0116] Through this implementation manner, it is possible to evaluate from the dimension of the comprehensibility of the translation results provided by each translation service provider, ensuring that a translation service provider with higher translation comprehensibility can be selected subsequently.

[0117] Optionally, when the plurality of dimensions include real-time performance, step 102 includes:

[0118] Determine the response time of each translation service provider among the at least two translation service providers for translating the source language text sample into the target language;

[0119] Based on the response time corresponding to each translation service provider, determine the real-time performance evaluation value of each translation service provider.

[0120] In some embodiments, the real-time performance of the first translation results of each translation service provider can be evaluated. Specifically, when implementing, the response time of each translation service provider can be recorded first. For example, when sending a translation request, record the timestamp T1 of the request sending, and when receiving the translation result, record the received timestamp T2, so that the delay can be calculated: response time RT = T2 - T1.

[0121] Network delay measurement can also be performed. For example, a Ping test is carried out to perform a Ping operation on the servers of each translation service provider to measure the network delay NL, and then delay correction is performed, that is, the actual translation processing time PT = RT - NL.

[0122] In this way, the response time of each translation service provider can be scored. Specifically, according to the actual translation processing time of each translation service provider, a scoring standard can be set. For example:

[0123] PT ≤ 500ms, get 100 points;

[0124] 500ms < PT ≤ 1000ms, get 80 points;

[0125] PT > 1000ms, get 60 points or lower.

[0126] In some embodiments, stability monitoring can also be performed on each translation service provider. For example, the response time of each translation service provider can be measured multiple times: record the response time within a period of time (e.g., multiple consecutive requests), calculate the average value and standard deviation; anomaly detection: identify and record the situations where each translation service provider times out or fails to respond. The fewer the abnormal situations occur, the better the stability. On the contrary, the more the abnormal situations occur, the worse the stability.

[0127] When performing real-time scoring, stability weighting can also be combined. For example, stability weighting is performed on the scoring of the response time. The higher the stability, the higher the score.

[0128] Through this implementation method, it is possible to evaluate from the real-time dimension of the translation results of each translation service provider, ensuring that a translation service provider with higher translation real-time performance is selected subsequently.

[0129] Optionally, when the multiple dimensions include accuracy, step 102 includes:

[0130] Evaluate each first translation result in the source language text sample and the at least two first translation results using a preset evaluation index, and calculate the accuracy score of each first translation result;

[0131] Identify and count the number of incorrect translations in each first translation result, where the incorrect translation includes at least one of mistranslation, omitting translation, and adding translation;

[0132] Determine the accuracy evaluation value of each translation service provider corresponding to each first translation result according to the accuracy score and the number of incorrect translations of each first translation result.

[0133] In some embodiments, the accuracy of the first translation results of each translation service provider can be evaluated. Specifically, the evaluation can be carried out from the following aspects:

[0134] 1) Comprehensive evaluation index:

[0135] Automatic evaluation index: Use automatic evaluation indexes such as BLEU, METEOR, and TER to evaluate the source language text sample and the translation result, and calculate the accuracy score. The automatic evaluation index is also the above-mentioned preset evaluation index.

[0136] Semantic similarity calculation: Use advanced semantic matching models such as Sentence-BERT and Universal Sentence Encoder (USE) to calculate the sentence-level semantic similarity.

[0137] 2) Mistranslation and omission detection:

[0138] Error type analysis: Identify types of errors such as mistranslation, omission, and addition in each first translation result.

[0139] Error statistics: Count the number and types of errors, and quantify the overall accuracy. For example, deduct points from the accuracy score based on the number of errors.

[0140] 3) Machine learning evaluation model:

[0141] Training quality evaluation model: Use the labeled translation quality data, such as including the samples to be translated and the corresponding standard translation results, extract various features including vocabulary, syntax, semantics, etc. as the input features of the model, train machine learning models such as random forest, neural network, etc., and use the trained quality evaluation model to predict the accuracy of each first translation result.

[0142] Finally, the comprehensive accuracy scores of each translation service provider can be calculated by combining the accuracy scores of the above automatic evaluation metrics, the sentence-level semantic similarity, the predicted accuracy, and the mistranslation and omission situations. For example, perform weighted averaging on each score to obtain the comprehensive accuracy score. And the scoring grades can be set according to the score range. For example:

[0143] Score ≥ 90, high accuracy, get 100 points;

[0144] 80 ≤ score < 90, good accuracy, get 80 points;

[0145] Score < 80, general or low accuracy, get 60 points or lower.

[0146] Through this implementation method, the accuracy of the translation results of each translation service provider can be evaluated, ensuring that a translation service provider with higher translation accuracy can be selected subsequently.

[0147] Optionally, when the multiple dimensions include price, step 102 includes:

[0148] Obtain the translation service price information of each translation service provider among the at least two translation service providers;

[0149] Determine the price evaluation value of each translation service provider according to the translation service price information of each translation service provider.

[0150] In some embodiments, considering the user's sensitivity to price, the prices of the first translation results of each translation service provider can be evaluated. Specifically, the price information of each translation service provider can be obtained. For example, by calling the Application Programming Interface (API) of each translation service provider, real-time translation service price information can be obtained, such as the charging standard per character / word; a price list of each translation service provider can also be established and updated regularly.

[0151] Then, calculate the translation costs of each translation service provider and calculate the cost of the current translation content. Specifically, according to the length (number of characters / words) of the translation text and the charging standards of each translation service provider, calculate the cost C required for each translation service provider to complete the current translation. The price C can also be normalized to facilitate comprehensive comparison with the scores of other evaluation dimensions.

[0152] Finally, conduct a price score for each translation service provider. For example, set a scoring standard according to the cost C. The lower the cost, the higher the score. For example:

[0153] If C ≤ 0.01 yuan / character, the score is 100 points;

[0154] If 0.01 yuan / character < C ≤ 0.02 yuan / character, the score is 80 points;

[0155] If C > 0.02 yuan / character, the score is 60 points or lower.

[0156] In some embodiments, the weight of the price score in the comprehensive evaluation value can also be adjusted according to the user's sensitivity to price.

[0157] Through this implementation method, it is possible to recommend translation service providers with relatively high cost performance to users by combining the prices of each translation service provider.

[0158] Step 103: Determine the comprehensive evaluation values of the at least two translation service providers according to the evaluation values of the at least two translation service providers in each dimension.

[0159] In this step, for each translation service provider, the evaluation values in each dimension can be comprehensively considered. For example, according to the importance of each dimension, weights are assigned to each dimension, and then the evaluation values of the translation service provider in each dimension are weighted and calculated to obtain the comprehensive evaluation value of the translation service provider.

[0160] For example, according to the importance of different evaluation dimensions, set their respective weights as follows:

[0161] Consistency: W1 = 0.1

[0162] Cultural adaptability: W2 = 0.1

[0163] Fidelity: W3 = 0.3

[0164] Understandability: W4 = 0.2

[0165] Real - time performance: W5 = 0.1

[0166] Accuracy: W6 = 0.1

[0167] Price: W7 = 0.1

[0168] Comprehensive score calculation: For each translation service provider, calculate the comprehensive score S:

[0169] S = W1 * Consistency score + W2 * Cultural adaptability score + W3 * Fidelity score + W4 * Understandability score + W5 * Real - time performance score + W6 * Accuracy score + W7 * Price score.

[0170] Optionally, the method further includes:

[0171] Identify the industry category to which the source - language text belongs;

[0172] Select test sentences of the industry category from the sample library, where the test sentences include sample sentences and expected translation results of the sample sentences in the target language;

[0173] Obtain the results of translating the sample sentences into the target language by the at least two translation service providers to get at least two second - translation results;

[0174] Determine the first evaluation values of the at least two translation service providers according to the similarity between each second - translation result in the at least two second - translation results and the expected translation result;

[0175] The determining the comprehensive evaluation values of the at least two translation service providers according to the evaluation values of the at least two translation service providers in each dimension includes:

[0176] Determine the comprehensive evaluation values of the at least two translation service providers according to the evaluation values of the at least two translation service providers in each dimension and the first evaluation values.

[0177] That is, in some embodiments, industry - translation tests can also be performed on each translation service provider to determine the translation accuracy of each translation service provider for texts in the industry category to which the current source - language text belongs.

[0178] Specifically, the industry category to which the source language text belongs can be identified first. For example, by performing word segmentation and keyword extraction on the source language text, and using a pre-trained industry classification model, such as a deep learning model, to identify the industry to which the content of the source language text belongs. The model outputs the identification result, that is, the industry label, for use by subsequent modules. Of course, in addition to pre-trained deep learning models, methods such as rule matching and Naive Bayes can also be used for industry identification.

[0179] Then, test sentences of the industry category are selected from a pre-established sample library, such as including sample sentences and the expected translation results of the sample sentences in the target language. Among them, the sample library can be a test sample library established by collecting typical expressions and common sentences in various industries and can be updated regularly.

[0180] Exemplarily, the test sentences can be as follows:

[0181] Sample sentence: Who knows what he's up to?

[0182] Expected translation: Who knows what he's up to?

[0183] Unexpected translation: Who knows what medicine he's selling in his gourd?

[0184] After selecting the test sentences, the sample sentences in the test sentences can be sent to each translation service provider to obtain the translation results of the test sentences in the target language by each translation service provider. Each translation result is the second translation result.

[0185] For the convenience of subsequent comparison, the second translation results can also be pre-processed, such as unifying the format of the translation results, removing special characters, and performing standardization processing.

[0186] Then, semantic comparison can be performed on each second translation result. For example, using semantic analysis technology, such as the BERT model, to calculate the semantic similarity between the actual translation and the expected translation respectively, that is, to calculate the semantic similarity between each second translation result and the expected translation result of the sample sentence respectively.

[0187] Finally, according to the similarity calculation results, corresponding scores can be given to each second translation result, that is, to score each translation service provider for the industry test. It can be that the higher the similarity, that is, the smaller the difference, the higher the score.

[0188] In this way, when calculating the comprehensive evaluation value of each translation service provider, the comprehensive evaluation value of each translation service provider can be determined by integrating the evaluation values and industry test scores of each translation service provider in the foregoing various dimensions.

[0189] Through this implementation, since the evaluation values of each translation service provider in multiple dimensions and the industry test scores are considered to evaluate the translation performance of each translation service provider, the accuracy and reliability of the evaluation can be guaranteed, which is conducive to accurately selecting the best target translation service provider subsequently.

[0190] Step 104: Select a target translation service provider from the at least two translation service providers according to the comprehensive evaluation values of the at least two translation service providers.

[0191] In this step, the comprehensive evaluation values of each translation service provider can be referred to for selecting the target translation service provider. For example, the translation service provider with the highest comprehensive evaluation value can be selected as the target translation service provider, or the top several translation service providers with relatively high comprehensive evaluation values can be output for the user to confirm and select, so as to perform subsequent translation through the finally selected target translation service provider.

[0192] Exemplarily, the translation service providers can be sorted according to the comprehensive score S of each translation service provider. In addition, the comprehensive evaluation values of each translation service provider can be regularly detected. When the detected comprehensive evaluation value decreases, the optimal translation service provider can be re-evaluated and selected.

[0193] Optionally, when the multiple dimensions do not include price, step 104 includes:

[0194] When the user selects quality priority, select the translation service provider with the highest comprehensive evaluation value from the at least two translation service providers as the target translation service provider;

[0195] And / or,

[0196] When the user selects cost performance priority, obtain the translation service price information of each translation service provider among the at least two translation service providers;

[0197] Calculate the cost performance of each translation service provider according to the ratio of the comprehensive evaluation value of each translation service provider to the translation service price information;

[0198] Select the translation service provider with the highest cost performance from the at least two translation service providers as the target translation service provider.

[0199] That is, in some embodiments, two selection strategies can be provided for users to choose from, such as including a quality - first mode and a cost - performance - first mode. For the quality - first mode, the translation service provider with the highest comprehensive score can be selected; for the cost - performance - first mode, if the user pays more attention to cost, a higher weight can be given to the price score to recalculate the comprehensive score, or the cost - performance of each translation service provider can be calculated by combining the comprehensive score of translation quality and price. For example, the ratio of the comprehensive evaluation value to the price can be used as the cost - performance index, that is, cost - performance = comprehensive evaluation value / service price, and finally the translation service provider with the highest cost - performance is selected.

[0200] Through this implementation method, it is possible to recommend and select a more suitable translation service provider for the user according to the translation quality or price that the user emphasizes, meeting the needs of different users.

[0201] It should be noted that the system can continuously monitor the translation quality and adjust the selection of the translation service provider in real - time according to the changes in the source - language content.

[0202] Step 105: Translate the source - language text through the target translation service provider.

[0203] After selecting the target translation service provider, the source - language text can be translated through the target translation service provider, that is, the source - language text is translated into the target - language text. It should be noted that the target translation service provider can be used for translation in this translation service, but the translation service provider is re - evaluated and selected next time, or the target translation service provider can be used for a period of time in the future, but the comprehensive evaluation values of each translation service provider are continuously monitored, and when a change is detected, the user is reminded whether to switch the translation service provider.

[0204] Optionally, step 103 includes:

[0205] According to the weights assigned to each dimension in advance, the evaluation values of the at least two translation service providers in each dimension are weighted and calculated to obtain a comprehensive evaluation value;

[0206] The method further includes:

[0207] Obtain the feedback information of the user on the translation results translated by each translation service provider among the at least two translation service providers;

[0208] According to the feedback information, adjust the weights of each dimension.

[0209] In some embodiments, user feedback can also be collected to adjust the weights of each dimension according to the user feedback. For example, a feedback entry can be provided on the user side to allow users to rate or provide opinions on the translation results of the currently selected translation service provider; and the data of the user feedback can be associated and stored with the corresponding translation results and translation service provider information.

[0210] The feedback data can be statistically analyzed to identify common problems and the performance of the translation service providers; the parameters of the aforementioned evaluation model and the weights of each dimension can also be adjusted using the feedback data to improve the accuracy of the evaluation.

[0211] In some embodiments, the long-term translation quality of each translation service provider can also be monitored to detect trends of quality fluctuations or declines; a warning mechanism can also be adopted. When it is found that the translation quality of a certain translation service provider continues to decline, a warning is issued, and its priority is reduced in subsequent selection of translation service providers.

[0212] Through this implementation method, the weights of each evaluation dimension can be adjusted based on user feedback to ensure that the subsequent selection of translation service providers better meets the expectations of users.

[0213] It should be noted that for the semantic comparison involved in the foregoing implementation method, more advanced language models such as GPT can also be used to improve the accuracy of semantic comparison. The evaluation dimensions can also be increased or decreased according to requirements, such as adding factors such as user feedback and the stability of translation service providers. The application scenarios of the solution of the present disclosure can also be expanded, including the following:

[0214] Multi-language translation scenario: Apply the solution to the translation of more languages to achieve a global real-time translation service.

[0215] Education and training: Apply it to language learning to help students compare different translation results and improve learning effects.

[0216] Cross-border e-commerce: In cross-border e-commerce platforms, provide high-quality real-time translation services for merchants and customers.

[0217] Through the above multi-dimensional comprehensive evaluation method for translation services, the system can deeply and comprehensively evaluate the translation quality of each simultaneous interpretation provider. With the help of advanced natural language processing technologies and machine learning algorithms, a detailed evaluation is carried out from aspects such as consistency, cultural adaptability, fidelity, comprehensibility, real-time performance, accuracy, and price, to help users select the most suitable translation service provider in different application scenarios and improve the user experience.

[0218] It should also be noted that in some embodiments, model optimization can also be carried out. For example, the model can be updated regularly. As languages evolve and new information emerges, the language model and evaluation model are updated regularly to maintain the accuracy of evaluation. It is also possible to expand to support more language pairs to achieve multilingual translation quality evaluation.

[0219] In some embodiments, the performance can also be optimized. For example, multi-threading or distributed computing can be used to speed up the evaluation process and reduce the impact on system performance. For frequently used translation results and evaluation data, a caching mechanism is introduced to reduce repeated calculations.

[0220] In some embodiments, the translation service can also be guaranteed from the perspectives of security and privacy. For example, during the transmission process, the user's voice and text data are encrypted to ensure data security. Relevant privacy regulations are complied with, and the user's sensitive information is not stored or leaked.

[0221] According to the above introduction, the embodiments of the present disclosure can be applied to simultaneous interpretation quality evaluation systems of electronic devices such as smart speakers, smart earphones, and smart watches. The system architecture includes the following modules:

[0222] Voice acquisition module: The voice signal of the speaker is collected in real time through the microphone of the intelligent electronic device.

[0223] Content analysis module: The voice signal is recognized and analyzed to extract the text content, and the natural language processing technology is used to identify the industry field to which the speech content belongs.

[0224] Sample test module: Select test samples corresponding to the industry from the sample library, send them to each translation service provider, and obtain the translation results for comparison and scoring.

[0225] Translation service provider translation module: Access the service interfaces of multiple translation service providers, send real-time voice samples, and obtain their respective translation results.

[0226] Translation quality evaluation module: Evaluate and score the translation results returned by the translation service provider from multiple dimensions.

[0227] Supplier selection module: Based on the evaluation results and price information, calculate the cost performance and intelligently select the optimal translation service provider.

[0228] User interaction module: Feed back the translation results to the user and provide setting and feedback interfaces.

[0229] In the embodiments of the present disclosure, the quality of simultaneous interpretation is automatically evaluated through AI technology, and the optimal provider is intelligently selected, significantly improving the translation accuracy and real-time performance. At the same time, considering the price factor, the service with the highest cost performance is provided for users. This solution can be widely applied to fields such as smart speakers and multimedia conference systems, and has important commercial value and social significance.

[0230] In the data processing method of the embodiments of the present disclosure, the results of translating a source language text sample into a target language by at least two translation service providers are obtained to obtain at least two first translation results; according to the at least two first translation results, the evaluation values of each dimension among the at least two translation service providers in multiple dimensions are respectively determined, where the multiple dimensions include at least two of the following: consistency, cultural adaptability, fidelity, comprehensibility, real-time performance, accuracy, price, the consistency is used to indicate the similarity of the translation results of the at least two translation service providers, the cultural adaptability is used to indicate the translation accuracy of the translation service provider for culturally characteristic vocabulary, the fidelity is used to indicate the matching degree of the translation results of the translation service provider with the source language text, and the comprehensibility includes at least one of grammatical accuracy, readability, fluency, and ambiguity; according to the evaluation values of the at least two translation service providers in each dimension, the comprehensive evaluation values of the at least two translation service providers are determined; according to the comprehensive evaluation values of the at least two translation service providers, a target translation service provider is selected from the at least two translation service providers; the source language text is translated by the target translation service provider. In this way, by evaluating the translation services of each translation service provider from multiple dimensions and comprehensively considering the multi-dimensional evaluation values to select a target translation service provider, it is possible to ensure that a translation service provider with good translation service quality or high cost performance is selected to translate the current source language text, ensure that the translation quality meets the translation requirements of the current scenario, and have a high selection efficiency, and can adapt to the interpretation requirements of real-time usage scenarios.

[0231] In the data processing method provided by the embodiments of the present disclosure, the execution subject may be a data processing device. In the embodiments of the present disclosure, taking the data processing device executing the data processing method as an example, the data processing device provided by the embodiments of the present disclosure is described.

[0232] Please refer to Figure 2 , Figure 2 which is the structural schematic diagram of the data processing device provided by the embodiments of the present disclosure. As Figure 2 shown, the data processing device 200 includes:

[0233] A first acquisition module 201, configured to obtain the results of translating a source language text sample into a target language by at least two translation service providers to obtain at least two first translation results;

[0234] A first determination module 202, configured to respectively determine evaluation values of each dimension in multiple dimensions for the at least two translation service providers according to the at least two first translation results, where the multiple dimensions include at least two of the following: consistency, cultural adaptability, fidelity, comprehensibility, real-time performance, accuracy, price, the consistency is used to indicate the similarity of the translation results of the at least two translation service providers, the cultural adaptability is used to indicate the translation accuracy of the translation service provider for culturally characteristic words, the fidelity is used to indicate the matching degree between the translation results of the translation service provider and the source language text, and the comprehensibility includes at least one of grammatical accuracy, readability, fluency, and ambiguity;

[0235] A second determination module 203, configured to determine comprehensive evaluation values of the at least two translation service providers according to the evaluation values of the at least two translation service providers in each dimension;

[0236] A first selection module 204, configured to select a target translation service provider from the at least two translation service providers according to the comprehensive evaluation values of the at least two translation service providers;

[0237] A processing module 205, configured to translate the source language text through the target translation service provider.

[0238] Optionally, the data processing device 200 further includes:

[0239] An identification module, configured to identify the industry category to which the source language text belongs;

[0240] A second selection module, configured to select test sentences of the industry category from a sample library, where the test sentences include sample sentences and expected translation results of the sample sentences in the target language;

[0241] A second acquisition module, configured to obtain results of translating the sample sentences into the target language by the at least two translation service providers, to obtain at least two second translation results;

[0242] A third determination module, configured to determine first evaluation values of the at least two translation service providers according to the similarity between each second translation result in the at least two second translation results and the expected translation result;

[0243] The second determination module 203 is configured to determine comprehensive evaluation values of the at least two translation service providers according to the evaluation values of the at least two translation service providers in each dimension and the first evaluation values.

[0244] Optionally, the first determination module 202 is configured to, when the multiple dimensions include consistency and the number of the at least two translation service providers is greater than 2, perform clustering calculation or similarity calculation on the at least two first translation results, determine that the consistency evaluation value of the first translation service provider is a first value, and determine that the consistency evaluation value of the second translation service provider is a second value, where the first translation service provider is the translation service provider with consistent first translation results among the at least two translation service providers, the second translation service provider is the translation service provider other than the target translation service provider among the at least two translation service providers, and the first value is greater than the second value.

[0245] Optionally, the first determination module 202 includes:

[0246] A first recognition unit, configured to, when the multiple dimensions include cultural adaptability, perform word segmentation and part-of-speech tagging on the source language text sample, and identify cultural feature words in the source language text sample according to the word segmentation and part-of-speech tagging results of the source language text sample;

[0247] A first inspection unit, configured to perform word segmentation and part-of-speech tagging on each of the at least two first translation results, and check whether each of the at least two first translation results contains target language cultural feature words corresponding to the cultural feature words in the source language text sample according to the word segmentation and part-of-speech tagging results of each of the at least two first translation results, to obtain a first inspection result;

[0248] A first calculation unit, configured to calculate the semantic similarity between the source language text sample and each of the at least two first translation results in the cultural feature dimension;

[0249] A first determination unit, configured to determine the cultural adaptability evaluation value of each translation service provider corresponding to each of the at least two first translation results according to the first inspection result and semantic similarity of each of the at least two first translation results.

[0250] Optionally, the first determination module 202 includes:

[0251] An extraction unit, configured to, when the multiple dimensions include fidelity, extract key sentences and phrases from the source language text sample, where the key sentences and phrases include at least one of key entities, keywords, and key phrases;

[0252] A second inspection unit, configured to check whether each of the at least two first translation results correctly translates the key sentences and phrases, to obtain a second inspection result;

[0253] A comparison unit for translating each of the first translation results back into the source language, and comparing the source language translation results of the first translation results with the source language text sample to obtain a consistency comparison result;

[0254] A second determination unit for determining a fidelity evaluation value of each translation service provider corresponding to each of the first translation results according to the second inspection results and the consistency comparison results of the first translation results.

[0255] Optionally, the first determination module 202 includes:

[0256] A second identification unit for, when the plurality of dimensions include understandability, performing syntactic analysis on each of the at least two first translation results using a syntactic analyzer of the target language, and identifying and recording the number of syntactic errors in each of the first translation results;

[0257] A second calculation unit for calculating a readability score of each of the first translation results using a readability algorithm;

[0258] A third calculation unit for calculating the fluency of each of the first translation results;

[0259] A detection unit for performing ambiguity detection on each of the first translation results to obtain an ambiguity detection result, where the ambiguity detection includes at least one of the following: analyzing whether there is ambiguity in each of the first translation results, and checking whether the reference of pronouns in each of the first translation results is clear;

[0260] A third determination unit for respectively determining an understandability evaluation value of each translation service provider corresponding to each of the first translation results according to the number of syntactic errors in each of the first translation results, the readability score of each of the first translation results, the fluency, and the ambiguity detection result.

[0261] Optionally, the first determination module 202 includes:

[0262] A fourth determination unit for, when the plurality of dimensions include real-time performance, determining the response time of each translation service provider among the at least two translation service providers for translating the source language text sample into the target language;

[0263] A fifth determination unit for determining a real-time performance evaluation value of each translation service provider according to the response time corresponding to each translation service provider.

[0264] Optionally, the first determination module 202 includes:

[0265] An evaluation unit, configured to, when the plurality of dimensions include accuracy, evaluate each of the first translation results between the source language text sample and the at least two first translation results by using a preset evaluation index, and calculate the accuracy scores of each of the first translation results;

[0266] A third identification unit, configured to identify and count the number of incorrect translations in each of the first translation results, where the incorrect translations include at least one of mistranslation, omission, and addition;

[0267] A sixth determination unit, configured to determine the accuracy evaluation values of the respective translation service providers corresponding to each of the first translation results according to the accuracy scores and the number of incorrect translations of each of the first translation results.

[0268] Optionally, the first determination module 202 includes:

[0269] A first acquisition unit, configured to, when the plurality of dimensions include price, acquire the translation service price information of each of the at least two translation service providers;

[0270] A seventh determination unit, configured to determine the price evaluation values of the respective translation service providers according to the translation service price information of each of the translation service providers.

[0271] Optionally, the second determination module 203 is configured to perform a weighted calculation on the evaluation values of the at least two translation service providers in each of the dimensions according to the weights assigned to each of the dimensions in advance, to obtain a comprehensive evaluation value;

[0272] The data processing device 200 further includes:

[0273] A third acquisition module, configured to acquire the feedback information of the user on the translation results translated by each of the at least two translation service providers;

[0274] An adjustment module, configured to adjust the weights of each of the dimensions according to the feedback information.

[0275] Optionally, when the plurality of dimensions do not include price, the first selection module 204 includes:

[0276] A first selection unit, configured to, when the user selects quality priority, select the translation service provider with the highest comprehensive evaluation value from the at least two translation service providers as the target translation service provider;

[0277] And / or,

[0278] A second acquisition unit, configured to, when the user selects cost performance priority, acquire the translation service price information of each of the at least two translation service providers;

[0279] A fourth calculation unit, configured to calculate the performance-price ratio of each translation service provider according to the ratio of the comprehensive evaluation value of each translation service provider to the translation service price information.

[0280] A second selection unit, configured to select the translation service provider with the highest performance-price ratio from the at least two translation service providers as the target translation service provider.

[0281] The data processing device in the embodiment of the present disclosure obtains the results of translating a source language text sample into a target language by at least two translation service providers, and obtains at least two first translation results; according to the at least two first translation results, respectively determine the evaluation values of the at least two translation service providers in each dimension among multiple dimensions, where the multiple dimensions include at least two of the following: consistency, cultural adaptability, fidelity, comprehensibility, real-time performance, accuracy, price, the consistency is used to indicate the similarity of the translation results of the at least two translation service providers, the cultural adaptability is used to indicate the translation accuracy of the translation service provider for culturally characteristic words, the fidelity is used to indicate the matching degree of the translation results of the translation service provider and the source language text, and the comprehensibility includes at least one of grammatical accuracy, readability, fluency, and ambiguity; determine the comprehensive evaluation values of the at least two translation service providers according to the evaluation values of the at least two translation service providers in each dimension; select a target translation service provider from the at least two translation service providers according to the comprehensive evaluation values of the at least two translation service providers; translate the source language text through the target translation service provider. In this way, by evaluating the translation services of each translation service provider from multiple dimensions and selecting the target translation service provider by synthesizing the multi-dimensional evaluation values, it is possible to ensure that a translation service provider with good translation service quality or high performance-price ratio is selected to translate the current source language text, ensure that the translation quality meets the translation requirements of the current scenario, and have a high selection efficiency, and can adapt to the simultaneous interpretation requirements of real-time usage scenarios.

[0282] The data processing device in the embodiments of the present disclosure may be an electronic device or a component in an electronic device, such as an integrated circuit or a chip. The electronic device may be a terminal or other devices other than the terminal. Exemplarily, the electronic device may be a mobile phone, a tablet computer, a laptop computer, a handheld computer, a vehicle-mounted electronic device, a Mobile Internet Device (MID), an augmented reality (AR) / virtual reality (VR) device, a robot, a wearable device, an ultra-mobile personal computer (UMPC), a netbook, or a personal digital assistant (PDA), etc., or may also be a server, a Network Attached Storage (NAS), a personal computer (PC), a television (TV), a teller machine, or a self-service machine, etc. The embodiments of the present disclosure do not make specific limitations.

[0283] The data processing device in the embodiments of the present disclosure may be a device with an operating system. The operating system may be the Android operating system, the iOS operating system, or other possible operating systems. The embodiments of the present disclosure do not make specific limitations.

[0284] The data processing device provided in the embodiments of the present disclosure can implement Figure 1 each process implemented by the method embodiments, and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0285] Optionally, as Figure 3 shown, the embodiments of the present disclosure further provide an electronic device 300, including a processor 301 and a memory 302. A program or instruction that can run on the processor 301 is stored on the memory 302. When the program or instruction is executed by the processor 301, it implements each step of the above data processing method embodiment, and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0286] It should be noted that the electronic devices in the embodiments of the present disclosure include the above-mentioned mobile electronic devices and non-mobile electronic devices.

[0287] The embodiments of the present disclosure further provide a readable storage medium. A program or instruction is stored on the readable storage medium. When the program or instruction is executed by a processor, it implements each process of the above data processing method embodiment, and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0288] Among them, the processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes computer-readable storage media such as computer read-only memory ROM, random access memory RAM, magnetic disks, or optical discs.

[0289] Another embodiment of the present disclosure provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement each process of the above data processing method embodiment, and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0290] It should be understood that the chip mentioned in the embodiments of the present disclosure may also be referred to as a system-on-chip, system chip, chip system, or system-on-chip, etc.

[0291] The embodiments of the present disclosure provide a computer program product. The program product is stored in a storage medium and is executed by at least one processor to implement each process of the above data processing method embodiment, and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0292] It should be noted that in this article, the term "including", "comprising", or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article, or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article, or device. Without more limitations, the element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, method, article, or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present disclosure is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in a reverse order according to the functions involved. For example, the described method may be executed in an order different from that described, and various steps may be added, omitted, or combined. Additionally, the features described with reference to certain examples may be combined in other examples.

[0293] Through the description of the above embodiments, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases, the former is a better implementation method. Based on such an understanding, the technical solution of the present disclosure, in essence or the part that contributes to the prior art, can be embodied in the form of a computer software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to enable a terminal (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in various embodiments of the present disclosure.

[0294] The embodiments of the present disclosure have been described above in conjunction with the accompanying drawings. However, the present disclosure is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present disclosure, those of ordinary skill in the art can also make many forms without departing from the purpose of the present disclosure and the scope protected by the claims, and all of them belong to the protection scope of the present disclosure.

Claims

1. A data processing method, characterized in that, Including: Obtain the results of translating a source language text sample into a target language by at least two translation service providers, and obtain at least two first translation results; According to the at least two first translation results, respectively determine the evaluation values of the at least two translation service providers in each dimension among multiple dimensions, where the multiple dimensions include at least two of the following: consistency, cultural adaptability, fidelity, comprehensibility, real-time performance, accuracy, price. The consistency is used to indicate the similarity of the translation results of the at least two translation service providers. The cultural adaptability is used to indicate the translation accuracy of the cultural feature vocabulary by the translation service provider. The fidelity is used to indicate the matching degree between the translation result of the translation service provider and the source language text. The comprehensibility includes at least one of grammatical accuracy, readability, fluency, and ambiguity; According to the evaluation values of the at least two translation service providers in each dimension, determine the comprehensive evaluation values of the at least two translation service providers; According to the comprehensive evaluation values of the at least two translation service providers, select a target translation service provider from the at least two translation service providers; Translate the source language text through the target translation service provider.

2. The method according to claim 1, characterized in that The method further includes: Identify the industry category to which the source language text belongs; Select test statements of the industry category from a sample library, where the test statements include sample statements and expected translation results of the sample statements in the target language; Obtain the results of translating the sample statements into the target language by the at least two translation service providers, and obtain at least two second translation results; According to the similarity between each second translation result in the at least two second translation results and the expected translation result, determine the first evaluation values of the at least two translation service providers; The determining the comprehensive evaluation values of the at least two translation service providers according to the evaluation values of the at least two translation service providers in each dimension includes: According to the evaluation values of the at least two translation service providers in each dimension and the first evaluation values, determine the comprehensive evaluation values of the at least two translation service providers.

3. The method according to claim 1 or 2, characterized in that, When the multiple dimensions include consistency and the number of the at least two translation service providers is greater than 2, the determining the evaluation values of the at least two translation service providers in each dimension among multiple dimensions according to the at least two first translation results includes: Perform clustering calculation or similarity calculation on the at least two first translation results, determine that the consistency evaluation value of a first translation service provider is a first value, and determine that the consistency evaluation value of a second translation service provider is a second value, where the first translation service provider is a translation service provider with consistent first translation results among the at least two translation service providers, the second translation service provider is a translation service provider other than the target translation service provider among the at least two translation service providers, and the first value is greater than the second value.

4. The method according to claim 1 or 2, characterized in that When the multiple dimensions include cultural adaptability, determining, according to the at least two first translation results, the evaluation values of each dimension in the multiple dimensions for each of the at least two translation service providers includes: Performing word segmentation and part-of-speech tagging on the source language text sample, and identifying cultural feature words in the source language text sample according to the word segmentation and part-of-speech tagging results of the source language text sample; Performing word segmentation and part-of-speech tagging on each first translation result among the at least two first translation results, and checking whether each first translation result contains target language cultural feature words corresponding to the cultural feature words in the source language text sample according to the word segmentation and part-of-speech tagging results of each first translation result, to obtain a first check result; Calculating the semantic similarity between the source language text sample and each first translation result in the cultural feature dimension; Determining the cultural adaptability evaluation values of each translation service provider corresponding to each first translation result according to the first check result and semantic similarity of each first translation result.

5. The method according to claim 1 or 2, characterized in that When the multiple dimensions include fidelity, determining, according to the at least two first translation results, the evaluation values of each dimension in the multiple dimensions for each of the at least two translation service providers includes: Extracting key sentences and phrases from the source language text sample, where the key sentences and phrases include at least one of key entities, keywords, and key phrases; Checking whether each first translation result among the at least two first translation results correctly translates the key sentences and phrases, to obtain a second check result; Translating each first translation result back to the source language, and performing a consistency comparison between the source language translation results of each first translation result and the source language text sample, to obtain a consistency comparison result; Determining the fidelity evaluation values of each translation service provider corresponding to each first translation result according to the second check result and consistency comparison result of each first translation result.

6. The method according to claim 1 or 2, characterized in that When the multiple dimensions include comprehensibility, determining, according to the at least two first translation results, the evaluation values of each dimension in the multiple dimensions for each of the at least two translation service providers includes: Using a grammar parser of the target language to perform grammar analysis on each first translation result among the at least two first translation results, and identifying and recording the number of grammar errors in each first translation result; Calculating the readability scores of each first translation result using a readability algorithm; Calculating the fluency of each first translation result; Performing ambiguity detection on each first translation result, and obtaining an ambiguity detection result, where the ambiguity detection includes at least one of the following: analyzing whether there is ambiguity in each first translation result, and checking whether the reference of pronouns in each first translation result is clear; Determining the comprehensibility evaluation values of each translation service provider corresponding to each first translation result respectively according to the number of grammar errors in each first translation result, the readability scores of each first translation result, the fluency, and the ambiguity detection result.

7. The method according to claim 1 or 2, characterized in that, Determining the comprehensive evaluation value of the at least two translation service providers according to the evaluation values of the at least two translation service providers in each dimension includes: Calculating the weighted values of the evaluation values of the at least two translation service providers in each dimension according to the weights assigned to each dimension in advance to obtain the comprehensive evaluation value; The method further includes: Obtaining feedback information of the user on the translation results translated by each of the at least two translation service providers; Adjusting the weights of each dimension according to the feedback information.

8. A data processing device, characterized in that, Including: A first obtaining module, configured to obtain the translation results of the source language text sample into the target language by at least two translation service providers to obtain at least two first translation results; A first determining module, configured to respectively determine the evaluation values of the at least two translation service providers in each of the multiple dimensions according to the at least two first translation results, where the multiple dimensions include at least two of the following: consistency, cultural adaptability, fidelity, comprehensibility, real-time performance, accuracy, price, the consistency is used to indicate the similarity of the translation results of the at least two translation service providers, the cultural adaptability is used to indicate the translation accuracy of the translation service provider for culturally characteristic vocabulary, the fidelity is used to indicate the matching degree of the translation results of the translation service provider and the source language text, and the comprehensibility includes at least one of grammar accuracy, readability, fluency, and ambiguity; A second determining module, configured to determine the comprehensive evaluation value of the at least two translation service providers according to the evaluation values of the at least two translation service providers in each dimension; A first selection module, configured to select a target translation service provider from the at least two translation service providers according to the comprehensive evaluation values of the at least two translation service providers; A processing module, configured to translate the source language text through the target translation service provider.

9. An electronic device, characterized in that, Including a processor and a memory, where the memory stores a program or instruction that can run on the processor, and when the program or instruction is executed by the processor, the steps of the data processing method according to any one of claims 1 to 7 are implemented.

10. A readable storage medium, characterized in that, A program or instruction is stored on the readable storage medium, and when the program or instruction is executed by a processor, the steps of the data processing method according to any one of claims 1 to 7 are implemented.

Images