Medical machine translation method, device and electronic equipment

Abstract

The present invention belongs to the field of translation technology. A medical machine translation method, device and electronic device are provided. The method comprises: receiving original medical text data to be translated, performing word segmentation processing on the original medical text data, and identifying the word segmentations belonging to the target field; calling the corresponding vertical translation model according to the target field, using the vertical model to translate the corresponding word segmentations, and obtaining the word segmentation translations; inputting the original medical text data and each word segmentation translation into a general medical translation model, and the general medical translation model outputs the medical translation data; and displaying the medical translation data to the operator. The present invention uses a vertical translation model to accurately translate professional vocabulary in the field of medicine and medical equipment, and at the same time uses a general medical translation model to convert the original text data into a fluent and coherent translation, thereby achieving accurate translation of medical content, and can significantly improve the translation rate of the general medical translation model.

Description

Technical Field

[0001] The present invention relates to the field of translation technology, and in particular to a medical machine translation method, device and electronic equipment. Background Art

[0002] Translation refers to the process of translating from one language into another. There are many popular and mature translation software on the market. However, there are many professional terms and documents in the medical field, which are very different from the fields we usually come into contact with. The existing general translation software cannot accurately translate the relevant information in the medical field, which brings a lot of inconvenience to users. Moreover, if there is a translation error, it is likely to bring serious consequences.

[0003] Therefore, how to improve the accuracy of translation of information in the medical field is a technical problem that urgently needs to be solved. Summary of the invention

[0004] In this regard, the present invention provides a medical machine translation method, device, electronic device, computer storage medium and computer program product to solve the above technical problems.

[0005] The present invention provides a medical machine translation method, which comprises the following steps: receiving original medical text data to be translated, performing word segmentation processing on the original medical text data, and identifying the word segments belonging to a target field; wherein the target field comprises a pharmaceutical field and a medical device field; calling a corresponding vertical translation model according to the target field, using the vertical model to translate the corresponding word segments, and obtaining word segmentation translations; wherein the vertical translation model comprises a pharmaceutical field translation model and a medical device field translation model; inputting the original medical text data and each of the word segmentation translations into a general medical translation model, and the general medical translation model outputs medical translation data; and displaying the medical translation data to an operator.

[0006] In some embodiments, the receiving of original medical data to be translated includes: receiving the original medical data input by a camera module and / or a text input module and / or a file input module.

[0007] In some embodiments, the word segmentation processing of the medical original text data and identifying the word segmentations belonging to the target domain includes: using a word segmentation model based on a dual encoder to perform word segmentation processing on the medical original text data to obtain a plurality of word segmentations; wherein the word segmentation model includes a main encoder and an auxiliary encoder, the auxiliary encoder is used to provide the main encoder with coarse-grained representation information about medical terminology, and the main encoder determines the word segmentation span according to the coarse-grained representation information; performing semantic analysis on each of the word segmentations, and performing matching analysis based on the semantic analysis results and the preset semantic meanings of each target domain, and determining the target domain to which it belongs based on the matching analysis results.

[0008] In some embodiments, determining the target domain to which it belongs based on the matching analysis results includes: determining the target sub-domain of the target domain to which it belongs based on the matching analysis results; then calling the corresponding vertical translation model based on the target domain includes: calling the corresponding vertical translation model based on each of the target sub-domains; and the universal medical translation model is constructed based on a pre-trained language model.

[0009] In some embodiments, the auxiliary encoder is used to provide the main encoder with coarse-grained representation information about medical terminology, including: the auxiliary encoder determines whether the input medical text data contains title content, and if so, extracts medical terminology from the title content, and predicts the length of medical vocabulary monomers corresponding to the medical text data based on the medical terminology; and determines the coarse-grained representation information based on the length of the medical vocabulary monomers.

[0010] In some embodiments, displaying the translated text data to the operator includes: evaluating the operator's language adaptability to the original medical data to obtain a language adaptability level; if the language adaptability level is higher than the ability level threshold, displaying the translated text data and the original text data display page to the operator at the same time; if the language adaptability level is not higher than the ability level threshold, only displaying the translated text data to the operator on the display page.

[0011] The present invention also discloses a medical machine translation device, which includes a word segmentation module, a retrieval module, a translation module, and an output module; the word segmentation module is used to receive medical original text data to be translated, perform word segmentation processing on the medical original text data, and identify the word segments belonging to the target field; wherein the target field includes the medical field and the medical device field; the retrieval module is used to retrieve the corresponding vertical translation model according to the target field, use the vertical model to translate the corresponding word segments, and obtain the word segmentation translation; wherein the vertical translation model includes a medical field translation model and a medical device field translation model; the translation module is used to input the medical original text data and each of the word segmentation translations into a general medical translation model, and the general medical translation model outputs the medical translation data; the output module is used to display the medical translation data to an operator.

[0012] The present invention also provides an electronic device, comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the method described in any of the preceding items.

[0013] The present invention also provides a computer storage medium, wherein the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement any of the above methods.

[0014] The present invention also provides a computer program product, which implements the method described in the above embodiment when the computer program product is executed by a processor of an electronic device.

[0015] The beneficial effect of the present invention is that the present invention organically combines the general medical translation model and the vertical translation model, uses the vertical translation model to accurately translate professional vocabulary in the medical field and the medical device field, and uses the general medical translation model to convert the original text data into a smooth and fluent translation, thereby achieving accurate translation of medical content and significantly improving the translation rate of the general medical translation model. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 It is a schematic diagram of a scenario of quality inspection of outer packaging of a product based on image recognition technology disclosed in an embodiment of the present invention.

[0018] Figure 2 It is a schematic diagram of an output display interface after translating a Spanish original text into Chinese text according to an embodiment of the present invention.

[0019] Figure 3 It is a schematic diagram of the structure of a medical machine translation device disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The following is a description of the implementation of the present application by specific specific embodiments. People familiar with the technology can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0021] In addition, the technical features involved in the different embodiments of the present application described below can be combined with each other as long as they do not conflict with each other.

[0022] like Figure 1 As shown, an embodiment of the present invention discloses a medical machine translation method, which includes the following steps: receiving medical original text data to be translated, performing word segmentation processing on the medical original text data, and identifying the word segments belonging to the target field; wherein the target field includes the medical field and the medical device field; calling the corresponding vertical translation model according to the target field, using the vertical model to translate the corresponding word segments, and obtaining the word segmentation translation; wherein the vertical translation model includes a medical field translation model and a medical device field translation model; inputting the medical original text data and each of the word segmentation translations into a general medical translation model, and the general medical translation model outputs medical translation data; and displaying the medical translation data to an operator.

[0023] In view of the problem that the existing general translation software in the background technology cannot accurately translate the relevant information in the medical field, the present invention designs a general medical translation model and a variety of vertical translation models, and realizes accurate translation in the medical field through the mutual cooperation of the general medical translation model and the vertical translation model. Specifically as follows: the operator inputs the original medical data to be translated into the translation software, and the translation software performs word segmentation processing on the original medical data, and analyzes which of the obtained word segments belong to the target fields including the medical field and the medical device field. Then, according to the target fields involved, the corresponding vertical translation model is retrieved. The vertical translation model is a pre-built and trained model suitable for special translation of vocabulary in the corresponding target field. The retrieved vertical model is used to translate the corresponding word segmentation to obtain the word segmentation translation. Finally, the general medical translation model is used to comprehensively analyze the above-mentioned obtained medical original data and each word segmentation translation to achieve accurate translation of the original text data, that is, to obtain medical translation data, and output it to the operator.

[0024] Therefore, the present invention organically combines the general medical translation model and the vertical translation model, uses the vertical translation model to accurately translate professional vocabulary in the fields of medicine and medical devices, and uses the general medical translation model to convert the original text data into a fluent and coherent translation, thereby achieving accurate translation of medical content and significantly improving the translation rate of the general medical translation model.

[0025] In some embodiments, the receiving of original medical data to be translated includes: receiving the original medical data input by a camera module and / or a text input module and / or a file input module.

[0026] In an embodiment of the present invention, the solution of the present invention is packaged as translation software, such as a translation APP installed on a mobile terminal, and the mobile terminal includes but is not limited to a mobile phone, a tablet computer, a wearable device, a virtual reality device, etc. The mobile terminal is generally equipped with a camera module, a text input module, and a file input module, and the above modules can respectively support the operator to input the captured image data, text characters, and files of a suitable format related to the content to be translated into the translation software, and the translation software then uses a corresponding analysis method to identify the above data related to the content to be translated, so as to extract the original medical text data.

[0027] In some embodiments, the word segmentation processing of the medical original text data and identifying the word segmentations belonging to the target domain includes: using a word segmentation model based on a dual encoder to perform word segmentation processing on the medical original text data to obtain a plurality of word segmentations; wherein the word segmentation model includes a main encoder and an auxiliary encoder, the auxiliary encoder is used to provide the main encoder with coarse-grained representation information about medical terminology, and the main encoder determines the word segmentation span according to the coarse-grained representation information; performing semantic analysis on each of the word segmentations, and performing matching analysis based on the semantic analysis results and the preset semantic meanings of each target domain, and determining the target domain to which it belongs based on the matching analysis results.

[0028] In an embodiment of the present invention, a word segmentation model is first used to extract multiple word segments from medical original text data. The word segmentation model includes a main encoder and an auxiliary encoder. The auxiliary encoder is used to determine coarse-grained representation information about medical terminology. The coarse-grained representation information represents the single interception word quantity that should be taken during word segmentation, that is, how many words are intercepted at one time for word segmentation processing; the main encoder determines the word segmentation span according to the coarse-grained representation information determined by the auxiliary encoder, that is, the word quantity intercepted for each word segmentation processing, such as 3 words, 5 words, 10 words, etc.

[0029] Then, the semantic meaning of each segmented word is determined by semantic analysis, and the semantic analysis result of the segmented word is matched and analyzed with the semantic meaning of each preset target field. If the match is successful, the segmented word is assigned to the corresponding target field.

[0030] In some embodiments, determining the target domain to which it belongs based on the matching analysis results includes: determining the target sub-domain of the target domain to which it belongs based on the matching analysis results; then calling the corresponding vertical translation model based on the target domain includes: calling the corresponding vertical translation model based on each of the target sub-domains; and the universal medical translation model is constructed based on a pre-trained language model.

[0031] In the embodiment of the present invention, multiple target sub-fields are further subdivided in each target field. For example, the pharmaceutical field includes 9 sub-fields, including medicine (human medicine), non-clinical (human medicine), clinical (human medicine), medicine (animal health), pharmacovigilance, regulations, imaging, protocols, and others. The medical device field includes 6 sub-fields, including IVD, electrical, orthopedic, clinical, vascular, and other (comprehensive and general). A matching vertical translation model is constructed for each sub-field. Therefore, the corresponding target sub-field can be determined during the aforementioned semantic analysis, and the corresponding vertical translation model is retrieved from the matched target sub-field.

[0032] The above vertical translation models are preferably built based on deep learning and natural language processing technologies, which mainly include several functional modules: context understanding module, professional terminology and concept recognition module, encoder-decoder architecture, and attention mechanism module. The explanations of each functional module are as follows: Context understanding module: Use deep learning technology, such as recurrent neural network (RNN) or Transformer architecture, to understand the context information in medical texts, which is very important for accurately translating medical terms and concepts.

[0033] Professional terminology and concept recognition module: specially designed to identify and process professional terminology and concepts in the medical field to ensure the accuracy and professionalism of the translation.

[0034] Encoder-Decoder Architecture: In Neural Machine Translation (NMT), the encoder is used to understand the source language text and the decoder is used to generate the target language text. This architecture allows the model to capture the complex mapping relationship between the source and target languages.

[0035] Attention mechanism module: In the Transformer model, the attention mechanism enables the model to focus on the key information in the source text during the translation process, improving the accuracy and fluency of the translation.

[0036] However, the above-mentioned universal medical translation model is built based on a pre-trained language model, which refers to BERT (Bidirectional Encoder Representations from Transformers) and its variant in the medical field, BioBERT. Pre-training on a medical corpus improves the performance of the universal medical translation model in medical text mining and translation tasks. The universal medical translation model is obtained by retraining the pre-trained language model (i.e., light training) using pre-collected small sample data.

[0037] In some embodiments, the auxiliary encoder is used to provide the main encoder with coarse-grained representation information about medical terminology, including: the auxiliary encoder determines whether the input medical text data contains title content, and if so, extracts medical terminology from the title content, and predicts the length of medical vocabulary monomers corresponding to the medical text data based on the medical terminology; and determines the coarse-grained representation information based on the length of the medical vocabulary monomers.

[0038] In an embodiment of the present invention, the auxiliary encoder is used to determine the coarse-grained representation information of the input medical original data, and provide it to the main encoder to indicate the amount of words intercepted at a single time during the word segmentation process. Specifically, medical terminology is extracted from the article title of the input medical original data. The medical terminology extracted from the article title roughly determines the subject content of the article, and then it can be determined which medical terminology will be involved in the article, and the length of the terminology in different medical sub-fields has its own characteristics. Therefore, based on the medical terminology extracted from the article title, the length of the vocabulary monomers of the medical vocabulary involved in the medical original data can be roughly determined, and then the corresponding coarse-grained representation information can be determined.

[0039] It should be noted that this function of the auxiliary encoder can also be realized with the help of an external model (such as the aforementioned pre-trained language model or vertical model), that is, the external model predicts and analyzes the medical terminology extracted from the article title to obtain the length of the vocabulary monomer, and the auxiliary encoder then determines the corresponding coarse-grained representation information based on this. Obviously, the corresponding coarse-grained representation information is positively correlated with the length of the vocabulary monomer, that is, the larger the predicted length of the vocabulary monomer, the higher the coarse-grainedness, that is, the more words are intercepted at a time; correspondingly, the smaller the predicted length of the vocabulary monomer, the lower the coarse-grainedness, that is, the fewer words are intercepted at a time.

[0040] In addition, if it is determined that the input medical original text data contains title content, it is possible that the title content is not placed in the viewfinder in the picture data taken by the operator. In this case, the camera module can be controlled to adjust the focus to place the title content in the viewfinder, so that medical terminology can be extracted from the title content. And if the text input module or file input module is used for input, the operator can be prompted to further input the title content corresponding to the original text data.

[0041] In some embodiments, displaying the translated text data to the operator includes: evaluating the operator's language adaptability to the original medical text data to obtain a language adaptability level; if the language adaptability level is higher than the ability level threshold, displaying the translated text data and the original text data to the operator on a display page at the same time; if the language adaptability level is not higher than the ability level threshold, only displaying the translated text data to the operator on the display page.

[0042] In the embodiments of the present invention, when using translation software, operators may not only need to view the translated text, but also need to view the original text at the same time based on their different needs, so as to verify and compare the translated text with the original text, for example, to verify the accuracy of vocabulary translation, grammar translation, etc. However, existing translation software can only provide operators with a simple translation page, or provide fixed original text page and translation page, and cannot automatically adjust the appropriate page display mode according to the actual viewing needs of the operator.

[0043] In view of the above technical problems, the present invention is provided to evaluate the language adaptability of the operator to the original medical text data to obtain the language adaptability level. For example, through the operator's login operation, relevant background information such as education, major, resume information, etc. can be obtained, and the language adaptability level of the operator to the original medical text data (such as English, French, Spanish, Arabic, etc.) can be evaluated and obtained. The higher the language adaptability level, the higher the language reading comprehension ability of the operator, and vice versa.

[0044] At the same time, when the operator's language adaptation ability level for the original medical text data is lower than the ability level threshold, it means that he or she basically has no reading and comprehension ability for the language. At this time, it is obvious that there is no need and necessity to verify and compare the translated text with the original text. Therefore, only the translated text data is displayed to the operator on the display page. When the operator's language adaptation ability level for the original medical text data is higher than the ability level threshold, it means that he or she obviously has the reading and comprehension ability for the language. At this time, there may be a need and necessity to verify and compare the translated text with the original text. Therefore, the translated text data and the original text data display page are displayed to the operator at the same time. Figure 2 Of course, the operator can also manually close the display subpage of the original text data in the display page according to their own needs, and only display the translated text.

[0045] It should be noted that the translated text data and the original text data in the display page may be arranged and displayed in any suitable manner such as up and down, left and right, etc., and the present invention is not limited to this.

[0046] An improved solution to this embodiment is further provided: when the translated text data and the original text data are simultaneously displayed to an operator on a display page, the method further includes: determining a display duration according to the language adaptation capability level, and when the display duration is reached, controlling the display page to close the display of the original text data; wherein the display duration is negatively correlated with the language adaptation capability level.

[0047] When the translated text data and the original text data are simultaneously displayed to the operator on the display page, the operator checks the translated text against the original text, and the checking time is different for different language adaptability levels. The higher the language adaptability level, the less time it takes to check, and the lower the language adaptability level, the longer it takes to check. In view of this situation, the present invention is set to determine the display time of the original text data on the display page according to the language adaptability level and the corresponding negative correlation, and close the original text data when the display time is reached, so that the operator can perform operations such as copying.

[0048] like Figure 3 As shown, an embodiment of the present invention also discloses a medical machine translation device, which includes a word segmentation module, a retrieval module, a translation module, and an output module; the word segmentation module is used to receive medical original text data to be translated, perform word segmentation processing on the medical original text data, and identify the word segments belonging to the target field; wherein the target field includes the medical field and the medical device field; the retrieval module is used to retrieve the corresponding vertical translation model according to the target field, use the vertical model to translate the corresponding word segments, and obtain the word segmentation translation; wherein the vertical translation model includes the medical field translation model and the medical device field translation model; the translation module is used to input the medical original text data and each of the word segmentation translations into a general medical translation model, and the general medical translation model outputs the medical translation data; the output module is used to display the medical translation data to the operator.

[0049] An embodiment of the present invention further discloses an electronic device, comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the method described in the above embodiment.

[0050] An embodiment of the present invention further discloses a computer storage medium, wherein the computer storage medium stores a computer program, and the computer program is executed by a processor to implement the method described in the above embodiment.

[0051] The embodiment of the present invention further discloses a computer program product. When the computer program product is executed by a processor of an electronic device, the method described in the above embodiment is implemented.

[0052] Computer readable media include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, disk storage, quantum memory, graphene-based storage media or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0053] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0054] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0055] The terms used in one or more embodiments of this specification are only for the purpose of describing specific embodiments, and are not intended to limit one or more embodiments of this specification. The singular forms of "a", "said" and "the" used in one or more embodiments of this specification and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0056] It should be understood that although the terms first, second, third, etc. may be used to describe various information in one or more embodiments of this specification, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of this specification, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0057] The above description is merely a preferred embodiment of one or more embodiments of the present specification and is not intended to limit one or more embodiments of the present specification. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of the present specification shall be included in the scope of protection of one or more embodiments of the present specification.

Claims

1. A medical machine translation method, characterized in that , the method comprises the following steps: Receiving original medical text data to be translated, performing word segmentation processing on the original medical text data, and identifying the word segments belonging to the target field; wherein the target field includes the pharmaceutical field and the medical device field; Retrieving a corresponding vertical translation model according to the target domain, and using the vertical translation model to translate the corresponding segmented words to obtain segmented word translations; wherein the vertical translation model includes a translation model for the medical field and a translation model for the medical device field; Inputting the original medical text data and each of the segmented translations into a universal medical translation model, the universal medical translation model outputting medical translation data; and displaying the medical translation data to an operator; Perform word segmentation processing on the original medical text data and identify the word segments belonging to the target field, including: The medical original text data is segmented using a dual-encoder-based segmentation model to obtain a plurality of segmented words; wherein the segmentation model includes a main encoder and an auxiliary encoder, the auxiliary encoder is used to provide the main encoder with coarse-grained representation information about medical terminology, and the main encoder determines the segmentation span according to the coarse-grained representation information; Performing semantic analysis on each of the segmented words, and performing matching analysis on the semantic analysis results with the semantic meanings of each preset target field, and determining the target field to which it belongs based on the matching analysis results; The auxiliary encoder is used to provide the main encoder with coarse-grained representation information about medical terminology, including: The auxiliary encoder determines whether the input medical original text data contains title content, and if so, extracts medical terminology from the title content, and predicts the length of medical vocabulary monomers corresponding to the medical original text data based on the medical terminology; The coarse-grained representation information is determined based on the length of the medical vocabulary monomer. Specifically: the larger the predicted length of the medical vocabulary monomer is, the higher the coarse-grainedness is, that is, the more words are intercepted at a single time; the smaller the predicted length of the medical vocabulary monomer is, the lower the coarse-grainedness is, that is, the fewer words are intercepted at a single time.

2. A medical machine translation method according to claim 1, characterized in that: Receiving original medical data to be translated includes: receiving the original medical data input by a camera module and / or a text input module and / or a file input module.

3. A medical machine translation method according to claim 1, characterized in that: The target areas to which it belongs are determined based on the matching analysis results, including: Determine the target subfield of the target field to which it belongs based on the matching analysis results; Then, calling the corresponding vertical translation model according to the target field includes: calling the corresponding vertical translation model according to each of the target sub-fields; and the universal medical translation model is constructed based on a pre-trained language model.

4. A medical machine translation method according to claim 1, characterized in that: The presenting the medical translation data to the operator comprises: Evaluate the language adaptability of the operator to the original medical data to obtain the language adaptability level; If the language adaptation capability level is higher than the capability level threshold, the medical translation data and the medical original data display page are simultaneously displayed to the operator; If the language adaptation capability level is not higher than the capability level threshold, only the medical translation data is displayed to the operator on the display page.

5. A medical machine translation device, comprising a word segmentation module, a retrieval module, a translation module, and an output module; characterized in that: The word segmentation module is used to receive the original medical text data to be translated, perform word segmentation processing on the original medical text data, and identify the word segments belonging to the target field; wherein the target field includes the pharmaceutical field and the medical device field; The retrieval module is used to retrieve the corresponding vertical translation model according to the target field, and use the vertical translation model to translate the corresponding segmented words to obtain segmented word translations; wherein the vertical translation model includes a translation model for the medical field and a translation model for the medical device field; The translation module is used to input the original medical text data and each of the segmented translations into a universal medical translation model, and the universal medical translation model outputs medical translation data; The output module is used to display the medical translation data to the operator; The medical original text data is segmented and the segmented words belonging to the target field are identified, including: The medical original text data is segmented using a dual-encoder-based segmentation model to obtain a plurality of segmented words; wherein the segmentation model includes a main encoder and an auxiliary encoder, the auxiliary encoder is used to provide the main encoder with coarse-grained representation information about medical terminology, and the main encoder determines the segmentation span according to the coarse-grained representation information; Performing semantic analysis on each of the segmented words, and performing matching analysis on the semantic analysis results with the semantic meanings of each preset target field, and determining the target field to which it belongs based on the matching analysis results; The auxiliary encoder is used to provide the main encoder with coarse-grained representation information about medical terminology, including: The auxiliary encoder determines whether the input medical original text data contains title content, and if so, extracts medical terminology from the title content, and predicts the length of medical vocabulary monomers corresponding to the medical original text data based on the medical terminology; The coarse-grained representation information is determined based on the length of the medical vocabulary monomer. Specifically: the larger the predicted length of the medical vocabulary monomer is, the higher the coarse-grainedness is, that is, the more words are intercepted at a single time; the smaller the predicted length of the medical vocabulary monomer is, the lower the coarse-grainedness is, that is, the fewer words are intercepted at a single time.

6. An electronic device comprising: At least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the method according to any one of claims 1 to 4.

7. A computer storage medium storing a computer program, characterized in that: The computer program is executed by a processor to implement the method according to any one of claims 1 to 4.

8. A computer program product comprising a computer program stored on a non-transitory computer readable medium, characterized in that: When the computer program product is executed by a processor of an electronic device, the method according to any one of claims 1 to 4 is implemented.

Images