A translation system and method, apparatus for translation service invocation

Abstract

The application provides a translation system and a method and device for calling a translation service, multiple basic translation function modules can be created to meet various complex translation requirements in multiple scenes and multiple fields, cloud-native container management and scheduling capabilities are combined on this basis, distributed containerized automatic deployment of various translation function modules in the translation system is realized, whole translation service on the cloud can be quickly and continuously delivered and dynamically expanded, service deployment cost and time are reduced, iteration efficiency is accelerated, and product update cycle is shortened.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cloud computing, and in particular to a translation system and a method and device for calling a translation service. BACKGROUND

[0002] With the development of economic globalization and trade internationalization, the rapidly rising cross-border e-commerce is reshaping the new pattern of global trade and has become a new driving force for trade growth that attracts attention from all countries. The first obstacle in front of cross-border e-commerce is the multilingual problem. The spread of commodity information and the effective communication of buyers and sellers are affected by the multilingual problem. For the e-commerce scene with huge volume and high time efficiency requirement, inefficient manual translation cannot meet the actual demand, so building a commercially available machine translation system has become the most suitable solution and is widely used.

[0003] However, the demand for machine translation is quite different in different scenarios and different fields, especially in the e-commerce scene. For example, the translation accuracy of the model, size, specification, etc. of the goods is extremely high, and the time efficiency requirement for the communication scene translation is higher. Even the most advanced neural network machine translation technology cannot meet the actual production demand with only one translation decoder, and a large number of upstream and downstream functional modules and engineering work need to be done. For example, spelling correction of the original text, forced translation intervention of the original text, optimization of multiple translation model results, etc. At the same time, the translation model needs to be iterated on the basis of corpus sedimentation and model optimization. If the deployment and operation of the entire translation system are very complex, it will limit the speed of updating and iterating the translation model. Therefore, how to quickly deploy and manage the translation system in one key is very important. SUMMARY

[0004] In order to solve the above problems, the present application provides a translation system and a method and device for calling a translation service, which can meet the demand of various complex translation service application scenarios in multiple scenarios and multiple fields. On this basis, the cloud-native container management and scheduling capabilities are combined to realize the distributed containerization automatic deployment of each translation functional module in the system, which can realize the rapid and continuous delivery and dynamic expansion of the overall translation service function on the cloud, reduce the service deployment cost and time, accelerate the iteration efficiency and shorten the product update cycle.

[0005] The embodiment of the present application provides a translation system, comprising: a deployment controller;

[0006] The deployment controller is used for creating a plurality of translation functional modules corresponding to a required translation service application scenario according to the translation service application scenario, and deploying the plurality of translation functional modules in different containers respectively.

[0007] Optionally, the system further comprises:

[0008] a business logic layer: obtaining information of a translation function module to be created, the information of the translation function module to be created including a translation service identifier and a translation function identifier, the translation service identifier representing a translation service application scenario, and the translation function identifier representing the translation function module;

[0009] a data layer: obtaining a data file of the translation function module to be created according to the translation service identifier and the translation function identifier;

[0010] a container processing layer: creating an image file of the translation function module in a container according to the data file.

[0011] Optionally,

[0012] the data layer: obtaining, by using a deep learning mechanism, the data file of each translation function module corresponding to the translation service application scenario according to the translation service application scenario; and cloud-storing the data file of each translation function module corresponding to the translation service application scenario.

[0013] Optionally,

[0014] the business logic layer: obtaining information of a translation function module to be updated, the information of the translation function module to be updated including a translation service identifier and a translation function identifier;

[0015] the data layer: querying, according to the translation service identifier and the translation function identifier, a data file of the translation function module to be updated on a cloud storage, and updating the data file;

[0016] the container processing layer: updating an image file of the translation function module in a container according to the updated data file.

[0017] Optionally,

[0018] the translation function module includes a total control module, an original text intervention module, a translated text intervention module, and a decoder.

[0019] The application further provides a translation service calling method, including:

[0020] preprocessing received original text;

[0021] determining a translation model for translating the original text according to a preprocessing result;

[0022] calling the translation model to translate the original text.

[0023] Optionally, preprocessing received original text includes:

[0024] The original text is parsed, and original text intervention processing is performed on the parsed original text to obtain an original text intervention result;

[0025] According to the original text intervention result, the original text after intervention is processed to obtain a translation intervention result.

[0026] Optionally, according to the pre-processing result, a translation model for translating the original text is determined, comprising:

[0027] According to the translation intervention result, a translation service application scenario to which the original text belongs is determined;

[0028] According to the translation service application scenario, a translation model corresponding to the translation service application scenario is determined;

[0029] The translation model is a translation model corresponding to the translation service application scenario, which is learned and calculated according to the translation service application scenario by using a deep neural network learning mechanism.

[0030] Optionally, after the original text is translated by the translation model, the method further comprises:

[0031] The post-processing of the translation includes translation optimization, case restoration, undecoded word processing, and sensitive word processing.

[0032] The application also provides a translation service calling device, comprising:

[0033] A total control module is configured to perform pre-processing on the received original text, determine a translation model for translating the original text according to a pre-processing result, and call a decoder corresponding to the translation model according to the translation model;

[0034] The decoder is configured to translate the original text by using the translation model according to a calling instruction of the total control module.

[0035] Optionally, the device further comprises an original text intervention module and a translation intervention module;

[0036] The total control module is configured to call the original text intervention module after parsing the original text;

[0037] The original text intervention module is configured to perform original text intervention processing on the parsed original text according to a calling instruction of the total control module, and return an original text intervention result to the total control module;

[0038] The total control module is further configured to call the translation intervention module according to the original text intervention result returned by the original text intervention module;

[0039] The translation intervention module is configured to perform translation intervention processing on the original text after the original text intervention according to the calling instruction of the total control module, and return the original text intervention result to the total control module.

[0040] Optionally, the total control module is further configured to determine a translation service application scenario to which the original text belongs according to the translation intervention result returned by the translation intervention module, and determine a translation model corresponding to the translation service application scenario according to the translation service application scenario.

[0041] The translation model is obtained by learning and calculating the translation model corresponding to the translation service application scenario by using a deep neural network learning mechanism according to the translation service application scenario.

[0042] Optionally, the total control module is further configured to perform post-processing on the translation, including translation optimization, case restoration, undecoded word processing, and sensitive word processing.

[0043] The application also provides a server, including a memory, a processor, and a communication component.

[0044] The memory is configured to store a computer program.

[0045] The processor is coupled with the memory and the communication component, and is configured to execute the computer program to perform the steps or operations of the above-described translation system or the above-described method.

[0046] The application also provides a computer-readable storage medium storing a computer program, which can implement the steps or operations of the above-described translation system or the above-described method when the computer program is executed by a computer.

[0047] The translation system distributed deployment method of the embodiment of the application can manufacture each translation function module into an image container, and manage the container by using a deployment controller to realize automatic deployment of the translation function module, support the creation of multiple sets of translation service systems, and isolate each set of translation service systems from each other.

[0048] In the embodiment of the application, the translation system is distributed and deployed, that is, each function of the translation system is modularized and packaged, so that each function module can cover the translation service requirements of multiple scenarios and multiple fields, and can be distributed and deployed on the cloud to quickly and automatically deploy the entire translation system service, support iterative updating of the model and dynamic expansion of the service, and have high flexibility and maintainability. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings described below are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort based on these drawings.

[0050] Figure 1 The distributed deployment system architecture diagram of the translation system provided by an embodiment of the present application is shown in FIG. 1.

[0051] Figure 2 The flowchart of the distributed deployment method of the translation system provided by an embodiment of the present application is shown in FIG. 2.

[0052] Figure 3 The structure diagram of the distributed deployment controller of the translation system provided by an embodiment of the present application is shown in FIG. 3.

[0053] Figure 4 The flowchart of the translation service calling method provided by an embodiment of the present application is shown in FIG. 4.

[0054] Figure 5 The distributed deployment system architecture diagram of the translation system provided by another embodiment of the present application is shown in FIG. 5.

[0055] Figure 6 The structure diagram of the translation service calling device provided by an embodiment of the present application is shown in FIG. 6.

[0056] Figure 7 The distributed deployment system architecture diagram of the translation system provided by another embodiment of the present application is shown in FIG. 7.

[0057] Figure 8 The structure diagram of the server provided by an embodiment of the present application is shown in FIG. 8. DETAILED DESCRIPTION

[0058] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following will combine the drawings in the embodiments of the present application to make a clear and complete description of the technical solutions in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present application.

[0059] The terminology used in the description of the implementation of the application is only for the purpose of describing specific embodiments and is not intended to limit the application. The singular forms "a", "said" and "the" used in the implementation of the application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. "A plurality" generally includes at least two, but does not exclude the possibility of including at least one.

[0060] It should be understood that the term "and / or" used herein is only to describe the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the three cases of A alone, A and B together, and B alone. In addition, the character " / " herein generally represents an "or" relationship between the front and rear associated objects.

[0061] It should also be noted that the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the goods or systems including a series of elements not only include those elements, but also include other elements not explicitly listed, or include elements inherent to such goods or systems. Without more limitations, the elements defined by the statement "including a" do not exclude the presence of other identical elements in the goods or systems including the elements.

[0062] The inventors found during the implementation of the application:

[0063] The existing machine translation is to integrate all functional modules into one module in a machine, that is, to add pre-processing and post-processing logic in the decoder module, including sentence segmentation, word segmentation, case conversion, original text intervention, translation text term intervention, translation sentence intervention, out-of-vocabulary processing, deduplication, sensitive word processing, brand word processing, and word segmentation removal modules. Coupling all functional modules together can easily cause logic stacking confusion, making it difficult to maintain and expand functions. Moreover, as the translation demand increases, the complexity of each module will increase when the translation model is iterated and the function is updated, making operation and maintenance extremely difficult.

[0064] To solve the above technical problems, in the embodiments of the present application, the translation system is distributedly deployed, that is, each function of the translation system is modularly packaged, so that each functional module can cover multiple scene and field translation service requirements, and distributedly deployed on the cloud, which can quickly and automatically deploy the entire translation system service, while supporting model iteration and service dynamic expansion, with high flexibility and maintainability.

[0065] Terminology:

[0066] Machine translation: Let the machine realize the mapping from one language to another language.

[0067] Neural machine translation: machine translation based on deep neural networks, usually consisting of an encoder and a decoder.

[0068] Kubernetes: an open source, for managing cloud platform in multiple hosts on containerized applications.

[0069] Docker container: an open source application container engine, which allows developers to package their applications and dependencies into a portable container, and then publish to any machine to achieve virtualization.

[0070] Figure 1 The distributed deployment system architecture of the translation system provided by an embodiment of the application is shown in Figure 1 The deployment controller deploys the plurality of translation function modules in different containers according to the required translation service application scenarios and the plurality of translation function modules corresponding to the translation service application scenarios.

[0071] In specific implementation, the deployment controller creates the plurality of translation function modules in each translation service application scenario by calling an application program interface (API), and each translation function module can be deployed in different containers on different hosts in a kubernetes cluster. Each translation function module obtains a data file required by the corresponding translation function module from cloud storage, and the data file is a data package required for deploying the corresponding translation function module.

[0072] Kubernetes is an open source application solution for managing containers (docker) and can be widely applied as a bottom-layer container solution, having certain standardization and universality. By using the container management and scheduling capabilities of kubernetes, a plurality of translation service suites with a plurality of container sets can be quickly built to provide complete translation services to the outside, and each translation service is isolated from each other and does not interfere with each other, so that different translation function modules can be built according to the requirements of different translation service application scenarios, and a plurality of translation service function models can be simultaneously iterated and updated.

[0073] Due to the great difference in the requirements of machine translation in different application scenarios and different application fields, especially in e-commerce scenarios, for example, the translation accuracy of the model, size, and specification of goods is extremely high, and the timeliness requirement of the communication application scenario for translation is higher. By the embodiment of the application, a plurality of translation function modules can be created, each set of translation function modules (including a plurality of translation function modules) supports a corresponding translation service application scenario, and each set of translation function modules can be distinguished by a translation service identifier representing the translation service application scenario.

[0074] Based on Figure 1A distributed deployment system architecture diagram of the translation system shown, Figure 2 A flowchart of a distributed deployment method of the translation system provided for an embodiment of the present application is shown in Figure 2 The method comprises the following steps:

[0075] 101. Obtain information of a translation function module to be created;

[0076] Figure 3 A structure diagram of a distributed deployment controller of the translation system provided for an embodiment of the present application is shown in Figure 3 The deployment controller comprises a business logic layer, a data layer and a container processing layer.

[0077] The business logic layer is mainly responsible for managing and organizing each translation function module of the translation system deployment described in the embodiments of the present application, and supports the creation, dynamic combination and elastic scaling of each translation function module.

[0078] For example, when a set of translation service system applied to a certain translation service application scenario needs to be created, each set of translation service system comprises a plurality of translation function modules. In step 101, the business logic layer of the deployment controller needs to obtain the information of the translation function module to be created according to the user's creation instruction, which comprises a translation service identifier and a translation function identifier, wherein the translation service identifier represents the translation service application scenario, and the translation function identifier represents the translation function module.

[0079] 102. Obtain the data file of the translation function module to be created according to the translation service identifier and the translation function identifier included in the information of the translation function module;

[0080] As shown in Figure 3 The data layer of the deployment controller processes some intermediate data according to the provided information of the translation function module to be deployed, and then generates the final deployment instruction and data file in combination with the deployment template file of each translation function module.

[0081] Before step 102, the data layer uses a deep neural network learning mechanism to learn and calculate the data file of each translation function module corresponding to the translation service application scenario in advance according to the translation service application scenario; and the data file of each translation function module corresponding to the translation service application scenario is stored in the cloud.

[0082] In the specific implementation of step 102, the data layer queries the cloud storage to obtain the data file of the translation function module to be created according to the translation service identifier and the translation function identifier.

[0083] 103. Based on the data file, create an image file of the translation function module in the container.

[0084] like Figure 3 As shown, the container processing layer of the deployment controller creates the image file of the translation function module by calling the Kubernetes application programming interface (API).

[0085] Specifically, when creating the image file of the translation function module in the container, it is necessary to determine the container corresponding to the translation function identifier on the host corresponding to the translation service identifier and the translation function identifier of the translation function module to create the image file of the translation function module.

[0086] Therefore, the deployment controller described in this embodiment of the invention can provide HTTP services to support adding, modifying, deleting, and querying instances of translation function modules on the Kubernetes cluster. For example, taking updating or iterating a translation function module as an example, a typical update or iteration of a translation function module usually refers to updating or iterating the reloading of the translation model data in the decoder.

[0087] Specifically, the implementation process of deploying the controller to update the translation function module is as follows:

[0088] Business logic layer: Obtain information about the translation function module to be updated, wherein the information about the translation function module to be updated includes the translation service identifier and the translation function identifier;

[0089] Data layer: Based on the translation service identifier and translation function identifier, query the data file of the translation function module to be updated on the cloud storage, and update the data file;

[0090] Container processing layer: Updates the image file of the translation function module in the container based on the updated data file.

[0091] The distributed deployment method of the translation system in this invention can create each translation function module as a mirror container, and manage the instantiation of the container through a deployment controller. This distributed deployment method is applicable to various cloud container technologies to achieve automated deployment of translation function modules. It also supports the creation of multiple translation service systems, with each translation service system isolated from each other and distinguished by a translation service identifier.

[0092] based on Figure 1 The diagram shown is a distributed deployment system architecture diagram of the translation system. Figure 4 This is a flowchart illustrating a translation service invocation method according to an embodiment of the present invention, as shown below. Figure 4 As shown, it includes:

[0093] 201. Perform preprocessing on the received original text;

[0094] In practical applications, when a translation request is received from a user, the request includes the original text to be translated. For example... Figure 1 As shown, the deployment controller provides HTTP services to the outside world. When it receives a user's translation request, it needs to process the original text to determine the application scenario for translating the original text, thereby determining the translation service identifier of the original text. Then, based on the translation service identifier, it determines the translation system (i.e., the translation function module) corresponding to the translation service identifier. Finally, it uses the translation model in the translation function module to translate the original text.

[0095] Specifically, step 201 includes the following in its implementation:

[0096] The original text is parsed, and the parsed original text is subjected to original text intervention processing to obtain the original text intervention result;

[0097] Based on the original text intervention results, the original text after the original text intervention is subjected to translation intervention processing to obtain the translation intervention results.

[0098] 202. Based on the preprocessing results, determine the translation model for translating the original text;

[0099] Specifically, step 202 includes the following in its implementation:

[0100] Based on the results of the translation intervention, determine the translation service application scenario to which the original text belongs;

[0101] Based on the application scenario of the translation service, determine the translation model corresponding to the application scenario of the translation service;

[0102] The translation model is obtained by learning and calculating a translation model corresponding to the translation service application scenario using a deep neural network learning mechanism.

[0103] 203. Use the translation model to translate the original text.

[0104] Optionally, after translating the original text using the translation model, the method further includes:

[0105] The translated text is post-processed, including translation optimization, case restoration, undecoded word processing, and sensitive word processing.

[0106] based on Figure 1 The diagram shown is a distributed deployment system architecture diagram of the translation system. Figure 5 A distributed deployment system architecture diagram of a translation system is provided for another embodiment of the present invention, such as... Figure 5As shown, the deployment controller is provided with an API interface, and the deployment controller calls each translation function module corresponding to each translation service application scenario through the API interface, wherein each translation function module corresponding to each translation service application scenario comprises a total control module, an original text intervention module, a translated text intervention module and a decoder.

[0107] In actual application, when receiving a translation request initiated by a user, the translation request comprises original text to be translated. Figure 1 As shown, the deployment controller is provided with an API interface, and the deployment controller calls each translation function module corresponding to each translation service application scenario through the API interface, wherein each translation function module corresponding to each translation service application scenario comprises a total control module, an original text intervention module, a translated text intervention module and a decoder.

[0108] Figure 6 A structure diagram of a translation service calling device provided by an embodiment of the present application is shown in FIG. 2. Figure 6 As shown, in the embodiment of the present application, the total control module serves as a unified entrance for a translation service request externally, and is responsible for service calling and integration processing of the original text intervention module, the translated text intervention module and the decoder internally, and is responsible for pre-processing and post-processing of a translation result in the translation system.

[0109] The total control module is configured to call the original text intervention module after analyzing the original text, wherein the analysis of the original text by the total control module comprises, for example, html analysis, paragraph segmentation, natural sentence segmentation and other analysis processing.

[0110] The original text intervention module is configured to perform original text intervention processing on the analyzed original text according to the calling instruction of the total control module, and return an original text intervention result to the total control module, wherein the original text intervention module mainly performs preprocessing on the original text, so that a subsequent translation model can more accurately and smoothly translate the original text. For example, a seller will pile up many words in a product title in order to hit more search keywords, which is not friendly to the translation model and needs to be rewritten and de-duplicated; spelling errors caused by too fast input in product search will also affect the search result after translation, and need to be corrected in advance; many repeated colloquial words in communication will cause the translation model to decode for too long, and need to be normalized in advance. Therefore, in the original text intervention module, functions such as the above-mentioned processing of the original text before decoding are mainly integrated.

[0111] The total control module is further configured to call the translated text intervention module according to the original text intervention result returned by the original text intervention module.

[0112] The translation intervention module is configured to perform translation intervention processing on the original text after original text intervention according to the calling instruction of the total control module, and return the original text intervention result to the total control module; since the training corpus of the translation model cannot cover all vocabularies, and some words will have different translation results in different application scenarios, especially in the e-commerce scenario, if an error translation result appears in the production environment, a mechanism is needed to quickly intervene in the result. Therefore, the main function of the translation intervention module is to ensure that the translation service returns the expected result by manually intervening in the input data in advance; in the translation intervention module, two intervention modes of terms and sentences can be provided to adjust the translation result.

[0113] The total control module is further configured to determine a translation model corresponding to the translation intervention result according to the translation intervention result returned by the translation intervention module, and call a decoder corresponding to the translation model according to the translation model.

[0114] The decoder is configured to perform translation on the original text by using the translation model according to the calling instruction of the total control module; the most core translation model in the decoder is used to decode and translate the input original text into a translation. Different language directions and different types of translation models (such as phrase table models and neural network models) can be loaded in the decoder to support decoding in a batch mode to reduce the corresponding time delay and accelerate the translation decoding operation efficiency.

[0115] The total control module is further configured to perform post-processing on the translation, including translation optimization, case restoration, undecoded word processing, and sensitive word processing.

[0116] Based on Figure 1 The distributed deployment system architecture of the translation system is shown in the figure, Figure 7 The distributed deployment system architecture of the translation system is shown in the figure, Figure 7 As shown in the figure, the deployment controller calls each set of translation service system for translation through an API interface, wherein each set of translation service system corresponds to a translation service application scenario, each set of translation service system shares a total control module, an original text intervention module, and a translation intervention module, but each set of translation service system corresponds to a respective decoder. Since the translation model in each decoder is obtained by using a deep neural network learning mechanism to learn and calculate a translation model corresponding to the respective translation service application scenario.

[0117] The total control module serves as a unified entrance for translation service requests, and is responsible for service calling, integration processing of the original text intervention module, the translation intervention module, and the decoder, and for pre-processing and post-processing of translation results in the translation system.

[0118] The total control module is configured to call a source text intervention module after the source text is parsed;

[0119] The source text intervention module is configured to perform source text intervention processing on the parsed source text according to the calling instruction of the total control module, and return a source text intervention result to the total control module;

[0120] The total control module is further configured to call a translated text intervention module according to the source text intervention result returned by the source text intervention module;

[0121] The translated text intervention module is configured to perform translated text intervention processing on the source text after the source text intervention according to the calling instruction of the total control module, and return a translated text intervention result to the total control module.

[0122] The total control module is further configured to determine a translation service application scenario to which the source text belongs according to the translated text intervention result returned by the translated text intervention module, determine a translation model corresponding to the translation service application scenario according to the translation service application scenario, and call a decoder corresponding to the translation model according to the translation model, wherein the translation model is a translation model corresponding to the translation service application scenario learned and calculated according to the translation service application scenario by using a deep neural network learning mechanism.

[0123] The decoder is configured to perform translated text on the source text by using the translation model according to the calling instruction of the total control module.

[0124] The total control module is further configured to perform post-processing on the translated text, including translated text optimization, case restoration, undecoded word processing, and sensitive word processing.

[0125] The translation system distributedly deployed by the application can create multiple basic translation function modules, can meet various complex translation requirements in multiple scenarios and multiple fields, and on this basis, combines the container management and scheduling capabilities of cloud native, realizes the distributed containerization automatic deployment of various translation function modules in the translation system, can perform fast and continuous delivery and dynamic expansion of overall translation services on the cloud, reduces service deployment cost and time, accelerates iteration efficiency, and shortens product update cycle.

[0126] Figure 8 The server structure diagram provided by an embodiment of the application is shown in FIG. 1, which includes: Figure 8

[0127] a memory, a processor, and a communication component;

[0128] The memory is configured to store a computer program.

[0129] ​The processor, coupled with the memory and the communication component, is configured to execute a computer program for performing the above Figure 2 or Figure 4 the steps or operations of the method embodiments shown in the above

[0130] Further, as shown in the above Figure 8 , the server further includes a display, a power component, an audio component, and other components. Figure 8 Some components are only shown schematically in the above Figure 8 , which does not mean that the server only includes the components shown in the above.

[0131] The server shown in the above embodiment can perform the above Figure 2 or Figure 4 the method embodiments shown in the above, and the implementation principles and technical effects thereof will not be described again.

[0132] Correspondingly, the embodiments of the present application further provide a computer readable storage medium storing a computer program, and the computer program can implement the above Figure 2 or Figure 4 the steps or operations related to the server in the above method embodiments when executed by a computer, which will not be described again herein.

[0133] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the above-mentioned embodiments of the present application have been described in detail, those skilled in the art should understand: it can still modify the technical solutions recorded in the above-mentioned embodiments, or make equivalent replacement for some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for deploying a translation system, characterized in that, The method comprises the following steps: According to the required translation service application scenario, a total control module, an original text intervention module, a translated text intervention module and a decoder corresponding to the translation service application scenario are created, wherein the total control module is used to call the original text intervention module after the original text is parsed, the original text intervention module is used to perform original text intervention processing on the parsed original text according to the calling instruction of the total control module, and the original text intervention result is returned to the total control module, the total control module is also used to call the decoder according to the original text intervention result returned by the original text intervention module, the decoder is used to call the translation model to perform translated text on the original text intervention result according to the calling instruction of the total control module, and the translated text is returned to the total control module, and the total control module is also used to call the translated text intervention module according to the translated text returned by the decoder, the translated text intervention module is used to perform translated text intervention processing on the translated text according to the calling instruction of the total control module, and the translated text intervention result is returned to the total control module; The total control module, the original text intervention module, the translated text intervention module and the decoder are respectively deployed in different containers based on the container technology on the cloud.

2. The method of claim 1, wherein, The method further comprises the following steps: Obtain the information of the translation function module to be created, wherein the information of the translation function module to be created comprises a translation service identifier and a translation function identifier, the translation service identifier represents the translation service application scenario, and the translation function identifier represents the translation function module; According to the translation service identifier and the translation function identifier, obtain the data file of the translation function module to be created; According to the data file, create the image file of the translation function module in the container.

3. The method of claim 2, wherein, The method further comprises the following steps: According to the translation service application scenario, utilize the deep learning mechanism to obtain the data file of each translation function module corresponding to the translation service application scenario, and perform cloud storage on the data file of each translation function module corresponding to the translation service application scenario.

4. The method of claim 3, wherein, The method further comprises the following steps: Obtain the information of the translation function module to be updated, wherein the information of the translation function module to be updated comprises a translation service identifier and a translation function identifier; According to the translation service identifier and the translation function identifier, query the data file of the translation function module to be updated on the cloud storage, and update the data file; According to the updated data file, update the image file of the translation function module in the container.

5. The method of claim 1, wherein, The method is realized by a deployment controller.

6. The method of claim 2, wherein, The method is realized by a deployment controller, and the deployment controller comprises a business logic layer, a data layer and a container processing layer. The business logic layer is used to perform the step of obtaining the information of the translation function module to be created, wherein the information of the translation function module to be created comprises a translation service identifier and a translation function identifier, the translation service identifier represents the translation service application scenario, and the translation function identifier represents the translation function module. The data layer is configured to perform a step of obtaining a data file of the translation function module to be created according to the translation service identifier and the translation function identifier; The container processing layer is configured to perform a step of creating an image file of the translation function module in a container according to the data file.

7. A translation service invocation method characterized by, The translation system corresponding to the translation service application scenario comprises a total control module, an original text intervention module, a translated text intervention module and a decoder, and the total control module, the original text intervention module, the translated text intervention module and the decoder are respectively deployed in different containers, and the translation system comprises: The total control module is configured to call the original text intervention module after analyzing the received original text; The original text intervention module is configured to perform original text intervention processing on the analyzed original text according to the calling instruction of the total control module, and return an original text intervention result to the total control module; The total control module is configured to call a decoder according to the original text intervention result returned by the original text intervention module; The decoder is configured to perform translated text on the original intervention result by calling a translation model according to the calling instruction of the total control module, and return the translated text to the total control module; The total control module is configured to call the translated text intervention module according to the translated text returned by the decoder; The translated text intervention module is configured to perform translated text intervention processing on the translated text according to the calling instruction of the total control module, and return a translated text intervention result to the total control module.

8. The method of claim 7, wherein, The method further comprises: Post-processing the translated text intervention result, wherein the post-processing comprises translated text optimization, case restoration, undecoded word processing and sensitive word processing.

9. The method of claim 8, wherein, The method further comprises: The total control module performs post-processing on the translated text intervention result, including translated text optimization, case restoration, undecoded word processing and sensitive word processing.

10. A server, characterized by It comprises: a memory, a processor and a communication component; The memory is configured to store a computer program; The processor is coupled with the memory and the communication component, and is configured to execute the computer program to perform the steps or operations of the method in any one of claims 1-9.

11. A computer-readable storage medium storing a computer program, characterized in that, The computer program can implement the steps or operations of the method in any one of claims 1-9 when executed by a computer.

Images