CN115688808A - Translation method, translation device, readable medium and electronic equipment - Google Patents

Abstract

The embodiment of the disclosure relates to a translation method, a translation device, a readable medium and electronic equipment. The method comprises the following steps: determining a source text to be translated and a target language style; determining a text with an undetermined style from a plurality of styles corresponding to the target language style according to the source text; and inputting the source text and the text with the undetermined style into a pre-generated text translation model to obtain a target translation text output by the text translation model, wherein the language style of the target translation text is a target language style. Therefore, the translation process of the text translation model can be assisted through the text with the undetermined style, so that the text translation model can realize accurate stylized translation without training a large amount of stylized bilingual texts, the training complexity of the text translation model is reduced, and the accuracy of the stylized translation of the text translation model is improved.

Description

Translation method, device, readable medium and electronic equipment

technical field

The present disclosure relates to the field of computer technology, and in particular, to a translation method, device, readable medium and electronic equipment.

Background technique

With the advancement of computer technology, machine translation has become an important research topic in natural language text processing. Machine translation refers to the process of translating a source language text into a semantically equivalent target language text through computers or other electronic devices. Natural language texts can be written in various styles with different vocabularies and syntaxes, while the semantics of the different styles remain the same. Language style plays an important role in communication in many languages, for example, there are American English style and British English style in English, honorific style and non-honorific style in Korean, and so on.

However, in related technologies, machine translation cannot achieve targeted language style after translation, and the language needs to be manually adjusted after the translation is completed, which affects the translation effect and translation efficiency of machine translation.

Contents of the invention

This Summary is provided to introduce a simplified form of concepts that are described in detail later in the Detailed Description. This summary of the invention is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to be used to limit the scope of the claimed technical solution.

According to a first aspect of an embodiment of the present disclosure, there is provided a translation method, the method comprising:

Determine the source text and target language style to be translated;

According to the source text, determine the undetermined style text from a plurality of style texts corresponding to the target language style;

Inputting the source text and the undetermined style text into a pre-generated text translation model to obtain a target translation text output by the text translation model, the language style of the target translation text being the target language style.

According to a second aspect of an embodiment of the present disclosure, there is provided a translation device, the device comprising:

The first determining module is used to determine the source text to be translated and the target language style;

The second determination module is configured to determine a pending style text from a plurality of style texts corresponding to the target language style according to the source text;

A translation module, configured to input the source text and the text of the undetermined style into a pre-generated text translation model to obtain a target translation text output by the text translation model, the language style of the target translation text is the target language style .

According to a third aspect of the embodiments of the present disclosure, there is provided a computer-readable medium on which a computer program is stored, and when the computer program is executed by a processing device, the steps of the method described in the first aspect of the present disclosure are implemented.

According to a fourth aspect of the embodiments of the present disclosure, there is provided an electronic device, including:

a storage device on which a computer program is stored;

A processing device configured to execute the computer program in the storage device to implement the steps of the method described in the first aspect of the present disclosure.

Using the above technical solution, determine the source text and target language style to be translated; determine the undetermined style text from multiple style texts corresponding to the target language style according to the source text; input the source text and undetermined style text into the pre-generated text translation model , to obtain the target translation text output by the text translation model, and the language style of the target translation text is the target language style. In this way, the translation process of the text translation model can be assisted by the undetermined style text, so that the text translation model can achieve accurate stylized translation without training a large number of stylized bilingual texts, which not only reduces the complexity of text translation model training , which improves the accuracy of the text translation model for stylized translation.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure.

Description of drawings

The above and other features, advantages and aspects of the various embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic and that elements and elements are not necessarily drawn to scale. In the attached picture:

Fig. 1 is a flowchart of a translation method according to an exemplary embodiment.

Fig. 2 is a flow chart showing a step of S102 according to the embodiment shown in Fig. 1 .

Fig. 3 is a schematic diagram of a translation method according to an exemplary embodiment.

Fig. 4 is a flowchart showing a method for generating a target style text set according to an exemplary embodiment.

Fig. 5 is a block diagram of a translation device according to an exemplary embodiment.

Fig. 6 is a block diagram of another translation device according to an exemplary embodiment.

Fig. 7 is a block diagram of an electronic device according to an exemplary embodiment.

Detailed ways

Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the drawings, it should be understood that the disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein; A more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are for exemplary purposes only, and are not intended to limit the protection scope of the present disclosure.

It should be understood that the various steps described in the method implementations of the present disclosure may be executed in different orders, and/or executed in parallel. Additionally, method embodiments may include additional steps and/or omit performing illustrated steps. The scope of the present disclosure is not limited in this regard.

As used herein, the term "comprise" and its variations are open-ended, ie "including but not limited to". The term "based on" is "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one further embodiment"; the term "some embodiments" means "at least some embodiments." Relevant definitions of other terms will be given in the description below.

It should be noted that concepts such as "first" and "second" mentioned in this disclosure are only used to distinguish different devices, modules or units, and are not used to limit the sequence of functions performed by these devices, modules or units or interdependence.

It should be noted that the modifications of "one" and "multiple" mentioned in the present disclosure are illustrative and not restrictive, and those skilled in the art should understand that unless the context clearly indicates otherwise, it should be understood as "one or more" multiple". In the description of the present disclosure, unless otherwise stated, "plurality" refers to two or more than two, and other quantifiers are similar; "at least one (one)", "one (one) or more (units)” or similar expressions refer to any combination of these items (units), including any combination of single item (units) or plural items (units). For example, at least one (one) a can represent any number of a; another example, one or more items (ones) in a, b and c can represent: a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, c can be single or multiple; "and/or" is an association relationship describing associated objects, indicating that there can be three relationships, for example, A and/ Or B, can mean: A alone exists, A and B exist at the same time, and B exists alone, where A and B can be singular or plural.

In the embodiments of the present disclosure, although operations or steps are described in a specific order in the drawings, it should not be understood as requiring that these operations or steps be performed in the specific order shown or in a serial order, or that All indicated actions or steps to achieve the desired result. In the embodiments of the present disclosure, these operations or steps may be performed in series; these operations or steps may also be performed in parallel; or a part of these operations or steps may be performed.

The names of messages or information exchanged between multiple devices in the embodiments of the present disclosure are used for illustrative purposes only, and are not used to limit the scope of these messages or information.

It can be understood that before using the technical solutions disclosed in the embodiments of the present disclosure, the user should be informed of the type, scope of use, and use scenarios of the personal information involved in the present disclosure in an appropriate manner in accordance with relevant laws and regulations, and the authorization of the user should be obtained. .

For example, in response to receiving the user's active request, send prompt information to the user to clearly remind the user that the requested operation will require the acquisition and use of the user's personal information. Thus, the user can independently choose whether to provide personal information to software or hardware such as electronic devices, application programs, servers, or storage media that perform the operations of the technical solution of the present disclosure according to the prompt information.

As an optional but non-limiting implementation, in response to receiving the active request from the user, the prompt information may be sent to the user, for example, in the form of a pop-up window, and the prompt information may be presented in text in the pop-up window. In addition, the pop-up window may also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device.

It can be understood that the above process of notifying and obtaining user authorization is only illustrative and does not limit the implementation of the present disclosure. Other methods that meet relevant laws and regulations may also be applied to the implementation of the present disclosure.

At the same time, it can be understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of data) should comply with the requirements of corresponding laws and regulations and relevant regulations.

The present disclosure will be described below in combination with specific embodiments.

First, the application scenarios of the present disclosure are described. The present disclosure can be applied to language translation scenarios, especially stylized language translation scenarios, for example, in the scenario of translating Chinese into American English, or in the scenario of translating English into Korean honorific style.

In order to enable the machine translation model to achieve stylized translation, a large number of stylized bilingual texts (such as text pairs of Chinese text and American English text, text pairs of English text and Korean honorific style text) can be collected, and the machine translation model can be Train to get a stylized translation model. However, using this method, a large amount of stylized bilingual texts needs to be collected, and due to the limited corpus of stylized bilingual texts, the translation accuracy of the trained model is not high.

Fig. 1 is a flowchart of a translation method according to an exemplary embodiment. The method can be applied to electronic equipment, and the electronic equipment can include terminal equipment, such as smart phones, smart wearable devices, smart speakers, smart tablets, PDA (Personal Digital Assistant, personal digital assistant), CPE (Customer PremiseEquipment, customer terminal equipment) ), a personal computer, a vehicle terminal, etc.; the electronic device may also include a server, such as a local server or a cloud server. As shown in Figure 1, the method may include:

S101. Determine the source text to be translated and the target language style.

Wherein, the target language style may be an expected language style after translation of the source text; the source text may be words, sentences, paragraphs or articles, which is not limited in the present disclosure.

The language corresponding to the source text may be called the source language, the expected language to be translated may be called the target language, and the expected language style after translation may be called the target language style. For example, if the source text input by the user is Chinese text, the source language is Chinese, the expected target language after translation may be English, and the expected target language style after translation may be American English.

Wherein, the translated target language and target language style may be specified by the user, or may be automatically detected and determined by the electronic device.

In some embodiments, the text to be translated input by the user may be used as the source text, and the language style input by the user may be used as the target language style.

For example, a language style selection box may be provided to the user, through which the user may select a corresponding language style as the target language style.

In some other embodiments, the text to be translated input by the user may be used as the source text, and the target language style may be automatically determined according to the state parameters of the electronic device used by the user to input the source text.

Wherein, the state parameters may include current time zone parameters, country and language parameters of the electronic device, and the like.

S102. According to the source text, determine a pending style text from a plurality of style texts corresponding to the target language style.

Wherein, the multiple style texts corresponding to the target language style may be pre-generated monolingual style texts. For example, if the target language style is American English style, then the multiple style texts are all American English style texts; if the target language style is Korean honorific style, then the multiple style texts are Korean honorific style of the text.

In some embodiments, the text closest to the source text among the multiple style texts may be used as the undetermined style text. For example, the style text and the source text can be coded separately by a pre-generated text encoding model, and the similarity between the source text and each style text can be calculated, and the style text with the highest similarity can be used as the undetermined style text.

In other embodiments, a style text may be randomly selected from multiple style texts as the pending style text.

S103. Input the source text and the undetermined style text into the pre-generated text translation model to obtain the target translation text output by the text translation model.

Wherein, the language style of the target translation text may be the target language style.

In some embodiments, the target spliced text can be obtained after splicing the source text and the undetermined style text, and input the target spliced text into the text translation model to obtain the target translated text.

For example, the source text and the undetermined style text can be spliced according to preset keywords, which can be spliced into undetermined style text + preset keywords + source text, or can also be spliced into source text + preset keywords + undetermined style text.

For example, the preset keyword can be <token>, if the source text is "I don't know, he won't tell me", the pending style text is "I tell thee, Kate, 'twas burnt and dried away" , the target spliced text obtained after splicing can be "I tell thee, Kate, 'twas burnt and dried away <token> I don't know, and he won't tell me", or "I don't know, and he won't Tell me <token>I tell thee, Kate, 'twas burnt and dried away".

In some other embodiments, the source text and the undetermined style text may be respectively input into the above-mentioned text translation model to obtain the target translation text.

In some embodiments, the undetermined style text can be used as a prompt monolingual corresponding to the source text, and is used to prompt the text translation model, so that the style text or the target language style corresponding to the style text is retained in the translated target translation text , that is, make the language style of the output target translation text the target language style.

In some other embodiments, the text translation model may determine the target language style according to the pending style text, so that the speech style of the translated target translated text is the same as or similar to the target language style.

It should be noted that the above-mentioned text translation model may adopt a machine translation model in the related art, for example, a Transformer model or a BERT model, which is not limited in the present disclosure.

Using the above method, determine the source text to be translated and the target language style; determine the undetermined style text from multiple style texts corresponding to the target language style according to the source text; input the source text and the undetermined style text into the pre-generated text translation model, The target translation text output by the text translation model is obtained, and the language style of the target translation text is the target language style. In this way, the translation process of the text translation model can be assisted by the undetermined style text, so that the text translation model can achieve accurate stylized translation without training a large number of stylized bilingual texts, which not only reduces the complexity of text translation model training , which improves the accuracy of the text translation model for stylized translation.

Fig. 2 is a flow chart showing a step of S102 according to the embodiment shown in Fig. 1 . As shown in Figure 2, the above S102 step may include the following sub-steps:

S1021. Determine a target style text set corresponding to the target language style.

Wherein, the target style text set may be a pre-generated text set, the target style text set may include multiple style texts, and the first vector corresponding to each style text, and different language styles may correspond to different style text sets, The language styles corresponding to the style texts in the same style text set can be the same.

For example, when the target language style is American English, the target style text set may be an American English style text set.

S1022. Input the source text into the pre-generated multilingual encoding model to obtain a second vector output by the multilingual encoding model.

It should be noted that the multilingual coding model may be developed based on XLM-Roberta, but this embodiment is not limited to the model network structure of XLM-Roberta, and may also be other neural network structures. Among them, XLM-Roberta is a typical multilingual pre-training model, which is a converter-based language model, which relies on the mask language model as the target, and can process text in more than 100 different languages.

In some embodiments, the second vector may be a 512-dimensional or 768-dimensional vector.

In some embodiments, the above-mentioned first vector may also be a vector obtained by encoding the style text through the same multilingual encoding model.

S1023. Determine the undetermined style text from the target style text set according to the second vector and the first vector.

In some embodiments, from the first vector of the target style text set, determine the third vector with the smallest vector distance (that is, the highest similarity) with the second vector; and use the style text corresponding to the third vector as an undetermined style text.

In some other embodiments, the third vector with the smallest vector distance from the second vector can be determined from the first vector of the target style text set; when the vector distance between the third vector and the second vector is less than or equal to the preset In the case of the distance threshold, the style text corresponding to the third vector is used as the undetermined style text. The preset distance threshold may be a preset threshold.

In this way, the pending style text can be determined from multiple style texts corresponding to the target language style according to the source text.

In this way, cross-language retrieval can be performed through the multilingual coding model to obtain the undetermined style text corresponding to the source text, and the retrieved undetermined style text can be spliced with the source text, so that the language style of the target translation text output by the text translation model can be intervened .

Fig. 3 is a schematic diagram of a translation method according to an exemplary embodiment. In Fig. 3, three source texts are taken as examples to illustrate the data flow of the translation method.

First, identify the source text to be translated.

Taking Figure 3 as an example, the three source texts can include: source text Case1 is "I don't know, and he won't tell me.", source text Case2 is "a woman more beautiful than the person I love?", The source text of Case3 is "Now I'll tell you so you don't have to ask.".

Secondly, the three source texts are input into the pre-generated multilingual encoding model to obtain a second vector output by the multilingual encoding model for each source text.

For example, the second vector corresponding to the source text Case1 is [0.01,0.02,-0.03,…,0.05,0.37]; the second vector corresponding to the source text Case2 is [0.09,0.04,-0.01,…,0.17,0.07] ; The second vector corresponding to the source text Case3 is [-0.01,0.07,0.05,...,0.23,0.02].

Again, for each source text, according to the second vector, determine the undetermined style text corresponding to the source text from the target style text set.

Wherein, the vector distance between the first vector and the second vector corresponding to the undetermined style text is the smallest (that is, the similarity is the highest).

For example, the vector distance between the source text Case1 and the top1 neighbor vector is 20; the undetermined style text 1 corresponding to the source text Case1 is "I tell thee, Kate, 'twas burnt and dried away."; the source text Case2 and the top1 neighbor vector The vector distance is 40; the undetermined style text 2 corresponding to the source text Case2 is "And I the king shall love thee."; the vector distance between the source text Case3 and the top1 neighbor vector is 30; the undetermined style text 3 corresponding to the source text Case3 is "Now Let me see if I can conster it.".

Then, the undetermined style text is spliced with the source text to obtain the target spliced text.

For example, the target splicing text 1 corresponding to the source text Case1 is: "I tell thee, Kate, 'twas burnt and dried away. <token> I don't know, and he won't tell me."; the target splicing corresponding to the source text Case2 Text 2 is: "And I the king shall love thee. <token> a woman more beautiful than the person I love?"; the target spliced text 3 corresponding to the source text Case3 is: "Now let me see if I can conster it .<token>I'll tell you now so you don't have to ask."

Finally, the target spliced text is input into the text translation model to obtain the target translated text output by the text translation model.

In this way, the language style of the target translation text output by the text translation model can be intervened by splicing the undetermined style text and the source text to achieve accurate stylized translation.

In some embodiments of the present disclosure, the above-mentioned target style text set may be a pre-generated text set. Fig. 4 is a flowchart showing a method for generating a target style text set according to an exemplary embodiment. As shown in Figure 4, this target style text set can be pre-generated by:

S301. Acquire multiple style texts corresponding to the target language style.

In this step, multiple style texts of the target language style can be manually collected, or multiple style texts corresponding to the target language style can be obtained from the Internet.

S302. For each style text, input the style text into the multilingual encoding model to obtain a first vector output by the multilingual encoding model.

Likewise, the multilingual coding model may be a model developed based on XLM-Roberta, and the first vector may also be a 512-dimensional or 768-dimensional vector.

In some embodiments, the dimensions of the first vector and the second vector may be the same, for example, both are 768 dimensions.

S303. Generate a target style text set according to the style text and the first vector.

In some embodiments, the target style text set may be generated based on a preset retrieval algorithm according to the style text and the first vector. The preset retrieval algorithm can realize the retrieval of text vectors, for example, it can be an ANN (Approximate Nearest Neighbor, approximate nearest neighbor) algorithm, or it can be a text vector retrieval engine in the related art, which is not limited in the present disclosure.

Exemplarily, a search pair (pair) can be generated according to each style text and the first vector corresponding to the style text, for example, the search pair can be (style text, first vector), or the search pair can also be (first vector, style text). The retrieval can be trained on an input text vector retrieval engine through an ANN (Approximate Nearest Neighbor, approximate nearest neighbor) algorithm to generate a target style text set. The target style text set may be a retrieval library, the style text may be used as the data (Value) of the retrieval database, and the first vector may be used as the retrieval keyword (Key) of the retrieval database.

In this way, the above-mentioned target style text set can be generated in advance in this manner.

In some embodiments of the present disclosure, the above-mentioned target style text set can be used for training a text translation model. For example, the text translation model is pre-generated in the following manner:

First, obtain translation training samples.

Wherein, the translation training sample includes a plurality of source sample texts, and a target sample text corresponding to each source sample text.

Secondly, according to the source sample text, the style sample text corresponding to the source sample text is determined from the target style text set.

In some embodiments, the source sample text can be input into the above-mentioned multilingual encoding model to obtain the third vector output by the multilingual encoding model; according to the third vector and the first vector in the target style text set, from the target style text set Determine the style sample text.

Finally, according to the style sample text, the source sample text and the target sample text, the preset translation model is trained to obtain a text translation model.

For example, the style sample text and the source sample text can be spliced to obtain the spliced source sample text, and the spliced source sample text and the target sample text can be used as a new translation training sample to train the preset translation model to obtain the target translation model .

In this way, the text translation model can be trained according to the target style text set, thereby further improving the accuracy of the text translation model.

The translation method provided in some embodiments of the present disclosure can be divided into two stages: offline and online. For example, the above-mentioned target style text set and the above-mentioned text translation model can be generated offline first; then according to the target style text set and text translation The model is translated online, so that the delay in the online translation stage can be reduced and the translation efficiency can be improved.

Furthermore, by adopting the method in the embodiment of the present disclosure, when a new language style is generated, the style text set corresponding to the language style can be generated offline, and the translation style of the text translation model can be intervened, and the text translation model can not need to Collect a large number of samples for training, thus realizing the ability of zero-sample learning.

Fig. 4 is a block diagram of a translation device 400 according to an exemplary embodiment. As shown in Fig. 4, the device 400 may include:

The first determination module 401 is used to determine the source text to be translated and the target language style;

The second determination module 402 is configured to determine a pending style text from a plurality of style texts corresponding to the target language style according to the source text;

A translation module 403, configured to input the source text and the text of the undetermined style into a pre-generated text translation model to obtain a target translation text output by the text translation model, the language style of the target translation text being the target language style.

According to one or more embodiments of the present disclosure, the second determining module 402 is configured to determine a target style text set corresponding to the target language style; the target style text set is a pre-generated text set, and the target The style text set includes a plurality of style texts, and the first vector corresponding to each style text; different language styles correspond to different style text sets, and the style texts in the same style text set correspond to the same language style; The source text is input into the pre-generated multilingual encoding model, and the second vector output by the multilingual encoding model is obtained; according to the second vector and the first vector, the undetermined style text is determined from the target style text set .

According to one or more embodiments of the present disclosure, the second determination module 402 is configured to determine, from the first vectors of the target style text set, a third vector with the smallest vector distance to the second vector; The style text corresponding to the third vector is used as the undetermined style text.

Fig. 5 is a block diagram of another translation device according to an exemplary embodiment. As shown in Fig. 5, the device 400 may include:

A generating module 404, configured to acquire a plurality of style texts corresponding to the target language style; for each style text, input the style text into the multilingual coding model to obtain the multilingual coding model Output the first vector; generate the target style text set according to the style text and the first vector.

According to one or more embodiments of the present disclosure, the generating module 404 is configured to generate the target style text set based on a preset retrieval algorithm according to the style text and the first vector.

According to one or more embodiments of the present disclosure, the generating module 404 is further configured to obtain translation training samples; the translation training samples include multiple source sample texts, and target sample texts corresponding to each source sample text; according to For the source sample text, determine the style sample text corresponding to the source sample text from the target style text set; perform a preset translation model according to the style sample text, the source sample text, and the target sample text Train to obtain the text translation model.

According to one or more embodiments of the present disclosure, the translation module 403 is configured to obtain a target spliced text after splicing the source text and the undetermined style text; input the target spliced text into the text translation model to obtain the target translation text.

Regarding the apparatus in the foregoing embodiments, the specific manner in which each module executes operations has been described in detail in the embodiments related to the method, and will not be described in detail here.

Referring now to FIG. 6 , it shows a schematic structural diagram of an electronic device 2000 (such as a terminal device or a server) suitable for implementing the embodiments of the present disclosure. The terminal equipment in the embodiment of the present disclosure may include but not limited to such as mobile phone, notebook computer, digital broadcast receiver, PDA (personal digital assistant), PAD (tablet computer), PMP (portable multimedia player), vehicle terminal (such as mobile terminals such as car navigation terminals) and fixed terminals such as digital TVs, desktop computers and the like. Servers in the embodiments of the present disclosure may include, but are not limited to, local servers, cloud servers, single servers, distributed servers, and the like. The electronic device shown in FIG. 6 is only an example, and should not limit the functions and application scope of the embodiments of the present disclosure.

As shown in FIG. 6, an electronic device 2000 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 2001, which may be randomly accessed according to a program stored in a read-only memory (ROM) 2002 or loaded from a storage device 2008. Various appropriate actions and processes are executed by programs in the memory (RAM) 2003 . In the RAM 2003, various programs and data necessary for the operation of the electronic device 2000 are also stored. The processing device 2001 , ROM 2002 , and RAM 2003 are connected to each other via a bus 2004 . An input/output (I/O) interface 2005 is also connected to the bus 2004 .

Typically, the following devices can be connected to the input/output interface 2005: input devices 2006 including, for example, a touch screen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; including, for example, a liquid crystal display (LCD), speaker, vibration an output device 2007 such as a computer; a storage device 2008 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 2009. The communication means 2009 may allow the electronic device 2000 to perform wireless or wired communication with other devices to exchange data. While FIG. 6 shows electronic device 2000 having various means, it should be understood that implementing or having all of the means shown is not a requirement. More or fewer means may alternatively be implemented or provided.

In particular, according to an embodiment of the present disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product, which includes a computer program carried on a non-transitory computer readable medium, where the computer program includes program code for executing the method shown in the flowchart. In such an embodiment, the computer program may be downloaded and installed from the network via the communication means 2009, or installed from the storage means 2008, or installed from the ROM 2002. When the computer program is executed by the processing device 2001, the above-mentioned functions defined in the methods of the embodiments of the present disclosure are executed.

It should be noted that the above-mentioned computer-readable medium in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. A computer readable storage medium may include, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, electrical connections with one or more wires, portable computer diskettes, hard disks, random access memory (RAM), read-only memory (ROM), erasable Programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present disclosure, however, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave carrying computer-readable program code therein. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. The computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, and the computer-readable signal medium may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device . Program code embodied on a computer readable medium may be transmitted by any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

In some embodiments, the client and the server can communicate using any currently known or future-developed network protocols such as HTTP (HyperText Transfer Protocol, Hypertext Transfer Protocol), and can communicate with digital data in any form or medium (eg, communication network) interconnections. Examples of communication networks include local area networks ("LANs"), wide area networks ("WANs"), internetworks (e.g., the Internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed network of.

The above-mentioned computer-readable medium may be included in the above-mentioned electronic device, or may exist independently without being incorporated into the electronic device.

The above-mentioned computer-readable medium carries one or more programs, and when the above-mentioned one or more programs are executed by the electronic device, the electronic device: determines the source text to be translated and the target language style; according to the source text, from Determining the undetermined style text among the plurality of style texts corresponding to the target language style; inputting the source text and the undetermined style text into a pre-generated text translation model to obtain the target translation text output by the text translation model, the The language style of the target translation text is the target language style.

Computer program code for carrying out operations of the present disclosure may be written in one or more programming languages, or combinations thereof, including but not limited to object-oriented programming languages—such as Java, Smalltalk, C++, and Includes conventional procedural programming languages - such as "C" or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (for example, using an Internet service provider to connected via the Internet).

The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in a flowchart or block diagram may represent a module, program segment, or portion of code that contains one or more logical functions for implementing specified executable instructions. It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or they may sometimes be executed in the reverse order, depending upon the functionality involved. It should also be noted that each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart illustrations, can be implemented by a dedicated hardware-based system that performs the specified functions or operations , or may be implemented by a combination of dedicated hardware and computer instructions.

The modules involved in the embodiments described in the present disclosure may be implemented by software or by hardware. Wherein, the name of the module does not constitute a limitation of the module itself under certain circumstances, for example, the first determination module may also be described as "a module for determining the style of the source text to be translated and the target language".

The functions described herein above may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: Field Programmable Gate Arrays (FPGAs), Application Specific Integrated Circuits (ASICs), Application Specific Standard Products (ASSPs), System on Chips (SOCs), Complex Programmable Logical device (CPLD) and so on.

In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media would include one or more wire-based electrical connections, portable computer discs, hard drives, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, compact disk read only memory (CD-ROM), optical storage, magnetic storage, or any suitable combination of the foregoing.

According to one or more embodiments of the present disclosure, there is provided a translation method, the method comprising:

Determine the source text and target language style to be translated;

According to the source text, determine the undetermined style text from a plurality of style texts corresponding to the target language style;

Inputting the source text and the undetermined style text into a pre-generated text translation model to obtain a target translation text output by the text translation model, the language style of the target translation text being the target language style.

According to one or more embodiments of the present disclosure, according to the source text, determining the undetermined style text from the multiple style texts corresponding to the target language style includes:

Determine the target style text set corresponding to the target language style; the target style text set is a pre-generated text set, the target style text set includes a plurality of style texts, and each style text corresponds to the first One vector; different language styles correspond to different style text sets, and the style texts in the same style text set correspond to the same language style;

inputting the source text into a pre-generated multilingual coding model to obtain a second vector output by the multilingual coding model;

According to the second vector and the first vector, a pending style text is determined from the target style text set.

According to one or more embodiments of the present disclosure, the determining the pending style text from the target style text set according to the second vector and the first vector includes:

From the first vectors of the target style text set, determine a third vector with the smallest vector distance from the second vector;

The style text corresponding to the third vector is used as the undetermined style text.

According to one or more embodiments of the present disclosure, the target style text set is pre-generated in the following manner:

Obtaining multiple style texts corresponding to the target language style;

For each style text, input the style text into the multilingual encoding model to obtain a first vector output by the multilingual encoding model;

Generate the target style text set according to the style text and the first vector.

According to one or more embodiments of the present disclosure, the generating the target style text set according to the style text and the first vector includes:

According to the style text and the first vector, the target style text set is generated based on a preset retrieval algorithm.

According to one or more embodiments of the present disclosure, the text translation model is pre-generated in the following manner:

Obtain translation training samples; the translation training samples include a plurality of source sample texts, and target sample texts corresponding to each source sample text;

determining the style sample text corresponding to the source sample text from the target style text set according to the source sample text;

According to the style sample text, the source sample text and the target sample text, the preset translation model is trained to obtain the text translation model.

According to one or more embodiments of the present disclosure, the input of the source text and the undetermined style text into a pre-generated text translation model, and obtaining the target translation text output by the text translation model includes:

After splicing the source text and the undetermined style text, the target spliced text is obtained;

Inputting the target spliced text into the text translation model to obtain the target translated text.

According to one or more embodiments of the present disclosure, there is provided a translation device, the device comprising:

The first determining module is used to determine the source text to be translated and the target language style;

The second determination module is configured to determine a pending style text from a plurality of style texts corresponding to the target language style according to the source text;

A translation module, configured to input the source text and the text of the undetermined style into a pre-generated text translation model to obtain a target translation text output by the text translation model, the language style of the target translation text is the target language style .

According to one or more embodiments of the present disclosure, the second determining module is configured to determine a target style text set corresponding to the target language style; the target style text set is a pre-generated text set, and the target style The text set includes a plurality of style texts, and the first vector corresponding to each style text; different language styles correspond to different style text sets, and the style texts in the same style text set correspond to the same language style; The source text is input into the pre-generated multilingual encoding model, and the second vector output by the multilingual encoding model is obtained; according to the second vector and the first vector, the undetermined style text is determined from the target style text set.

According to one or more embodiments of the present disclosure, the second determination module is configured to determine, from the first vectors of the target style text set, a third vector with the smallest vector distance to the second vector; The style text corresponding to the third vector is used as the pending style text.

According to one or more embodiments of the present disclosure, the device further includes:

A generating module, configured to obtain a plurality of style texts corresponding to the target language style; for each style text, input the style text into the multilingual coding model to obtain the output of the multilingual coding model the first vector; generating the target style text set according to the style text and the first vector.

According to one or more embodiments of the present disclosure, the generating module is configured to generate the target style text set based on a preset retrieval algorithm according to the style text and the first vector.

According to one or more embodiments of the present disclosure, the generating module is further configured to obtain translation training samples; the translation training samples include multiple source sample texts, and target sample texts corresponding to each source sample text; according to the The source sample text, determine the style sample text corresponding to the source sample text from the target style text set; train the preset translation model according to the style sample text, the source sample text and the target sample text , to obtain the text translation model.

According to one or more embodiments of the present disclosure, the translation module is configured to obtain a target spliced text after splicing the source text and the undetermined style text; input the target spliced text into the text translation model , to get the target translation text.

The above description is only a preferred embodiment of the present disclosure and an illustration of the applied technical principle. Those skilled in the art should understand that the disclosure scope involved in this disclosure is not limited to the technical solution formed by the specific combination of the above-mentioned technical features, but also covers the technical solutions formed by the above-mentioned technical features or Other technical solutions formed by any combination of equivalent features. For example, a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features with similar functions disclosed in this disclosure.

In addition, while operations are depicted in a particular order, this should not be understood as requiring that the operations be performed in the particular order shown or performed in sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Likewise, while the above discussion contains several specific implementation details, these should not be construed as limitations on the scope of the disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination.

Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are merely example forms of implementing the claims. Regarding the apparatus in the foregoing embodiments, the specific manner in which each module executes operations has been described in detail in the embodiments related to the method, and will not be described in detail here.

Claims (10)

1. A translation method, characterized in that the method comprises: Determine the source text and target language style to be translated; According to the source text, determine the undetermined style text from a plurality of style texts corresponding to the target language style; Inputting the source text and the undetermined style text into a pre-generated text translation model to obtain a target translation text output by the text translation model, the language style of the target translation text being the target language style. 2. The method according to claim 1, wherein, according to the source text, determining the undetermined style text from a plurality of style texts corresponding to the target language style comprises: Determine the target style text set corresponding to the target language style; the target style text set is a pre-generated text set, the target style text set includes a plurality of style texts, and each style text corresponds to the first One vector; different language styles correspond to different style text sets, and the style texts in the same style text set correspond to the same language style; inputting the source text into a pre-generated multilingual coding model to obtain a second vector output by the multilingual coding model; According to the second vector and the first vector, a pending style text is determined from the target style text set. 3. The method according to claim 2, wherein, according to the second vector and the first vector, determining the style text to be determined from the target style text set comprises: From the first vectors of the target style text set, determine a third vector with the smallest vector distance from the second vector; The style text corresponding to the third vector is used as the undetermined style text. 4. The method according to claim 2, wherein the target style text set is pre-generated in the following manner: Obtaining multiple style texts corresponding to the target language style; For each style text, input the style text into the multilingual encoding model to obtain a first vector output by the multilingual encoding model; Generate the target style text set according to the style text and the first vector. 5. The method according to claim 4, wherein, according to the style text and the first vector, generating the target style text set comprises: According to the style text and the first vector, the target style text set is generated based on a preset retrieval algorithm. 6. The method according to claim 2, wherein the text translation model is pre-generated in the following manner: Obtain translation training samples; the translation training samples include a plurality of source sample texts, and target sample texts corresponding to each source sample text; determining the style sample text corresponding to the source sample text from the target style text set according to the source sample text; According to the style sample text, the source sample text and the target sample text, the preset translation model is trained to obtain the text translation model. 7. The method according to any one of claims 1 to 6, wherein the input of the source text and the undetermined style text into a pre-generated text translation model obtains the output of the text translation model Target translated texts include: After splicing the source text and the undetermined style text, the target spliced text is obtained; Inputting the target spliced text into the text translation model to obtain the target translated text. 8. A translation device, characterized in that the device comprises: The first determining module is used to determine the source text to be translated and the target language style; The second determination module is configured to determine a pending style text from a plurality of style texts corresponding to the target language style according to the source text; A translation module, configured to input the source text and the text of the undetermined style into a pre-generated text translation model to obtain a target translation text output by the text translation model, the language style of the target translation text is the target language style . 9. A computer-readable medium, on which a computer program is stored, characterized in that, when the computer program is executed by a processing device, the steps of the method according to any one of claims 1 to 7 are implemented. 10. An electronic device, characterized in that it comprises: a storage device on which a computer program is stored; A processing device configured to execute the computer program in the storage device to implement the steps of the method according to any one of claims 1-7.

Images