CN117540757A - Method for automatic translation, electronic device, and computer-readable storage medium - Google Patents

Method for automatic translation, electronic device, and computer-readable storage medium Download PDF

Info

Publication number
CN117540757A
CN117540757A CN202311568764.6A CN202311568764A CN117540757A CN 117540757 A CN117540757 A CN 117540757A CN 202311568764 A CN202311568764 A CN 202311568764A CN 117540757 A CN117540757 A CN 117540757A
Authority
CN
China
Prior art keywords
translation
text
translated
content
language model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202311568764.6A
Other languages
Chinese (zh)
Inventor
徐冀韬
黄瑾
段亦涛
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Netease Youdao Information Technology Beijing Co Ltd
Original Assignee
Netease Youdao Information Technology Beijing Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Netease Youdao Information Technology Beijing Co Ltd filed Critical Netease Youdao Information Technology Beijing Co Ltd
Priority to CN202311568764.6A priority Critical patent/CN117540757A/en
Publication of CN117540757A publication Critical patent/CN117540757A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/40—Processing or translation of natural language
    • G06F40/58—Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/40—Processing or translation of natural language
    • G06F40/42—Data-driven translation
    • G06F40/49—Data-driven translation using very large corpora, e.g. the web
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Machine Translation (AREA)

Abstract

Embodiments of the present invention provide a method, electronic device, and computer-readable storage medium for automatic translation. Wherein the method comprises the following steps: acquiring reference translation content of a text to be translated; determining task indication information about text to be translated, wherein the task indication information is used for assisting a large language model in learning a thinking chain related to a translation analysis process; and inputting the text to be translated, the reference translation content and the task indication information into a large language model, and analyzing the text to be translated and the reference translation content according to the learned thinking chain through the large language model so as to output a translation result of the text to be translated. By the technical scheme, the invention breaks through the inertia thinking of improving the capability of a large language model through fine adjustment or retraining, and improves the efficiency and the accuracy of automatic translation.

Description

Method for automatic translation, electronic device, and computer-readable storage medium
Technical Field
Embodiments of the present invention relate to the field of machine translation technology, and more particularly, to a method for automatic translation, and an electronic device and a computer-readable storage medium that perform the foregoing method.
Background
This section is intended to provide a background or context to the embodiments of the invention that are recited in the claims. The description herein may include concepts that could be pursued, but are not necessarily ones that have been previously conceived or pursued. Accordingly, unless indicated otherwise, what is described in this section is not prior art to the description and claims in this application and is not admitted to be prior art by inclusion in this section.
At present, products capable of performing machine translation by using a large language model in the market generally need to perform fine adjustment or continuous training on the large language model, and then perform translation by using the trained large language model. The machine translation technology needs a large number of training set supports, and in the actual technology landing process, the problems that training data are difficult to obtain, the training time is too long, a lot of resources are occupied and the like exist, and finally, the training effect of a large language model is poor, so that the machine translation result is not ideal.
Disclosure of Invention
The known effect on machine translation is not ideal, which is a very annoying procedure.
For this reason, an improved scheme for automatic translation is highly needed, which can efficiently and accurately perform automatic translation to improve the machine translation effect.
In this context, embodiments of the present invention desire to provide a method, an electronic device, and a computer-readable storage medium for automatic translation.
In a first aspect of the embodiments of the present invention, a method for automatic translation is presented, comprising: acquiring reference translation content of a text to be translated; determining task instruction information about the text to be translated, wherein the task instruction information is used for assisting a large language model in learning a thinking chain related to a translation analysis process; and inputting the text to be translated, the reference translation content and the task instruction information into the large language model, and analyzing the text to be translated and the reference translation content according to the learned thinking chain through the large language model so as to output a translation result of the text to be translated.
In one embodiment of the invention, obtaining reference translation content for text to be translated includes: and searching translation contents matched with the text to be translated from a translation memory library, and taking the searched translation contents as the reference translation contents.
In another embodiment of the present invention, the translation memory stores translated text and corresponding translation results, and searching translation content matching the text to be translated from the translation memory includes: and searching a translation text matched with the text to be translated and a corresponding translation result from the translation memory based on an editing distance algorithm, wherein the searched translation text and the text to be translated are in the same language, and the translation result corresponding to the searched translation text and the translation result of the text to be translated are in the same language.
In yet another embodiment of the present invention, obtaining reference translation content for text to be translated includes: and acquiring a translation text which is set by a user and is matched with the text to be translated and a corresponding translation result.
In yet another embodiment of the present invention, determining task indication information about the text to be translated includes: acquiring first prompt information about a translation target of the large language model; acquiring second prompt information about the translation analysis process; and determining the task indication information according to the first prompt information, the second prompt information and the reference translation content.
In one embodiment of the present invention, the analyzing the text to be translated and the reference translation content includes: analyzing the reference translation content to determine similar parts in the translation text and the text to be translated; and determining a part to be edited in a translation result corresponding to the translation text.
In another embodiment of the present invention, outputting the translation result of the text to be translated includes: obtaining translation results corresponding to the similar parts; editing the part to be edited to obtain an editing result; and determining the translation result of the text to be translated based on the translation result and the editing result corresponding to the similar part.
In a second aspect of the embodiments of the present invention, there is provided an electronic device, including: a processor; and a memory storing computer instructions for automatic translation, which when executed by the processor, cause the electronic device to perform the method according to the previous and following embodiments.
In a third aspect of embodiments of the present invention, a computer readable storage medium is provided, containing program instructions for automatic translation, which when executed by a processor, cause the implementation of a method according to the foregoing and the following examples.
According to the method, the electronic device and the computer readable storage medium for automatic translation, which are provided by the embodiment of the invention, learning of the thinking chain can be performed by utilizing the acquired auxiliary large language model such as the reference translation content and the task instruction information related to the content to be translated, and the learning thinking chain is utilized to analyze the text to be translated and the reference translation content to obtain a final translation result. It can be seen that in the automatic translation process, the scheme of the invention does not need to finely tune or retrain the large language model additionally, but the task indication information is used for assisting the large language model to learn the thinking chain related to the translation analysis process, so that the analysis capability of the large language model is improved, and the large language model is promoted to accurately compare and analyze the text to be translated and the reference translation content by means of the learned thinking chain, so that the translation result of the text to be translated is obtained. Therefore, the method breaks through the inertia thinking of improving the capacity of the large language model through fine adjustment or retraining, and improves the efficiency and the accuracy of automatic translation.
Drawings
The above, as well as additional purposes, features, and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description when read in conjunction with the accompanying drawings. Several embodiments of the present invention are illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings and in which:
FIG. 1 schematically illustrates a block diagram of an exemplary computing system 100 suitable for implementing embodiments of the invention;
FIG. 2 schematically illustrates a flow diagram of a method for automatic translation according to one embodiment of the invention;
FIG. 3 schematically shows a flow diagram of a method for automatic translation according to another embodiment of the invention;
FIG. 4 schematically shows a flow diagram of a method for automatic translation according to yet another embodiment of the invention;
FIG. 5 schematically shows a comparison of the translation effect of different translation techniques on a test set; and
fig. 6 schematically shows a structural schematic diagram of an electronic device according to an embodiment of the invention.
In the drawings, the same or corresponding reference numerals indicate the same or corresponding parts.
Detailed Description
The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are presented merely to enable those skilled in the art to better understand and practice the invention and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
FIG. 1 illustrates a block diagram of an exemplary computing system 100 suitable for implementing embodiments of the invention. As shown in fig. 1, a computing system 100 may include: a Central Processing Unit (CPU) 101, a Random Access Memory (RAM) 102, a Read Only Memory (ROM) 103, a system bus 104, a hard disk controller 105, a keyboard controller 106, a serial interface controller 107, a parallel interface controller 108, a display controller 109, a hard disk 110, a keyboard 111, a serial peripheral 112, a parallel peripheral 113, and a display 114. Of these devices, coupled to the system bus 104 are a CPU 101, a RAM 102, a ROM 103, a hard disk controller 105, a keyboard controller 106, a serial controller 107, a parallel controller 108, and a display controller 109. The hard disk 110 is coupled to the hard disk controller 105, the keyboard 111 is coupled to the keyboard controller 106, the serial external device 112 is coupled to the serial interface controller 107, the parallel external device 113 is coupled to the parallel interface controller 108, and the display 114 is coupled to the display controller 109. It should be understood that the block diagram depicted in FIG. 1 is for illustrative purposes only and is not intended to limit the scope of the present invention. In some cases, some devices may be added or subtracted as the case may be.
Those skilled in the art will appreciate that embodiments of the invention may be implemented as a system, method, or computer program product. Accordingly, the present disclosure may be embodied in the following forms, namely: all hardware, all software (including firmware, resident software, micro-code, etc.), or a combination of hardware and software, is generally referred to herein as a "circuit," module, "" unit, "or" system. Furthermore, in some embodiments, the invention may also be embodied in the form of a computer program product in one or more computer-readable media, which contain computer-readable program code.
Any combination of one or more computer readable media may be employed. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive example) of the computer-readable storage medium could include, for example: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
The computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, either in baseband or as part of a carrier wave. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, smalltalk, C ++ and conventional procedural programming languages, such as the "C" programming language 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 the case of a remote computer, the remote computer may be connected to the user's 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, through the internet using an internet service provider.
Embodiments of the present invention will be described below with reference to flowchart illustrations of methods and block diagrams of apparatus (or systems) according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable medium that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable medium produce an article of manufacture including instruction means which implement the function/act specified in the flowchart and/or block diagram block or blocks.
The computer program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks.
According to an embodiment of the invention, a method, an electronic device and a computer readable storage medium for automatic translation are provided. Furthermore, any number of elements in the figures is for illustration and not limitation, and any naming is used for distinction only and not for any limiting sense.
The principles and spirit of the present invention are explained in detail below with reference to several representative embodiments thereof.
Summary of The Invention
The inventors have found that the translation effect for machine translation is currently not ideal. In particular, when products capable of performing machine translation based on a large language model on the market are utilized, contents to be translated are generally directly input into the large language model for translation, and the large language model itself lacks analysis capability for some professions or specific fields, so that the translation scenes suitable for the translation products are limited. In contrast, in the related art, fine tuning or continuous training of a large language model is adopted to improve the analysis capability of the large language model in a manner of changing parameters of the large language model. However, the realization of the technology requires a large number of data set supports in the professional field, and the problems of difficult acquisition of the data set, long training time and excessive occupation of resources exist, so that the technology is difficult to land, and the final translation effect cannot be effectively improved.
In this regard, the inventor has found through research that the thinking chain related to the translation analysis process can be learned by using the reference translation content, task instruction information and the like related to the text to be translated to assist the large language model, so that the analysis capability of the large language model is improved by learning the thinking chain, and the efficient and accurate translation of the text to be translated is realized based on the large language model with the improved analysis capability.
Having described the basic principles of the present invention, various non-limiting embodiments of the invention are described in detail below.
Exemplary method
A method for automatic translation according to an exemplary embodiment of the present invention is described below with reference to fig. 2. It should be noted that embodiments of the present invention may be applied to any scenario where applicable.
FIG. 2 schematically illustrates a flow diagram of a method 200 for automatic translation according to one embodiment of the invention.
As shown in fig. 2, at step S201, reference translation content of text to be translated may be acquired. The reference translation content in this embodiment may be understood as translation content highly related to the text to be translated and/or the target text (and the translation result) thereof, and may assist the large language model in learning the translation logic of the reference translation content, and the like. For example, in some embodiments, the language of the text to be translated is a, the language of the translation result is B, and the reference translation content may include the translation result of the reference text in the a language and the reference text in the B language. The reference translation content can assist the large language model in learning translation logic between different languages, etc.
In practical applications, the reference translation content may be obtained in various manners, for example, may be found from some public databases, or may be customized by a user. It should be noted that the detailed description herein with reference to the translation is merely illustrative, and the present invention is not limited thereto. For example, the reference translation content may be related not only to the text to be translated in language but also to the text to be translated semantically, and so on.
At step S202, task instruction information about text to be translated may be determined. The task instruction information can be used for assisting a large language model to learn a thinking chain related to a translation analysis process. For example, the task instruction information may include information such as specific execution tasks and execution logic of the large language model in the translation process, and the large language model may learn the thought chain related to the translation analysis process through learning the task instruction information so as to improve the analysis capability of the large language model.
At step S203, the text to be translated, the reference translation content, and the task instruction information may be input to a large language model, and the foregoing text to be translated and the reference translation content may be analyzed according to the learned thought chain through the large language model, so as to output a translation result of the text to be translated. The large language model in this embodiment may include a commercially available large language model. After the reference translation content and the task indication information are obtained, the reference translation content, the task indication information and the text to be translated can be used as input of the large language model, so that the large language model can learn a thinking chain, and the text to be translated and the reference translation content are efficiently and accurately analyzed based on the learned thinking chain, so that a final translation result is obtained.
According to the scheme, in the automatic translation process, the task indication information can be used for assisting the large language model in learning the thinking chain related to the translation analysis process, so that the analysis capability of the large language model is improved, the large language model is promoted to accurately compare and analyze the text to be translated and the reference translation content by means of the learned thinking chain, and the translation result of the text to be translated is obtained. Therefore, the method breaks through the inertia thinking of improving the capacity of the large language model through fine adjustment or retraining, and improves the efficiency and the accuracy of automatic translation. Compared with the traditional training set, the order of magnitude and the difficulty in acquiring the reference translation content are greatly reduced, and the text to be translated is not excessively limited, so that the scheme of the invention is easy to fall to the ground in implementation, is suitable for any translation scene, and is beneficial to improving the market competitiveness.
Fig. 3 schematically shows a flow diagram of a method 300 for automatic translation according to another embodiment of the invention. It is to be appreciated that the method 300 is a further definition and/or extension of the method 200 of fig. 2. Accordingly, the foregoing detailed description in connection with fig. 2 applies equally as well to the following.
As shown in fig. 3, at step S301, reference translation content of text to be translated may be acquired based on the translation memory. The translation memory (Translation Memory, TM) is a database of computer program software, and typically a translation memory unit includes text segments of a source language and translations thereof, where the text segments may be text blocks, chapters, a sentence or several sentences, or words.
In some embodiments, the translation content matched with the text to be translated can be searched from the translation memory library, and the searched translation content is used as the reference translation content. Specifically, the translated text and the corresponding translation result are stored in the translation memory, and the translated text and the corresponding translation result matched with the text to be translated can be searched from the translation memory based on an edit distance algorithm. The searched translation text and the text to be translated are in the same language, and the translation result corresponding to the searched translation text and the translation result of the text to be translated are in the same language.
It should be noted that, the searching of the reference translation content related to the text to be translated from the translation memory by using the edit distance algorithm is merely illustrative, and the present invention is not limited thereto. For example, the reference translation content may be obtained by other similarity searching methods, or, for example, a user may directly input the customized reference translation content. In addition, regarding the relevant details of the reference translation content, reference may be made to the relevant description in fig. 2, which is not repeated in this example.
After the reference translation content is obtained, task indication information about the text to be translated may then be determined. Specifically, at step S302, first hint information regarding a translation target of a large language model may be acquired. The first prompt information may include prompt information for prompting functions and/or tasks to be executed in the translation process of the large language model. For example, the first prompt information may include a prompt information such as "translate text to be translated from a language to B language". It should be noted that, the detailed description about the first prompt information is merely an exemplary description, and the specific content or the expression form of the first prompt information in the solution of the present invention is not limited.
In this example, it is also necessary to acquire second hint information regarding the translation analysis process, and determine task instruction information based on the first hint information, the second hint information, and the reference translation content. The second hint information may be hint information about the translation analysis logic, that is, information that includes some hint how the large language model performs translation analysis. Based on the first prompt information, the second prompt information and the reference translation content, the translation analysis logic used by the reference translation content can be effectively learned by the large language model, so that the large language model can learn a thinking chain related to the translation analysis process and simulate the thinking habit of people in the translation process.
Finally, at step S303, the text to be translated, the reference translation content, and the task instruction information described above may be input to a large language model, and the text to be translated and the reference translation content may be analyzed by the large language model according to the learned thought chain, so as to output a translation result of the text to be translated.
Fig. 4 schematically shows a flow diagram of a method 400 for automatic translation according to yet another embodiment of the invention. It is to be appreciated that method 400 is a further definition and/or extension of method 200 in fig. 2 and method 300 in fig. 3. Accordingly, the foregoing detailed description in connection with fig. 2 and 3 applies equally as well to the following.
As shown in fig. 4, at step S401, reference translation content of a text to be translated may be acquired, where the reference translation content includes a translation text matching the text to be translated and a translation result corresponding to the translation text. As previously described, the reference translation content may be obtained in a variety of ways. In some embodiments, the translation text matching the text to be translated and the translation result corresponding to the translation text may be looked up by a translation memory. For example, the reference translation content may be looked up from a translation memory by an edit distance algorithm or other similarity algorithm.
In other examples, the translation text and the corresponding translation result set by the user and matched with the text to be translated may also be obtained. In practical application, the reference translation content can be set in a manual self-defining mode according to practical requirements (for example, when the translation memory library cannot find out scenes such as the reference translation content). For example, an input interface of reference translation content may be provided, in which a user can directly input the reference translation content, or the like. It should be noted that, the description of the manual customization process of the reference translation content is merely an exemplary illustration, and the scheme of the present invention is not limited thereto, and the acquisition mode of the reference translation content may be specifically determined through man-machine interaction design.
At step S402, first hint information regarding a translation target of a large language model may be acquired, and second hint information regarding a translation analysis process may be acquired, and task instruction information is determined based on the first hint information, the second hint information, and reference translation content. The first prompt information, the second prompt information, and the task indication information in this embodiment may be described with reference to the related details in fig. 3, which are not described herein.
At step S403, the text to be translated, the reference translation content and the task instruction information may be input to a large language model, and the reference translation content may be analyzed by the large language model according to the learned thought chain to determine a similar portion in the translation text to be translated, and determine a portion to be edited in the translation result corresponding to the translation text. The reference translation content is highly correlated with the text to be translated, the two are analyzed through a large language model, and the similar part and the part to be modified (namely the part to be edited) between the two are determined, so that the translation result of the similar part can be directly multiplexed later, and the resource is saved and the translation time is shortened.
At step S404, a translation result corresponding to the similar portion may be obtained, and the foregoing portion to be edited may be edited (for example, insertion, replacement, and/or deletion may be performed on the portion to be edited) to obtain an edit result, and a translation result of the text to be translated may be determined based on the translation result and the edit result corresponding to the foregoing similar portion. Thereby, an automated translation of the text to be translated is achieved.
The implementation of the steps in fig. 4 is described below with reference to specific examples:
for example, the text to be translated is a section of english "Madam President, honestly, I am one of those who have visited Hong Kong since the transfer" that needs to be translated into french, and the reference translation content of the text to be translated may be determined based on a translation memory library or manual customization, where the reference translation content specifically includes: for example, the translation text of "Mr President, I am also one of those who have an enormously positive view of the building up", and the corresponding translation result "Monsieur le President, moi aussi je fais partie de ceux qui proc. Di t a une. Activating trre positive de ce base.
For the text to be translated, the first prompt information and the second prompt information can be obtained. For example, the first hint may be a hint that "the goal is to obtain a good translation of the source sentence by modifying a given similar French sentence" while the second hint may include "first, analyze the similar sentence to find out which portions of the source sentence are covered by the similar English sentence". Then, a French sentence that needs to be edited is indicated, wherein the editing mode may include insertion, deletion, and substitution. And finally, giving prompt information of translation analysis logic such as final translation.
For the reference translation content, a sample of analysis of the reference translation content according to the translation analysis logic may be provided for learning of the thought chain by the large language model.
And finally, analyzing the text to be translated and the reference translation content according to the learned thinking chain by the large language model, and outputting the translation result of the text to be translated according to the translation result corresponding to the similar part and the editing result of the part to be edited.
The translation effect of the translation scheme in the above embodiment of the present invention and other existing machine translation schemes will be described in comparison with fig. 5. It should be noted that fig. 5 is a diagram of a multi-domain english method (EnFr) data set, and the data set specifically relates to data sets of 11 fields, such as "ECB", "EMEA", "Epps", … and "WiKi". In addition, in this embodiment, the existing translation techniques "large language model zero-shot", "large language model one-shot", the translation technique combining only TM and the translation technique of the present invention are adopted to perform translation tests with respect to the data sets of the above 11 fields, respectively. Where the specific values in FIG. 5 represent BLUE scores, and "All" represents the test scores obtained by concatenating All the data sets. It can be seen that the translation effect of the scheme of the invention is obviously better than that of other technical schemes.
Having described the method of an exemplary embodiment of the present invention, a description of related products for automatic translation of an exemplary embodiment of the present invention follows with reference to fig. 6.
Fig. 6 schematically shows a schematic block diagram of an electronic device 600 according to an embodiment of the invention. As shown in fig. 6, an electronic device 600 may include a processor 601 and a memory 602. Wherein the memory 602 stores computer instructions for automatic translation that, when executed by the processor 601, cause the electronic device 600 to perform the method according to the previous description in connection with fig. 2-4. For example, in some embodiments, the electronic device 600 may obtain reference translation content for the text to be translated, determine task instruction information regarding the text to be translated, analyze the text to be translated and the reference translation content according to a learned chain of thought through a large language model, and so forth. Based on this, the electronic device 600 may assist the large language model to learn the thought chain related to the translation analysis process through the task indication information, thereby improving the analysis capability of the large language model, and causing the large language model to perform accurate comparative analysis on the text to be translated and the reference translation content by means of the learned thought chain, so as to obtain the translation result of the text to be translated. Therefore, the method breaks through the inertia thinking of improving the capacity of the large language model through fine adjustment or retraining, and improves the efficiency and the accuracy of automatic translation.
It should be noted that although several means or sub-means of the device are mentioned in the above detailed description, this division is not mandatory only. Indeed, the features and functions of two or more of the devices described above may be embodied in one device, in accordance with embodiments of the present invention. Conversely, the features and functions of one device described above may be further divided into multiple devices to be embodied.
Use of the verb "comprise," "include" and its conjugations in this application does not exclude the presence of elements or steps other than those stated in the application. The article "a" or "an" preceding an element does not exclude the presence of a plurality of such elements.
While the spirit and principles of the present invention have been described with reference to several particular embodiments, it is to be understood that the invention is not limited to the disclosed embodiments nor does it imply that features of the various aspects are not useful in combination, nor are they useful in any combination, such as for convenience of description. The invention is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims. The scope of the following claims is to be accorded the broadest interpretation so as to encompass all such modifications and equivalent structures and functions.

Claims (9)

1. A method for automatic translation, comprising:
acquiring reference translation content of a text to be translated;
determining task instruction information about the text to be translated, wherein the task instruction information is used for assisting a large language model in learning a thinking chain related to a translation analysis process; and
inputting the text to be translated, the reference translation content and the task indication information into the large language model, and analyzing the text to be translated and the reference translation content according to the learned thinking chain through the large language model so as to output a translation result of the text to be translated.
2. The method of claim 1, wherein obtaining reference translation content for text to be translated comprises:
and searching translation contents matched with the text to be translated from a translation memory library, and taking the searched translation contents as the reference translation contents.
3. The method of claim 2, wherein the translation memory stores translated text and corresponding translation results, and wherein searching translation content matching the text to be translated from the translation memory comprises:
and searching a translation text matched with the text to be translated and a corresponding translation result from the translation memory based on an editing distance algorithm, wherein the searched translation text and the text to be translated are in the same language, and the translation result corresponding to the searched translation text and the translation result of the text to be translated are in the same language.
4. The method of claim 1, wherein obtaining reference translation content for text to be translated comprises:
and acquiring a translation text which is set by a user and is matched with the text to be translated and a corresponding translation result.
5. The method of claim 1, wherein determining task indication information about the text to be translated comprises:
acquiring first prompt information about a translation target of the large language model;
acquiring second prompt information about the translation analysis process; and
and determining the task indication information according to the first prompt information, the second prompt information and the reference translation content.
6. The method according to any one of claims 1 to 5, wherein the reference translation content includes translation text matching the text to be translated and translation results corresponding to the translation text, and analyzing the text to be translated and the reference translation content includes:
analyzing the reference translation content to determine similar parts in the translation text and the text to be translated; and
and determining a part to be edited in a translation result corresponding to the translation text.
7. The method of claim 6, wherein outputting the translation result of the text to be translated comprises:
obtaining translation results corresponding to the similar parts;
editing the part to be edited to obtain an editing result; and
and determining the translation result of the text to be translated based on the translation result corresponding to the similar part and the editing result.
8. An electronic device, comprising:
a processor; and
a memory storing computer instructions for automatic translation, which when executed by the processor, cause the electronic device to perform the method of any of claims 1-7.
9. A computer readable storage medium containing program instructions for automatic translation, which when executed by a processor, cause the method according to any one of claims 1-7 to be implemented.
CN202311568764.6A 2023-11-22 2023-11-22 Method for automatic translation, electronic device, and computer-readable storage medium Pending CN117540757A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311568764.6A CN117540757A (en) 2023-11-22 2023-11-22 Method for automatic translation, electronic device, and computer-readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311568764.6A CN117540757A (en) 2023-11-22 2023-11-22 Method for automatic translation, electronic device, and computer-readable storage medium

Publications (1)

Publication Number Publication Date
CN117540757A true CN117540757A (en) 2024-02-09

Family

ID=89789571

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311568764.6A Pending CN117540757A (en) 2023-11-22 2023-11-22 Method for automatic translation, electronic device, and computer-readable storage medium

Country Status (1)

Country Link
CN (1) CN117540757A (en)

Similar Documents

Publication Publication Date Title
US10664660B2 (en) Method and device for extracting entity relation based on deep learning, and server
CN107908635B (en) Method and device for establishing text classification model and text classification
US10657325B2 (en) Method for parsing query based on artificial intelligence and computer device
US9767095B2 (en) Apparatus, system, and method for computer aided translation
RU2643467C1 (en) Comparison of layout similar documents
US20220318275A1 (en) Search method, electronic device and storage medium
CN108875059B (en) Method and device for generating document tag, electronic equipment and storage medium
US20190095526A1 (en) Keyphrase extraction system and method
CN111143226B (en) Automatic test method and device, computer readable storage medium and electronic equipment
AU2015360997A1 (en) Localization complexity of arbitrary language assets and resources
CN111552766B (en) Using machine learning to characterize reference relationships applied on reference graphs
US10255047B2 (en) Source code analysis and adjustment system
CN110245232B (en) Text classification method, device, medium and computing equipment
CN109710951B (en) Auxiliary translation method, device, equipment and storage medium based on translation history
CN107861948B (en) Label extraction method, device, equipment and medium
CN110609910B (en) Medical knowledge graph construction method and device, storage medium and electronic equipment
US11227119B2 (en) Cognitive word processing
CN109657056B (en) Target sample acquisition method and device, storage medium and electronic equipment
CN111143556A (en) Software function point automatic counting method, device, medium and electronic equipment
US11557381B2 (en) Clinical trial editing using machine learning
EP3404553A1 (en) Open information extraction method and system for extracting reified ternary relationship
CN111240971B (en) Method and device for generating wind control rule test case, server and storage medium
CN112287077A (en) Statement extraction method and device for combining RPA and AI for document, storage medium and electronic equipment
CN108733702B (en) Method, device, electronic equipment and medium for extracting upper and lower relation of user query
US8965750B2 (en) Acquiring accurate machine translation

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination