KR20220067371A - Translation method - Google Patents

Abstract

According to one embodiment of the present invention, disclosed is a translation method by a processor of a computing device, which can increase productivity of machine translation. The method may include: inputting a first sentence to a machine translation model trained by using pair data consisting of a pair of start sentence data and arrival sentence data to obtain a translated second sentence; inputting the second sentence to a grammar correction model trained by using a learning data set consisting of non-text data and plain text data labeled on the non-text data to obtain a corrected third sentence; and obtaining a fourth sentence, which is a final translation sentence for the first sentence, by performing an additional correction operation on the third sentence, using a preset rule.

Description

TRANSLATION METHOD

The present disclosure relates to a translation method, and more particularly, to a method for a processor of a computing device to translate using an artificial neural network-based translation model.

Neural Machine Translation (NMT) has a method based on an encoder-decoder mechanism. The method based on the encoder-decoder mechanism uses a Recurrent Neural Network with Long Short Term Memory (RNN-LSTM) or Convolutional Neural Network (CNN), Artificial neural network learning that compresses (or abstracts) the input text of the . Prediction Mechanism. A lot of research is being done on the neural network structure and learning method to implement this mechanism. Also, neural network-based machine translation technology is currently known as a core technology of services provided by companies such as Google and Naver.

On the other hand, the performance of the translator improved as the neural network-based machine translation technology developed and the amount of available learning data increased. However, grammatical errors or some words were omitted/added due to the black box characteristics of the artificial neural network. Errors, etc., occur repeatedly.

In order to improve this, there are currently two aspects of method being used. First, we are trying to improve the performance of machine translation itself. Specifically, in order to improve the quality of the existing machine translation results, techniques such as modifying the learning model and adding data are used. However, this cannot guarantee performance improvement compared to investment time and cost, and the performance improvement range is small. Second, there is a way to directly modify the translation result. A method of automatically correcting the results of machine translation as a rule-base or entrusting the correction to a human translator is being used. It is also assisted by a module that corrects grammar and a device that detects translation errors.

However, in order to obtain a translation result through these, two separate steps of translation and proofing are required, so there is a limit to productivity improvement, and consumption of human resources may be inevitable.

Therefore, there is a need for research and development of a translation method that can overcome the limitations of the existing translation method.

Republic of Korea Patent Publication No. 2000-0031651

The present disclosure has been devised in response to the above-described background art, and is intended to increase the productivity of machine translation by performing more accurate translation using a computing device.

The technical problems of the present disclosure are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the following description.

According to some embodiments of the present disclosure for solving the above-described problems, a method for a processor of a computing device to translate is disclosed. The method includes: inputting a first sentence into a machine translation model learned using pair data consisting of a pair of starting sentence data and arriving sentence data to obtain a translated second sentence; obtaining a corrected third sentence by inputting the second sentence into a grammar correction model learned using a learning dataset composed of inscription data and plaintext data labeled with the inscription data; and obtaining a fourth sentence that is a final translated sentence for the first sentence through an additional proofreading operation for the third sentence by using a preset rule.

In addition, the first sentence may be a text preset from the translation target sentence through a pre-processing process or a sentence in which special symbols are removed.

In addition, the departure sentence data is data in which a label indicating the type of the second language of the arrival sentence is added to the front of the departure sentence written in the first language through the preprocessing process, and the arrival sentence data is It may be data in which a label indicating the type of the first language is added to the front of the arrival sentence written in the second language.

In addition, the inscription data is sentence data including at least one of an article error, a preposition error, a position error of a number, a sentence structure error, or a spacing error, and the plaintext data is a sentence in which errors included in the inscription data are corrected. It can be data.

In addition, the step of obtaining a corrected third sentence by inputting the second sentence into a grammar correction model learned using a learning dataset consisting of the inscription data and the plaintext data labeled in the inscription data includes: when a line break symbol exists in a sentence, removing the line break symbol and storing location information of the line break symbol in a memory; and obtaining the third sentence by inputting the second sentence from which the newline symbol has been removed into the grammar correction model.

In addition, the step of obtaining a fourth sentence that is a final translated sentence for the first sentence through an additional proofing operation for the third sentence using the preset rule may include: , recognizing whether the insertion position of the line break symbol in the third sentence exists within a word; and when it is recognized that the insertion position exists within the word, inserting the newline symbol in front of the word to obtain the fourth sentence.

In addition, the step of obtaining a corrected third sentence by inputting the second sentence into a grammar correction model learned using a learning dataset consisting of the inscription data and the plaintext data labeled in the inscription data includes: comparing the sentence with the second sentence and recognizing whether there is a number lost in the translation process through the machine translation model; adding the missing number to the end of the second sentence when it is recognized that there is the missing number; and obtaining the third sentence by inputting the second sentence in which the missing number is added to the end of the sentence into the grammar correction model.

In addition, the step of obtaining the third sentence by inputting the second sentence to which the lost number is added to the grammar correction model may include: the third sentence in which the lost number is restored from the grammar correction model to an appropriate position may include; obtaining

In addition, the step of obtaining a fourth sentence that is a final translated sentence for the first sentence through an additional proofing operation on the third sentence using the preset rule includes: among the punctuation marks included in the third sentence and obtaining the fourth sentence by inserting a newline symbol after a punctuation mark related to the end of the sentence.

In addition, the step of using the preset rule to obtain a fourth sentence that is a final translation sentence for the first sentence through an additional proofing operation for the third sentence may include, based on the preset rule, It may include; adding a space to the third sentence to obtain the fourth sentence.

According to some embodiments of the present disclosure for solving the problems as described above, a computer program stored in a computer-readable storage medium is disclosed. The computer program includes instructions for causing a processor of a computing device to perform the following steps: A machine translation model learned using pair data consisting of pairs of starting sentence data and arrival sentence data. inputting the first sentence to to obtain a translated second sentence; obtaining a corrected third sentence by inputting the second sentence into a grammar correction model learned using a learning dataset composed of inscription data and plaintext data labeled with the inscription data; and obtaining a fourth sentence that is a final translated sentence for the first sentence through an additional proofreading operation for the third sentence by using a preset rule.

The technical solutions obtainable in the present disclosure are not limited to the above-mentioned solutions, and other solutions not mentioned are clearly to those of ordinary skill in the art to which the present disclosure belongs from the description below. can be understood

The present disclosure provides a translation method in which a computing device can perform more accurate translation, thereby increasing productivity of machine translation.

Effects obtainable in the present disclosure are not limited to the above-mentioned effects, and other effects not mentioned will be clearly understood by those of ordinary skill in the art to which the present disclosure belongs from the description below. .

Various aspects are now described with reference to the drawings, wherein like reference numbers are used to refer to like elements collectively. In the following example, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of one or more aspects. It will be evident, however, that such aspect(s) may be practiced without these specific details.
1 is a block diagram of a computing device performing translation according to some embodiments of the present disclosure.
2 is a flowchart illustrating an example of a translation method according to some embodiments of the present disclosure.
3 is a view for explaining an example of a training dataset for learning a grammar correction model according to some embodiments of the present disclosure.
4 is a flowchart illustrating an example of a translation process according to some embodiments of the present disclosure.
5 is a flowchart illustrating another example of a translation process according to some embodiments of the present disclosure.
6 is a view for explaining an example of each of the translation process according to some embodiments of the present disclosure.
7 depicts a simplified, general schematic diagram of an exemplary computing environment in which embodiments of the present disclosure may be implemented.

Various embodiments and/or aspects are now disclosed with reference to the drawings. In the following description, for purposes of explanation, numerous specific details are set forth to provide a thorough understanding of one or more aspects. However, it will also be appreciated by one of ordinary skill in the art that such aspect(s) may be practiced without these specific details. The following description and accompanying drawings set forth in detail certain illustrative aspects of one or more aspects. These aspects are illustrative, however, and some of various methods may be employed in the principles of the various aspects, and the descriptions set forth are intended to include all such aspects and their equivalents. Specifically, as used herein, “embodiment”, “example”, “aspect”, “exemplary”, etc. are not to be construed as advantageous or advantageous over any aspect or design described herein. It may not be.

Further, various aspects and features will be presented by a system that may include one or more apparatuses, terminals, servers, devices, components and/or modules, and the like. The various systems may include additional apparatuses, terminals, servers, devices, components and/or modules, etc. and/or the apparatuses, terminals, servers, etc. discussed in connection with the drawings; It should also be understood and appreciated that it may not include all of devices, components, modules, etc.

As used herein, the terms "computer program," "component," "module," "system," and the like, may be used interchangeably with each other, and include computer-related entities, hardware, firmware, software, combinations of software and hardware; or to the execution of software. For example, a component can be, but is not limited to being, a process running on a processor, a processor, an object, a thread of execution, a program, and/or a computer. For example, both an application running on a computing device and the computing device may be a component. One or more components may reside within a processor and/or thread of execution. A component may be localized within one computer. A component may be distributed between two or more computers.

In addition, these components can execute from various computer readable media having various data structures stored therein. Components may communicate via a network such as the Internet with another system, for example, via a signal having one or more data packets (eg, data and/or signals from one component interacting with another component in a local system, distributed system, etc.) may communicate via local and/or remote processes depending on the data being transmitted).

Hereinafter, the same or similar components are assigned the same reference numerals regardless of reference numerals, and overlapping descriptions thereof will be omitted. In addition, in describing the embodiments disclosed in the present specification, if it is determined that detailed descriptions of related known technologies may obscure the gist of the embodiments disclosed in the present specification, the detailed description thereof will be omitted. In addition, the accompanying drawings are only for easy understanding of the embodiments disclosed in the present specification, and the technical ideas disclosed in the present specification are not limited by the accompanying drawings.

The terminology used herein is for the purpose of describing the embodiments and is not intended to limit the present disclosure. In this specification, the singular also includes the plural unless specifically stated otherwise in the phrase. As used herein, “comprises” and/or “comprising” does not exclude the presence or addition of one or more other components in addition to the stated components.

Although the first, second, etc. are used to describe various elements or elements, these elements or elements are not limited by these terms, of course. These terms are only used to distinguish one element or component from another. Accordingly, it goes without saying that the first element or component mentioned below may be the second element or component within the spirit of the present disclosure.

Unless otherwise defined, all terms (including technical and scientific terms) used herein may be used with the meaning commonly understood by those of ordinary skill in the art to which this disclosure belongs. In addition, terms defined in a commonly used dictionary are not to be interpreted ideally or excessively unless clearly defined in particular.

In addition, the term “or” is intended to mean an inclusive “or” rather than an exclusive “or.” That is, unless otherwise specified or clear from context, "X employs A or B" is intended to mean one of the natural implicit substitutions. That is, X employs A; X employs B; or when X employs both A and B, "X employs A or B" may apply to either of these cases. It should also be understood that the term “and/or” as used herein refers to and includes all possible combinations of one or more of the listed related items.

In addition, as used herein, the terms “information” and “data” can often be used interchangeably.

The suffixes “module” and “part” for components used in the following description are given or mixed in consideration of only the ease of writing the specification, and do not have distinct meanings or roles by themselves.

Objects and effects of the present disclosure, and technical configurations for achieving them will become clear with reference to the embodiments described below in detail in conjunction with the accompanying drawings. In describing the present disclosure, if it is determined that a detailed description of a well-known function or configuration may unnecessarily obscure the subject matter of the present disclosure, the detailed description thereof will be omitted. In addition, the terms to be described later are terms defined in consideration of functions in the present disclosure, which may vary according to intentions or customs of users and operators.

However, the present disclosure is not limited to the embodiments disclosed below and may be implemented in various different forms. Only the present embodiments are provided so that the present disclosure is complete, and to fully inform those of ordinary skill in the art to which the present disclosure belongs, the scope of the disclosure, and the present disclosure is only defined by the scope of the claims . Therefore, the definition should be made based on the content throughout this specification.

The scope of the steps in the claims of the present disclosure is generated by the functions and features described in each step, and unless the precedence of the order in each step is specified, in the claims The order of description of each step is not affected. For example, in a claim described as a step comprising step A and step B, even if step A is described before step B, the scope of the rights is not limited that step A must precede step B.

1 is a block diagram of a computing device performing translation according to some embodiments of the present disclosure.

Referring to FIG. 1 , a computing device 100 may include a processor 110 , a communication unit 120 , and a memory 130 . However, since the above-described components are not essential in implementing the computing device 100 , the computing device 100 may have more or fewer components than those listed above.

Computing apparatus 100 may include any type of computer system or computer device, such as, for example, microprocessors, mainframe computers, digital processors, portable devices and device controllers, and the like. In addition, the computing device 100 is a mobile phone, a smart phone (smart phone), a laptop computer (laptop computer), personal digital assistants (PDA), slate PC (slate PC), tablet PC (tablet PC), ultrabook (ultrabook) , a wearable device, and the like may be included. However, the present invention is not limited thereto.

The processor 110 of the computing device 100 typically controls the overall operation of the computing device 100 . The processor 110 processes signals, data, information, etc. input or output through components included in the computing device 100 or drives an application program stored in the memory 130 to provide appropriate information or functions to the user. or can be dealt with.

In addition, the processor 110 may control at least some of the components of the computing device 100 in order to drive an application program stored in the memory 130 . Furthermore, the processor 110 may operate by combining at least two or more of the components included in the computing device 100 to drive the application program.

According to some embodiments of the present disclosure, the processor 110 of the computing device 100 pre-processes the translation target sentence, and then inputs the pre-processed translation target sentence to an artificial neural network model related to translation to obtain a final translation sentence. can

Specifically, the processor 110 may obtain a translated second sentence by inputting the first sentence into the machine translation model learned using pair data consisting of a pair of departure sentence data and arrival sentence data. . Here, the first sentence may be a pre-processed translation target sentence.

On the other hand, when the second sentence is obtained, the processor 110 inputs the second sentence to the grammar correction model learned using a learning dataset consisting of inscription data and plaintext data labeled in the inscription data, You can get 3 sentences. In addition, the processor 110 may obtain a fourth sentence that is a final translated sentence for the first sentence through an additional proofreading operation on the third sentence using a preset rule. Here, the preset rule is a correction rule preset by the user, and may be pre-stored in the memory 130 .

Accordingly, the translation method of the present disclosure does not require a resource for a separate person in the translation process, thereby improving the productivity of translation.

Hereinafter, a description of how the computing device 100 of the present disclosure translates will be described later with reference to FIGS. 2 to 6 .

Meanwhile, the communication unit 120 of the computing device 100 may include one or more modules that enable communication between the computing device 100 and the user terminal and between the computing device 100 and servers. Also, the communication unit 120 may include one or more modules for connecting the computing device 100 to one or more networks.

A network connecting the communication between the computing device 100 and the user terminal and between the computing device 100 and the servers is a Public Switched Telephone Network (PSTN), x Digital Subscriber Line (xDSL), and Rate Adaptive (RADSL). DSL), MDSL (Multi Rate DSL), VDSL (Very High Speed DSL), UADSL (Universal Asymmetric DSL), HDSL (High Bit Rate DSL), and a variety of wired communication systems such as local area network (LAN) can be used.

In addition, the networks presented herein include Code Division Multi Access (CDMA), Time Division Multi Access (TDMA), Frequency Division Multi Access (FDMA), Orthogonal Frequency Division Multi Access (OFDMA), Single Carrier-FDMA (SC-FDMA) and Various wireless communication systems may be used, such as other systems.

The network according to the embodiments of the present disclosure may be configured regardless of its communication mode, such as wired and wireless, and is composed of various communication networks such as a local area network (LAN) and a wide area network (WAN). can be In addition, the network may be a well-known World Wide Web (WWW), and may use a wireless transmission technology used for short-range communication such as Infrared Data Association (IrDA) or Bluetooth.

The techniques described herein may be used in the networks mentioned above, as well as in other networks.

The memory 130 of the computing device 100 may store a program for the operation of the processor 110 , and may temporarily or permanently store input/output data. The memory 130 may include a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (eg, SD or XD memory, etc.), RAM (Random Access Memory, RAM), SRAM (Static Random Access Memory), ROM (Read-Only Memory, ROM), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory), magnetic memory, magnetic It may include at least one type of storage medium among disks and optical disks. The memory 130 may be operated under the control of the processor 110 .

According to the software implementation, embodiments such as the procedures and functions described in this specification may be implemented as separate software modules. Each of the software modules may perform one or more functions and operations described herein. The software code may be implemented as a software application written in a suitable programming language. The software code may be stored in the memory 130 of the computing device 100 and executed by the processor 110 of the computing device 100 .

2 is a flowchart illustrating an example of a translation method according to some embodiments of the present disclosure. 3 is a view for explaining an example of a training dataset for learning a grammar correction model according to some embodiments of the present disclosure.

Referring to FIG. 2 , the processor 110 of the computing device 100 inputs a first sentence into a machine translation model learned using pair data consisting of a pair of departure sentence data and arrival sentence data, and translates it. The second sentence can be obtained (S110).

Here, the machine translation model may refer to a model in which an artificial neural network is trained and, upon receiving a sentence written in the first language, outputs a sentence translated into a second language different from the first language. Such a machine translation model may support multilingual input and output through a preprocessing process and a learning process. Also, the machine translation model of the present disclosure may be a model including at least one encoder and at least one decoder. Here, a known machine translation model such as a transformer model, a model using a sequence-to-sequence technique, and a T5 model may be used as the machine translation model of the present disclosure. However, the present invention is not limited thereto.

Specifically, when a sentence written in the first language is input to the machine translation model, the sentence written in the first language may be divided into at least one token and input to at least one encoder. In this case, at least one encoder may output data compressed into multiple N-dimensional vectors. And, when the output data is input to the at least one decoder, the at least one decoder may output a sentence translated into the second language as a result of the translation. However, the present invention is not limited thereto.

The processor 110 of the computing device 100 may input the preprocessed translation target sentence to the machine translation model. Specifically, the first sentence input by the processor 110 to the machine translation model may be a text preset from the translation target sentence through a preprocessing process or a sentence in which special symbols are removed. For example, when the translation target sentence is English, the processor 110 may remove non-English sentences (eg, Korean) from the translation target sentence. However, the present invention is not limited thereto.

On the other hand, according to some embodiments of the present disclosure, when the processor 110 of the computing device 100 inputs the first sentence to the machine translation model, the language of the arrival sentence (the translated sentence) is generated through a preprocessing process. It can be added as a label for 1 sentence. That is, since the machine translation model has been trained to support multiple languages, when the first sentence is input to the machine translation model, a label for which language should be translated may be added to the first sentence.

Meanwhile, the method of training the machine translation model so that the machine translation model can support multiple languages is as follows.

When the processor 110 of the computing device 100 trains the machine translation model, the departure sentence data obtained by adding a label indicating the type of the second language of the arrival sentence in front of the starting sentence (translation target sentence) written in the first language can create Also, the processor 110 may generate arrival sentence data in which a label indicating the type of the first language is added to the front of the arrival sentence (translation completed sentence) written in the second language. In addition, the processor 110 may train the machine translation model by using pair data composed of the departure sentence data and the arrival sentence data.

For example, when the language type of the departure sentence is Korean and the language type of the arrival sentence is English, the processor 110 displays a label such as 'EN' indicating English as the second language type in front of the departure sentence. In addition, starting sentence data can be generated. Also, the processor 110 may generate arrival sentence data by adding a label such as 'KR' indicating Korean, which is the language of the departure sentence, in front of the arrival sentence written in English. In addition, the processor 110 may train the machine translation model by using pair data in which the departure sentence data and the arrival sentence data are a pair. However, the present invention is not limited thereto.

That is, the starting sentence data constituting the pair data used in the learning process may be data in which a label indicating the type of the second language of the arrival sentence is added to the front of the starting sentence written in the first language through the preprocessing process. And, the arrival sentence data constituting the pair data used in the learning process may be data in which a label indicating the type of the first language is added to the front of the arrival sentence written in the second language through the preprocessing process. However, the present invention is not limited thereto.

Meanwhile, according to some embodiments of the present disclosure, the language of the departure sentence is added to the arrival sentence data as a label, and since the departure sentence exists as pair data, the processor 110 may translate the second language into the first language. It is also possible to train a machine translation model. As a result, the processor 110 may train the machine translation model to translate the first language into the second language through one pair of data, or train the machine learning model to translate the second language into the first language. You can also learn However, the present invention is not limited thereto.

Additionally, the machine translation model of the present disclosure may be fine-tuned through additional data related to the field to be used according to the field used. For example, when the word 'justice' is in the sentence to be translated, it may mean 'the meaning of the term' or 'correct reason for the truth' depending on the situation. In the machine translation model, weight adjustment may be performed using data with different weights according to fields used to classify them.

The machine translation model of the present disclosure may provide a more specialized machine translation function in the field used through the fine adjustment process as described above.

According to some embodiments of the present disclosure, after obtaining the second sentence, the processor 110 of the computing device 100 corrects the learned grammar using a learning dataset including inscription data and plaintext data labeled in the inscription data. By inputting the second sentence to the model, a corrected third sentence may be obtained (S120).

Here, the grammar correction model may refer to a model in which an artificial neural network is trained and, when a sentence in which a grammatical error exists is inputted, a sentence in which the grammatical error is removed is output. This grammar correction model can support multilingual input and output through a preprocessing process and a learning process. Also, the grammar correction model of the present disclosure may be a text generation model including at least one encoder and at least one decoder. In addition, the grammar correction model of the present disclosure may be trained to infer a corrected main sentence from a large amount of plaintext (plaintext data) and inscription text (inscription data) generated based on the large amount of plaintext (plaintext data) for languages to be corrected in advance. . For example, the Bert-base-multilingual-cased model may be used for the grammar correction model of the present disclosure. However, the present invention is not limited thereto.

The inscription data used for learning the grammar correction model may be sentence data including at least one of an article error, a preposition error, a position error of a number, a sentence structure error, or a spacing error. Additionally, the inscription data may include not only sentence data including grammatical errors, but also text data related to new words and buzzwords by collecting the latest text data from the Internet. Meanwhile, the plaintext data used for learning the grammar correction model may be sentence data in which errors, new words, or buzzwords included in the inscription data are corrected in a standard language. However, the present invention is not limited thereto.

3, the underline in FIG. 3 is for indicating a position where an error exists in the input data or a corrected position in a sentence output from the grammar correction model for convenience of explanation, and the underline does not exist in the actual input data or the output sentence. it may not be In addition, the underline and bold marks present in the description related to FIG. 3 are for indicating the position where an error exists in the input data or the corrected position in the sentence output from the grammar correction model, and the actual input data or the output sentence underline and bold may not exist.

For example, referring to FIG. 3A , inscription data, which is input data used to train a grammar correction model in a learning process, is a sentence in which a position error of a number exists, 'The redevelopment of the Angolan mining industry started in the late s. It can be 2000 '. In addition, the plaintext data labeled on the inscription data may be 'The redevelopment of the Angolan mining industry started in the late 2000 s.' The grammar correction model trained using this inscription data and the training dataset consisting of the plaintext data labeled in the inscription data is 'The redevelopment of the Angolan mining industry started in the late s. When the sentence ' 2000 ' is received as input, it may be learned to output the sentence 'The redevelopment of the Angolan mining industry started in the late 2000 s.'

For another example, referring to (b) of FIG. 3 , the inscription data, which is input data used to train the grammar correction model in the learning process, is a sentence in which a positional error of a number exists, 'The Angolan government loses million annually from diamond smuggling. It can be $375 '. In addition, the plaintext data labeled on the inscription data may be 'The Angolan government loses $ 375 million annually from diamond smuggling.' The grammar correction model learned using the training dataset consisting of the inscription data and the plaintext data labeled in the inscription data is 'The Angolan government loses million annually from diamond smuggling. When a sentence of ' $ 375 ' is received as an input, it may be learned to output the sentence 'The Angolan government loses $ 375 million annually from diamond smuggling.'

As another example, referring to (c) of FIG. 3 , inscription data, which is input data used to train a grammar correction model in the learning process, is a sentence in which a preposition (or conjunction) error exists, 'Andorra was admitted' to UEFA and FIFA and the same year, 1996.' can be In addition, the plaintext data labeled on the inscription data may be 'Andorra was admitted to UEFA and FIFA in the same year, 1996.'. A grammar correction model trained using this inscription data and a training dataset consisting of plaintext data labeled on the inscription data is subjected to an input of the sentence 'Andorra was admitted to UEFA and FIFA and the same year, 1996.' case, it may be trained to output the sentence 'Andorra was admitted to UEFA and FIFA in the same year, 1996.'.

As another example, referring to (d) of FIG. 3 , the inscription data, which is input data used to train the grammar correction model in the learning process, is a sentence with an article error, 'Andorra first participated at an Olympic Games in 1976.' In addition, the plaintext data labeled on the inscription data may be 'Andorra first participated at the Olympic Games in 1976.'. The grammar correction model trained using the training dataset consisting of the inscription data and the plaintext data labeled in the inscription data is 'Andorra first participated at an Olympic Games in 1976.' It can be learned to output the sentence 'first participated at the Olympic Games in 1976.'

According to some additional embodiments of the present disclosure, the processor 110 of the computing device 100 may generate inscription data by using the plaintext data. In addition, the processor 110 may label the generated inscription data with plaintext data and use it as a learning data set for learning the grammar correction model.

Specifically, the processor 110 may recognize each type (preposition, article, space, etc.) of words included in plaintext data (sentences composed of plaintext). In addition, when the processor 110 generates inscription data related to a specific error (eg, a preposition error), at least one word that may generate a specific error (eg, a preposition error) in the plaintext data can recognize In addition, the processor 110 may generate inscription data including a grammatical error by changing at least one recognized word to another word having the same type or a space. Then, the processor 110 may learn the grammar correction model by labeling the generated inscription data with plaintext data.

For example, the processor 110 may recognize each type of words included in the plaintext data 'The redevelopment of the Angolan mining industry started in the late 2000 s.' In addition, the processor 110 may recognize the prepositions of and in in the plaintext data in order to generate a preposition error. Then, the processor 110 changes each of and in to different words for and space, 'The redevelopment for the Angolan mining industry started', which is inscription data including a grammatical error. the late 2000 s.' The processor 110 'The redevelopment for the Angolan mining industry started The grammar correction model can be trained by labeling the plaintext data 'The redevelopment of the Angolan mining industry started in the late 2000 s.' to the late 2000 s.'

However, the present disclosure is not limited thereto, and the processor 110 of the present disclosure may generate inscription data by changing a part of a sentence included in the plaintext data in various ways. In this way, when the grammar correction model is trained by generating inscription data, there is no need to separately collect inscription data, thereby saving time and cost.

Additionally, the grammar correction model of the present disclosure may be fine-tuned through additional data related to the field to be used according to the field to be used. Specifically, in the grammar correction model, weight adjustment may be performed on a plurality of words capable of correcting a specific error existing in a sentence to be corrected by using data each having different weights.

The grammar correction model of the present disclosure may provide a more professional grammar correction function in the field used through the fine adjustment process as described above.

Referring back to FIG. 2 , the processor 110 of the computing device 100 of the present disclosure obtains the third sentence, and then uses a preset rule to add the third sentence to the first sentence through additional proofreading. It is possible to obtain a fourth sentence that is the final translated sentence for the text (S130). Here, the preset rule is a correction rule preset by the user, and may be pre-stored in the memory 130 . However, the present invention is not limited thereto.

For example, the processor 110 may obtain the fourth sentence by inserting a line break symbol after the punctuation related to the end of the sentence among the punctuation marks included in the third sentence.

Here, punctuation marks related to the end of a sentence may include, for example, a period (.), an exclamation point (!), and a question mark (?). That is, when a dot, an exclamation point, and a question mark exist in the third sentence, the processor 110 may obtain the fourth sentence by inserting a line break symbol after the dot, exclamation point, and question mark.

For example, assuming that the newline symbol is ' \n ', the processor 110 will 'inflict 130% physical damage to one enemy. There is a 50% chance that the blindness will be applied for 2 turns.' In the third sentence, inserts a line break symbol after the dot, and in the fourth sentence, 'Inflicts 130% physical damage to one enemy. ＼n There is a 50% chance that the blindness will be applied for 2 turns.' However, the present invention is not limited thereto.

As another example, the processor 110 may obtain the fourth sentence by adding a space to the third sentence based on a preset rule.

For example, if there is a preset rule that the processor 110 has no space to the right of the color tag, '[fff31b]<QUEST> The fourth sentence, '[fff31b]<QUEST>[-] Unlocked!!', can be obtained by removing the space on the right side of the color tag in the 3rd sentence, '[-] Unlocked!!'. However, the present invention is not limited thereto.

FENSA'라는 제 3 문장에서 % 기호 앞에 공백을 추가하여, 제 4 문장인 'Ignore 40 % de D

FENSA'를 획득할 수 있다. 다만, 이에 한정되는 것은 아니다.For another example, if there is a preset rule that French requires a space before the % sign, 'Ignore 40% de D'

By adding a space before the % sign in the third sentence of 'FENSA', the fourth sentence 'Ignore 40' % de D

FENSA' can be obtained. However, the present invention is not limited thereto.

Since the computing device 100 of the present disclosure outputs the results of machine translation and grammar proofreading, the quality of translation may be improved. In addition, the computing device 100 of the present disclosure may resolve errors related to new words and buzzwords that were not considered in the existing grammar correction method.

Through this, the computing device 100 of the present disclosure can reduce the cost of collecting numerous human resources and data required to improve the performance of machine translation itself.

In addition, according to the translation method of the present disclosure, it is possible to provide a highly complete translation with a small infrastructure in a global service that needs to translate and provide data into various languages. For example, the translation method of the present disclosure may be used for services such as mail, announcement, and in-game text translation.

4 is a flowchart illustrating an example of a translation process according to some embodiments of the present disclosure.

According to some embodiments of the present disclosure, the processor 110 of the computing device 100 may obtain a translated second sentence by inputting the first sentence into the machine translation model. Also, the processor 110 may obtain the third sentence by inputting the second sentence into the grammar correction model. In addition, the processor 110 may obtain a fourth sentence that is a final translated sentence for the first sentence through an additional proofreading operation for the third sentence.

Hereinafter, the processor 110 of the computing device 100 inputs the second sentence to the grammar correction model to obtain the third sentence, and the processor 110 performs an additional correction operation on the third sentence. Steps will be described in detail.

Referring to FIG. 4 , before inputting the second sentence into the grammar correction model, the processor 110 of the computing device 100 removes the newline symbol and removes the newline if the second sentence contains a newline symbol. The location information of the symbol may be stored in the memory (S210). Then, the processor 110 may obtain the third sentence by inputting the second sentence from which the newline symbol is removed to the grammar correction model ( S220 ). Here, the position information of the newline symbol may include a string index. However, the present invention is not limited thereto.

According to some embodiments of the present disclosure, the processor 110 removes the line break symbol from the second sentence, and then inputs the second sentence from which the line break symbol is removed to the grammar correction model, so that the grammar correction model is more accurately applied to the grammar correction model. This can lead to grammar correction. That is, the grammar correction model of the present disclosure may perform grammar correction with higher accuracy on a sentence without a newline symbol.

Meanwhile, after obtaining the third sentence, the processor 110 may recognize whether the insertion position of the newline symbol in the third sentence exists within the word by using the position information of the newline symbol ( S230 ). . Then, when recognizing that the insertion position exists within the word, the processor 110 may insert a newline symbol in front of the word to obtain the fourth sentence ( S240 ).

For example, suppose the 3rd sentence is 'This is a translated string', and the insertion position of the newline symbol ( \n ) is the string index '14' (i.e., 'This is a tran \n slated string'). , the processor 110 may recognize that the insertion position of the newline symbol exists within the word 'translated'. In this case, the processor 110 may obtain the fourth sentence 'This is a \n translated string' by inserting a newline symbol before the word 'translated'. However, the present invention is not limited thereto.

According to some embodiments of the present disclosure, the processor 110 inputs the second sentence from which the newline symbol has been removed into the grammar correction model, and not only induces the grammar to be more accurately corrected by the grammar correction model, but also corrects the grammar. Line break marks can be inserted at appropriate positions so that the sentences corrected by the model are natural.

5 is a flowchart illustrating another example of a translation process according to some embodiments of the present disclosure.

According to some embodiments of the present disclosure, the processor 110 of the computing device 100 may obtain a translated second sentence by inputting the first sentence into the machine translation model. Also, the processor 110 may obtain the third sentence by inputting the second sentence into the grammar correction model. In addition, the processor 110 may obtain a fourth sentence that is a final translated sentence for the first sentence through an additional proofreading operation for the third sentence.

Hereinafter, a process in which the processor 110 of the computing device 100 performs pre-processing on the second sentence before inputting the second sentence into the grammar correction model will be described in detail.

Referring to FIG. 5 , the processor 110 of the computing device 100 may compare the first sentence with the second sentence and recognize whether there is a number lost in the translation process through the machine translation model ( S310 ). .

Specifically, the processor 110 may recognize the missing number by using a sequence of digits for each of the first sentence and the second sentence. For example, the processor 110 uses the Number Detecting function (a function that selects and compares only the isdigit() part of the letters of all sentences) to determine whether there is a difference between the numbers included in each of the first sentence and the second sentence. can be checked, and the lost number can be recognized according to whether the number has changed. For example, the processor 110 may indicate '50%' included in the first sentence as '*%' in the second sentence after the translation is performed through the machine translation model. In this case, the processor 110 may recognize that the number '50' in front of the % has been lost.

Meanwhile, when the processor 110 recognizes that there is a missing number in the second sentence, the processor 110 may add the missing number to the end of the second sentence ( S320 ). Then, the processor 110 may obtain the third sentence by inputting the second sentence in which the missing number is added to the end of the sentence into the grammar correction model (S330).

For example, if the first sentence is '50% chance that your real name is applied for 2 turns.' and the second sentence is 'There is a *% chance that your real name is applied for 2 turns.' The processor 110 may recognize that the number '50' in front of the % has been lost. In this case, the processor 110 may add '50', which is a number lost in the second sentence, to the end of the second sentence. Then, the processor 110 inputs 'There is a *% chance that your real name is applied for 2 turns.50', which is the second sentence in which the missing number is added to the end of the sentence, into the grammar correction model, and the third You can obtain the sentence 'There is a 50% chance that your real name is applied for 2 turns.' However, the present invention is not limited thereto.

In general, when a number is lost in a translation process, it may be nearly impossible to predict the missing number in a sentence. Accordingly, the processor 110 of the present disclosure restores the lost number in the translation process, thereby preventing the loss of the number included in the sentence.

6 is a view for explaining an example of each of the translation process according to some embodiments of the present disclosure.

According to some embodiments of the present disclosure, the processor 110 of the computing device 100 may obtain a translated second sentence by inputting the first sentence into the machine translation model. Also, the processor 110 may obtain the third sentence by inputting the second sentence into the grammar correction model. In addition, the processor 110 may obtain a fourth sentence that is a final translated sentence for the first sentence through an additional proofreading operation for the third sentence.

Hereinafter, referring to FIG. 6 , the computing device 100 'Inflicts 130% physical damage to one enemy. ＼n 50% chance to apply blindness for 2 turns.' Explain each example of the process of translating the sentence.

First, the processor 110 of the computing device 100 'Inflicts 130% physical damage to one enemy. If you obtain a translation target sentence (1) saying ＼n 50% chance that your real name will be applied for 2 turns.

For example, the processor 110 may remove a line break symbol from the translation target sentence 1 and store [27], which is location information on the line break symbol, in the memory 130 . Here, the position information on the newline symbol may mean a string index. That is, the processor 110 of the present disclosure may remember that the newline symbol is removed from the 27th character string in the translation target sentence 1 .

The processor 110 can obtain the first sentence (2) with the line break symbol removed, 'Inflicts 130% physical damage to one enemy. There is a 50% chance that the blindness will be applied for 2 turns.' The processor 110 inputs the first sentence 2 into the machine translation model, and the second sentence 3 (ie, 'Deals % Physical Damage to 1 enemy.50% chance to Blind of target for 2 turns.') translation) can be obtained. Here, in the second sentence (3), the number '130' in front of % is lost.

Meanwhile, the processor 110 may compare the first sentence 2 and the second sentence 3 to recognize whether there is a number lost in the translation process through the machine translation model. The processor 110 may add the missing number '130' to the end of the second sentence 3 . Then, the third sentence 5 may be obtained by inputting the second sentence 4 in which the missing number is added to the end of the sentence into the grammar correction model. Here, the grammar correction model may correct an error in the sentence structure by restoring the lost number '130' and correcting 'of' to 'the'.

That is, the processor 110 may obtain the third sentence 5 'Deals 130% Physical Damage to 1 enemy.50% chance to Blind the target for 2 turns.' output by the grammar correction model.

In addition, the processor 110 may acquire the fourth sentence 6 that is the final translation sentence for the first sentence 2 through an additional proofreading operation on the third sentence 5 .

For example, the processor 110 inserts a newline symbol after the dot (.), which is a punctuation mark related to the end of a sentence, and inserts a newline symbol after the fourth sentence, 'Deals 130% Physical Damage to 1 enemy. ＼n 50% chance to Blind the target for 2 turns.'

The description of FIG. 6 describes each translation process for a specific sentence in order to help the understanding of the overall flow of the present disclosure, and the present disclosure is not limited to the embodiments disclosed in FIG. 6 .

7 depicts a simplified, general schematic diagram of an exemplary computing environment in which embodiments of the present disclosure may be implemented.

Although the present disclosure has been described above generally in the context of computer-executable instructions that may be executed on one or more computers, those skilled in the art will appreciate that the present disclosure may be implemented in combination with other program modules and/or in a combination of hardware and software. will be.

Generally, modules herein include routines, procedures, programs, components, data structures, etc. that perform particular tasks or implement particular abstract data types. In addition, those skilled in the art will appreciate that the methods of the present disclosure can be applied to single-processor or multiprocessor computer systems, minicomputers, mainframe computers as well as personal computers, handheld computing devices, microprocessor-based or programmable consumer electronics, etc. (each of which is It will be appreciated that other computer system configurations may be implemented, including those that may operate in connection with one or more associated devices.

The described embodiments of the present disclosure may also be practiced in distributed computing environments where certain tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote memory storage devices.

Computers typically include a variety of computer-readable media. Any medium accessible by a computer can be a computer-readable medium, and such computer-readable media includes volatile and nonvolatile media, transitory and non-transitory media, removable and non-transitory media. including removable media. By way of example, and not limitation, computer-readable media may include computer-readable storage media and computer-readable transmission media.

Computer readable storage media includes volatile and nonvolatile media, temporary and non-transitory media, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. includes media. A computer-readable storage medium may be RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital video disk (DVD) or other optical disk storage device, magnetic cassette, magnetic tape, magnetic disk storage device, or other magnetic storage device. device, or any other medium that can be accessed by a computer and used to store the desired information.

A computer readable transmission medium typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal, such as a carrier wave or other transport mechanism, and the like. Includes all information delivery media. The term modulated data signal means a signal in which one or more of the characteristics of the signal is set or changed so as to encode information in the signal. By way of example, and not limitation, computer-readable transmission media includes wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, RF, infrared, and other wireless media. Combinations of any of the above are also intended to be included within the scope of computer-readable transmission media.

An example environment 1100 implementing various aspects of the disclosure is shown including a computer 1102 , the computer 1102 including a processing unit 1104 , a system memory 1106 , and a system bus 1108 . do. A system bus 1108 couples system components, including but not limited to system memory 1106 , to the processing device 1104 . The processing device 1104 may be any of a variety of commercially available processors. Dual processor and other multiprocessor architectures may also be used as processing unit 1104 .

The system bus 1108 may be any of several types of bus structures that may be further interconnected to a memory bus, a peripheral bus, and a local bus using any of a variety of commercial bus architectures. System memory 1106 includes read only memory (ROM) 1110 and random access memory (RAM) 1112 . A basic input/output system (BIOS) is stored in non-volatile memory 1110, such as ROM, EPROM, EEPROM, etc., the BIOS is the basic input/output system (BIOS) that helps transfer information between components within computer 1102, such as during startup. contains routines. RAM 1112 may also include high-speed RAM, such as static RAM, for caching data.

The computer 1102 may also include an internal hard disk drive (HDD) 1114 (eg, EIDE, SATA) - this internal hard disk drive 1114 may also be configured for external use within a suitable chassis (not shown). Yes—a magnetic floppy disk drive (FDD) 1116 (eg, for reading from or writing to removable diskette 1118), and an optical disk drive 1120 (eg, a CD-ROM) for reading from, or writing to, disk 1122, or other high capacity optical media, such as DVD. The hard disk drive 1114 , the magnetic disk drive 1116 , and the optical disk drive 1120 are connected to the system bus 1108 by the hard disk drive interface 1124 , the magnetic disk drive interface 1126 , and the optical drive interface 1128 , respectively. ) can be connected to Interface 1124 for external drive implementation includes, for example, at least one or both of Universal Serial Bus (USB) and IEEE 1394 interface technologies.

These drives and their associated computer-readable media provide non-volatile storage of data, data structures, computer-executable instructions, and the like. In the case of computer 1102, drives and media correspond to storing any data in a suitable digital format. Although the description of computer-readable storage media above refers to HDDs, removable magnetic disks, and removable optical media such as CDs or DVDs, those skilled in the art will use zip drives, magnetic cassettes, flash memory cards, cartridges, It will be appreciated that other tangible computer-readable storage media and the like may also be used in the exemplary operating environment and any such media may include computer-executable instructions for performing the methods of the present disclosure. .

A number of program modules may be stored in the drive and RAM 1112 , including an operating system 1130 , one or more application programs 1132 , other program modules 1134 , and program data 1136 . All or portions of the operating system, applications, modules, and/or data may also be cached in RAM 1112 . It will be appreciated that the present disclosure may be implemented in various commercially available operating systems or combinations of operating systems.

A user may enter commands and information into the computer 1102 via one or more wired/wireless input devices, for example, a pointing device such as a keyboard 1138 and a mouse 1140 . Other input devices (not shown) may include a microphone, IR remote control, joystick, game pad, stylus pen, touch screen, and the like. Although these and other input devices are connected to the processing unit 1104 through an input device interface 1142 that is often connected to the system bus 1108, parallel ports, IEEE 1394 serial ports, game ports, USB ports, IR interfaces, It may be connected by other interfaces, etc.

A monitor 1144 or other type of display device is also coupled to the system bus 1108 via an interface, such as a video adapter 1146 . In addition to the monitor 1144, the computer typically includes other peripheral output devices (not shown), such as speakers, printers, and the like.

Computer 1102 may operate in a networked environment using logical connections to one or more remote computers, such as remote computer(s) 1148 via wired and/or wireless communications. Remote computer(s) 1148 may be workstations, server computers, routers, personal computers, portable computers, microprocessor-based entertainment devices, peer devices, or other common network nodes, and are generally Although including many or all of the components described, only memory storage device 1150 is shown for simplicity. The logical connections shown include wired/wireless connections to a local area network (LAN) 1152 and/or a larger network, eg, a wide area network (WAN) 1154 . Such LAN and WAN networking environments are common in offices and companies, and facilitate enterprise-wide computer networks, such as intranets, all of which can be connected to a worldwide computer network, for example, the Internet.

When used in a LAN networking environment, the computer 1102 is coupled to the local network 1152 through a wired and/or wireless communication network interface or adapter 1156 . Adapter 1156 may facilitate wired or wireless communication to LAN 1152 , which LAN 1152 also includes a wireless access point installed therein for communicating with wireless adapter 1156 . When used in a WAN networking environment, the computer 1102 may include a modem 1158 , connected to a communication server on the WAN 1154 , or otherwise establishing communications over the WAN 1154 , such as over the Internet. have the means A modem 1158 , which may be internal or external and a wired or wireless device, is coupled to the system bus 1108 via a serial port interface 1142 . In a networked environment, program modules described for computer 1102 , or portions thereof, may be stored in remote memory/storage device 1150 . It will be appreciated that the network connections shown are exemplary and other means of establishing a communication link between the computers may be used.

The computer 1102 may be associated with any wireless device or object that is deployed and operates in wireless communication, for example, a printer, scanner, desktop and/or portable computer, portable data assistant (PDA), communication satellite, wireless detectable tag. It operates to communicate with any device or place, and phone. This includes at least Wi-Fi and Bluetooth wireless technologies. Accordingly, the communication may be a predefined structure as in a conventional network or may simply be an ad hoc communication between at least two devices.

Wi-Fi (Wireless Fidelity) makes it possible to connect to the Internet, etc. without a wired connection. Wi-Fi is a wireless technology such as cell phones that allows these devices, eg, computers, to transmit and receive data indoors and outdoors, ie anywhere within range of a base station. Wi-Fi networks use a radio technology called IEEE 802.11 (a, b, g, etc.) to provide secure, reliable, and high-speed wireless connections. Wi-Fi can be used to connect computers to each other, to the Internet, and to wired networks (using IEEE 802.3 or Ethernet). Wi-Fi networks may operate in unlicensed 2.4 and 5 GHz radio bands, for example, at 11 Mbps (802.11a) or 54 Mbps (802.11b) data rates, or in products that include both bands (dual band). have.

Those of ordinary skill in the art of the present disclosure will recognize that the various illustrative logical blocks, modules, processors, means, circuits, and algorithm steps described in connection with the embodiments disclosed herein include electronic hardware, (convenience For this purpose, it will be understood that it may be implemented by various forms of program or design code (referred to herein as "software") or a combination of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. A person skilled in the art of the present disclosure may implement the described functionality in various ways for each specific application, but such implementation decisions should not be interpreted as a departure from the scope of the present disclosure.

The various embodiments presented herein may be implemented as methods, apparatus, or articles of manufacture using standard programming and/or engineering techniques. The term “article of manufacture” includes a computer program, carrier, or media accessible from any computer-readable device. For example, computer-readable storage media include magnetic storage devices (eg, hard disks, floppy disks, magnetic strips, etc.), optical disks (eg, CDs, DVDs, etc.), smart cards, and flash drives. memory devices (eg, EEPROMs, cards, sticks, key drives, etc.). The term “machine-readable medium” includes, but is not limited to, wireless channels and various other media that can store, hold, and/or convey instruction(s) and/or data.

It is understood that the specific order or hierarchy of steps in the presented processes is an example of exemplary approaches. Based on design priorities, it is understood that the specific order or hierarchy of steps in the processes may be rearranged within the scope of the present disclosure. The appended method claims present elements of the various steps in a sample order, but are not meant to be limited to the specific order or hierarchy presented.

The description of the presented embodiments is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the scope of the present disclosure. Thus, the present disclosure is not intended to be limited to the embodiments presented herein, but is to be construed in the widest scope consistent with the principles and novel features presented herein.

Claims (11)

A method for a processor of a computing device to translate, comprising:
obtaining a translated second sentence by inputting a first sentence into a machine translation model learned using pair data consisting of a pair of departure sentence data and arrival sentence data;
obtaining a corrected third sentence by inputting the second sentence into a grammar correction model learned using a learning dataset composed of inscription data and plaintext data labeled with the inscription data; and
obtaining a fourth sentence that is a final translated sentence for the first sentence through an additional proofing operation on the third sentence using a preset rule;
containing,
How to translate.
The method of claim 1,
The first sentence is
Pre-set text or special symbols removed from the text to be translated through the pre-processing process,
How to translate.
The method of claim 1,
The starting sentence data is,
It is data in which a label indicating the type of the second language of the arrival sentence is added to the front of the departure sentence written in the first language through the preprocessing process,
The arrival sentence data is
Data in which a label indicating the type of the first language is added in front of the arrival sentence written in the second language through the preprocessing process,
How to translate.
The method of claim 1,
The inscription data is
It is sentence data including at least one of an article error, a preposition error, a position error of a number, a sentence structure error, or a space error,
The plaintext data is
Sentence data in which errors included in the inscription data are corrected,
How to translate.
The method of claim 1,
The step of obtaining a corrected third sentence by inputting the second sentence into a grammar correction model learned using a learning dataset consisting of the inscription data and the plaintext data labeled in the inscription data includes:
removing the line break symbol when the line break symbol is present in the second sentence, and storing location information of the line break symbol in a memory; and
obtaining the third sentence by inputting the second sentence from which the newline symbol is removed into the grammar correction model;
containing,
How to translate.
6. The method of claim 5,
Using the preset rule, obtaining a fourth sentence that is a final translated sentence for the first sentence through additional proofreading for the third sentence includes:
recognizing whether the insertion position of the newline symbol in the third sentence exists within a word by using the position information of the newline symbol; and
when it is recognized that the insertion position exists within the word, inserting the newline symbol in front of the word to obtain the fourth sentence;
containing,
How to translate.
The method of claim 1,
The step of obtaining a corrected third sentence by inputting the second sentence into a grammar correction model learned using a learning dataset consisting of the inscription data and the plaintext data labeled in the inscription data includes:
comparing the first sentence with the second sentence, and recognizing whether there is a number lost in a translation process through the machine translation model;
adding the missing number to the end of the second sentence when it is recognized that there is the missing number; and
obtaining the third sentence by inputting the second sentence in which the missing number is added to the end of the sentence into the grammar correction model;
containing,
How to translate.
8. The method of claim 7,
The step of obtaining the third sentence by inputting the second sentence to which the missing number is added to the grammar correction model includes:
obtaining the third sentence in which the lost number is restored to an appropriate position from the grammar correction model;
containing,
How to translate.
The method of claim 1,
Using the preset rule, obtaining a fourth sentence that is a final translated sentence for the first sentence through an additional proofreading operation for the third sentence includes:
obtaining the fourth sentence by inserting a line break symbol after a punctuation mark related to an end of a sentence among punctuation marks included in the third sentence;
containing,
How to translate.
The method of claim 1,
Using the preset rule, obtaining a fourth sentence that is a final translated sentence for the first sentence through additional proofreading for the third sentence includes:
obtaining the fourth sentence by adding a space to the third sentence based on the preset rule;
containing,
How to translate.
A computer program stored in a computer readable storage medium, comprising:
The computer program includes instructions for causing a processor of a computing device to perform the following steps:
obtaining a translated second sentence by inputting a first sentence into a machine translation model learned using pair data consisting of a pair of departure sentence data and arrival sentence data;
obtaining a corrected third sentence by inputting the second sentence into a grammar correction model learned using a learning dataset composed of inscription data and plaintext data labeled with the inscription data; and
obtaining a fourth sentence that is a final translated sentence for the first sentence through additional proofreading for the third sentence using a preset rule;
containing,
A computer program stored on a computer-readable storage medium.

Images