Information Processing Apparatus, Information Processing System, Information Processing Method, and Program

Abstract

Provided are an information processing apparatus, an information processing system, an information processing method, and a program that can improve the convenience regarding the evaluation of editing accuracy. 【Solution means】The information processing apparatus includes at least one memory configured to store a program, and at least one processor configured to execute processing based on the program. The at least one processor generates a first prompt including the original text, an edited text obtained by editing the original text, and an instruction to determine an editing error in the edited text in association with an error classification, and acquires the editing error corresponding to the error classification by inputting the first prompt to a language model.

Description

Technical Field

[0001] The present invention relates to an information processing apparatus, an information processing system, an information processing method, and a program.

Background Art

[0002] Conventionally, for example, as disclosed in Patent Document 1, there has been disclosed an information processing system in which, after an evaluator determines whether a translation result translated by a learning model is appropriate, the determination result by the evaluator is input by the evaluator. In such an information processing system, the learning model can be evaluated by evaluating the translation result, which is an example of the editing result.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in such an information processing system, as a result of the evaluator determining whether the editing result is appropriate, the determination result is input by the evaluator. Therefore, it takes time and effort for the evaluator. Thus, it is desired to improve the convenience regarding the evaluation of the editing accuracy.

Means for Solving the Problems

[0005] The information processing apparatus for solving the above problems includes at least one memory configured to store a program, and at least one processor configured to execute processing based on the program. The at least one processor generates a first prompt including the original text, an edited text obtained by editing the original text, and an instruction to determine an editing error in the edited text in association with an error classification, and acquires the editing error corresponding to the error classification by inputting the first prompt into a language model.

[0006] The information processing system for solving the above problems includes at least one memory configured to store a program, and at least one processor configured to execute processing based on the program. The at least one processor generates a first prompt including the original text, an edited text obtained by editing the original text, and an instruction to determine an editing error in the edited text in association with an error classification, and acquires the editing error corresponding to the error classification by inputting the first prompt into a language model.

[0007] The information processing method for solving the above problems includes: a first prompt including the original text, an edited text obtained by editing the original text, and an instruction to determine an editing error in the edited text in association with an error classification is generated by at least one processor; and the editing error corresponding to the error classification is acquired by inputting the first prompt into a language model.

[0008] The program for solving the above problems causes at least one processor to generate a first prompt including the original text, an edited text obtained by editing the original text, and an instruction to determine an editing error in the edited text in association with an error classification, and acquire the editing error corresponding to the error classification by inputting the first prompt into a language model.

Advantages of the Invention

[0009] According to the present invention, the convenience regarding the evaluation of editing accuracy can be improved.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Mode for Carrying Out the Invention

[0011] [First Embodiment] One embodiment of an information processing apparatus, an information processing system, an information processing method, and a program will be described.

[0012] <Configuration of Information Processing System 10> As shown in FIG. 1, the information processing system 10 is a system that analyzes the translation accuracy of a translated text. The translated text is a text obtained by translating the original text in the first language into the second language. The translated text is also an edited text in which the original text has been edited. The translated text is also a processed text in which the original text has been processed. For example, the first language may be Japanese and the second language may be English. For example, the first language may be English and the second language may be Japanese. The information processing system 10 may be a system that obtains a translated text from the original text.

[0013] The original text is at least one sentence and may include a plurality of sentences. For example, the original text may be an explanatory text in an e-commerce service. For example, the original text may be an explanatory text in a travel service.

[0014] The information processing system 10 includes an information processing device 11. The information processing device 11 obtains a translated text. The information processing device 11 obtains a translation error in the translated text. The information processing device 11 analyzes the translated text. The information processing device 11 evaluates the translated text. The information processing device 11 displays an image related to the translation error.

[0015] The information processing system 10 may include at least one translation server 12. The translation server 12 may be a server for translating the input original text. The information processing system 10 may include a first translation server 13 and a second translation server 14.

[0016] The first translation server 13 may include a first translation language model 13A. The first translation language model 13A is a language model for translating the original text. The first translation language model 13A may be a machine translation model for translating the original text. The first translation language model 13A corresponds to an example of a first machine translation model, and the translated text obtained by translating the original text by the first translation language model 13A corresponds to an example of a first translated text.

[0017] The second translation server 14 may include a second translation language model 14A. The second translation language model 14A is a language model for translating the original text. The second translation language model 14A may be a machine translation model for translating the original text. The second translation language model 14A corresponds to an example of a second machine translation model, and the translated text obtained by translating the original text by the second translation language model 14A corresponds to an example of a second translated text.

[0018] The information processing system 10 may include a large language model (LLM) server 15. Hereinafter, the large language model is referred to as an LLM. The LLM server 15 may include an LLM 15A. The LLM 15A is a language model constructed by a large amount of data and deep learning technology. The LLM server 15 has a function of controlling various texts using generative artificial intelligence. The LLM server 15 may also have a function of controlling various images using generative artificial intelligence. The LLM server 15 includes various learning models of generative artificial intelligence.

[0019] As a specific example, the LLM server 15 has a function of detecting a translation error based on the input original text and the translated text. The LLM server 15 has a function of generating an analysis result obtained by analyzing the translation error. The analysis of the translation error may include the error classification of the translation error, the determination of the appropriateness of the translation error, and the content regarding the severity of the translation error.

[0020] The information processing device 11, the translation server 12, and the LLM server 15 may be able to communicate with each other via the network 19. Hereinafter, the description of communicating via the network 19 for the communication between the information processing device 11, the translation server 12, and the LLM server 15 is omitted.

[0021] <Configuration of the information processing device 11> The information processing apparatus 11 may be realized by at least one computer. The information processing apparatus 11 includes at least one processor 20 and at least one memory 21. The information processing apparatus 11 includes a communication interface 22. In the figure, the interface is indicated as I / F. The information processing apparatus 11 may include an input device 23 and a display device 24.

[0022] The processor 20 controls the information processing apparatus 11. The processor 20 is configured to execute processing based on a program 26 stored in the memory 21. The processor 20 may be a CPU (Central Processing Unit), a GPU (Graphic Processing Unit), or an NPU (Neural network Processing Unit). The processor 20 may include an integrated circuit, such as an application-specific integrated circuit, or may be an integrated circuit. The processor 20 may be a combination of these.

[0023] The memory 21 is configured to store the program 26. The memory 21 is a non-transitory computer-readable medium that stores the program 26, but may include a transitory computer-readable medium. The program 26 may include a dedicated application for using the information processing system 10. The memory 21 stores a database 27.

[0024] The communication interface 22 is implemented as hardware, software, or a combination thereof. The communication interface 22 transmits and receives data to and from the translation server 12 and the LLM server 15.

[0025] The input device 23 inputs data according to a user's operation. The input device 23 may be a touch panel integrated with the display device 24. The input device 23 may be a pointing device with operation buttons. The display device 24 displays information according to an output instruction from the processor 20.

[0026] The translation server 12 is configured in the same way as the information processing device 11. Therefore, descriptions of the processor, memory, and communication interface included in the translation server 12 are omitted. The LLM server 15 is configured in the same way as the information processing device 11. Therefore, descriptions of the processor, memory, and communication interface included in the LLM server 15 are omitted.

[0027] <Data Configuration of Database 27> As shown in FIG. 2, an error management database 30 is stored in the memory 21 of the information processing device 11 as a database 27. The database 27 includes the error management database 30. The error management database 30 is a database for managing translation errors.

[0028] The error management database 30 includes at least one data set 30A. As the data set 30A, a source text identifier, a translated text identifier, and an evaluation value are associated. In the figure, the identifier is shown as ID. As the data set 30A, at least one error identifier is associated with the source text identifier. As the data set 30A, an error identifier, error content, error applicability, main error classification, error classification, group, and severity are associated. As the data set 30A, at least one error classification is associated with the error identifier.

[0029] The source text identifier is data for identifying the source text. The translated text identifier is data for identifying the translated text obtained by translating the source text. The evaluation value is data indicating the evaluation result of the translated text. The evaluation value is a value indicating the degree of translation error. The evaluation value may be a larger value as the translation error is more. That is, it is evaluated that the higher the translation accuracy is, the smaller the evaluation value is.

[0030] The error identifier is data for identifying a translation error. The error content is data indicating the content of the translation error. The error applicability re-determines the applicability of the translation error after it is determined as a translation error from the translated text, and is data indicating the result of the re-determination of the applicability of the translation error.

[0031] The main error classification is data indicating the main error classification for translation errors. The error classification is data indicating the error classification for translation errors. The main error classification is an appropriate error classification among at least one error classification.

[0032] Specifically, the error classification may include a first error classification, a second error classification, and a third error classification. As a specific example, the error classification may include accuracy, fluency, and style.

[0033] The error classification may be divided into multiple levels. The error classification may include a major classification and a minor classification. As a specific example, the major classification of the error classification may include accuracy, fluency, and style. The minor classification of accuracy may include mistranslation and omission.

[0034] A group is data indicating a group related to the error content of translation errors. A group is different from the error classification. Multiple translation errors with overlapping error content are grouped as the same group.

[0035] The severity is data indicating the severity of translation errors. The severity may include high severity, medium severity, and low severity. As a specific example, high severity includes translation errors that have a great impact on users. Low severity includes translation errors that do not change the meaning but degrade the quality of the sentence expression.

[0036] In this way, in the error management database 30, the evaluation value of the translated text and data related to translation errors are managed. Specifically, in the error management database 30, based on the result of detecting at least one translation error from the translated text, the analysis result of the translation error and the evaluation result of the translated text are managed.

[0037] As shown in FIG. 3, a coefficient database 31 is stored in the memory 21 of the information processing apparatus 11 as a database 27. The database 27 includes the coefficient database 31. The coefficient database 31 is a database regarding coefficients used when calculating an evaluation value of a translation error.

[0038] In the coefficient database 31, a combination of a main error classification and a severity level is associated with a coefficient. An evaluation value of a translated sentence is calculated based on an operation result of the number of translation errors corresponding to the combination of the main error classification and the severity level and the coefficient corresponding to the combination of the main error classification and the severity level.

[0039] The coefficients may be different for each of the first error classification, the second error classification, and the third error classification. The coefficient may be a value that is larger for the first error classification than for the second error classification, and may be a value that is larger for the second error classification than for the third error classification. The coefficient may be a value such that the evaluation value is larger for the first error classification than for the second error classification, and may be a value such that the evaluation value is larger for the second error classification than for the third error classification. Thus, the priority of evaluating translation errors is higher for the first error classification than for the second error classification. The priority of evaluating translation errors is higher for the second error classification than for the third error classification.

[0040] The coefficients may be different for each of severity, medium severity, and low severity. The coefficient may be a value that is larger for severity than for medium severity, and may be a value that is larger for medium severity than for low severity. The coefficient may be a value such that the evaluation value is larger for severity than for medium severity, and may be a value such that the evaluation value is larger for medium severity than for low severity. Thus, the priority of evaluating translation errors is higher for severity than for medium severity. The priority of evaluating translation errors is higher for medium severity than for low severity.

[0041] <Original translation processing> Next, with reference to FIG. 4, the original text translation process will be described. The original text translation process is executed by the processor 20 when a translation instruction for the original text is received. The translation instruction includes an original text identifier, a language after translation, and a translation server 12 that is the subject of translation. The translation instruction may include the language of the original text. The translation instruction includes an instruction to cause all the specified translation servers 12 to translate the original text into the specified language with the specified translation servers 12 as the subjects of translation.

[0042] As shown in FIG. 4, in step S10, the processor 20 executes a translation server determination process. In this process, the processor 20 determines the translation server 12 that is the subject of translation from the translation servers 12 included in the translation instruction.

[0043] The translation server 12 that is the subject of translation does not include the translation server 12 that has already been the subject of translation among the translation servers 12 included in the translation instruction. The translation server 12 that is the subject of translation is determined from the translation servers 12 that have not yet been the subject of translation among the translation servers 12 included in the translation instruction.

[0044] For example, when the first translation server 13 has already been the subject of translation and the second translation server 14 has not yet been the subject of translation among the first translation server 13 and the second translation server 14 included in the translation instruction, the processor 20 determines the second translation server 14 as the subject of translation. In this way, the processor 20 sequentially determines the translation servers 12 included in the translation instruction as the subjects of translation.

[0045] In step S11, the processor 20 executes an original text acquisition process. In this process, the processor 20 reads out the original text corresponding to the original text identifier included in the translation instruction from the memory 21. Thereby, the processor 20 acquires the original text based on the translation instruction.

[0046] In step S12, the processor 20 executes a translation prompt generation process. In this process, the processor 20 generates a translation prompt. The translation prompt includes the original text. The translation prompt may include the language of the original text. The translation prompt includes the language after translation. The translation prompt includes an instruction to translate the original text into the language after translation.

[0047] In step S13, the processor 20 executes a translation prompt input process. In this process, the processor 20 transmits the translation prompt generated in step S12 to the translation server 12 that is the subject of translation. The translation prompt is input into the translation language model in the translation server 12 that is the subject of translation. In this way, the processor 20 inputs the translation prompt into the translation language model of the translation server 12 that is the subject of translation.

[0048] In step S14, the processor 20 executes a translated text acquisition process. In this process, the processor 20 receives the translated text from the translation server 12 that is the subject of translation. Thereby, the processor 20 acquires the translated text. The processor 20 generates a translated text identifier and registers it in the error management database 30 so as to correspond to the original text identifier. The processor 20 stores the translated text in the memory 21 so as to correspond to the translated text identifier.

[0049] In step S15, the processor 20 determines whether all the translated texts have been acquired from all the translation servers 12 that are the subjects of translation. If the processor 20 determines that not all the translated texts have been acquired, the process proceeds to step S10. If the processor 20 determines that all the translated texts have been acquired, the original text translation process ends. In this way, the processor 20 repeatedly executes steps S10 to S14 until all the translated texts have been acquired from all the translation servers 12 that are the subjects of translation.

[0050] In this way, the processor 20 obtains a translated text by inputting the original text into a translation language model. Specifically, the processor 20 obtains a first translated text as the translated text by inputting the original text into the first translation language model 13A. The processor 20 obtains a second translated text as the translated text by inputting the original text into the second translation language model 14A.

[0051] <Translation text analysis process> Next, with reference to FIGS. 5 to 9, the translation text analysis process will be described. The error management process is executed by the processor 20 when an analysis instruction is received. The analysis instruction includes an original text identifier and at least one translated text identifier.

[0052] In step S20, the processor 20 executes an analysis target determination process. In this process, the processor 20 determines a translated text to be analyzed from at least one translated text identifier corresponding to the original text identifier.

[0053] The translated text to be analyzed does not include the translated text corresponding to the translated text identifier that has already been the analysis target among the translated text identifiers included in the analysis instruction. The translated text to be analyzed is determined from the translated texts corresponding to the translated text identifiers that have not yet been the analysis target among the translated text identifiers included in the analysis instruction. For example, when the first translated text has already been the analysis target and the second translated text has not been the analysis target among the first translated text and the second translated text included in the analysis instruction, the processor 20 determines the second translated text as the analysis target.

[0054] <Error determination process> In step S21, the processor 20 executes an error determination process. In this process, the processor 20 makes a determination regarding translation errors for the translated text determined as the analysis target.

[0055] As shown in FIG. 6, in the error determination process, in step S30, the processor 20 executes the original text acquisition process. In this process, the processor 20 acquires the original text corresponding to the original text identifier included in the analysis instruction from the memory 21. The processor 20 registers the original text identifier in the error management database 30.

[0056] In step S31, the processor 20 executes the translated text acquisition process. In this process, the processor 20 acquires the translated text corresponding to the translated text identifier determined as the analysis target from the memory 21. The processor 20 registers the translated text identifier in the error management database 30 so as to correspond to the original text identifier.

[0057] In step S32, the processor 20 executes the error determination prompt generation process. In this process, the processor 20 generates an error determination prompt. Specifically, the processor 20 generates an error determination prompt including the original text acquired in step S30, the translated text acquired in step S31, and an instruction regarding the determination of translation errors.

[0058] The instruction regarding the determination of translation errors includes extracting the error content that becomes at least one translation error included in the translated text with respect to the original text. The instruction regarding the determination of translation errors includes determining the error classification of at least one error content. The instruction regarding the determination of translation errors includes returning at least one error content in association with the error classification.

[0059] Thus, the error determination prompt includes the original text, the translated text obtained by translating the original text, and an instruction for determining the translation error in the translated text in association with the error classification. The error determination prompt corresponds to an example of the first prompt.

[0060] Instructions for translation error determination may include error guideline data for determining error classification. The error guideline data includes data indicating error classification. The error guideline data may include data indicating the major and minor classifications of error classification. For example, the error guideline data may include accuracy, fluency, and style as the major classifications of error classification.

[0061] In step S33, the processor 20 executes error determination prompt input processing. In this processing, the processor 20 transmits the error determination prompt generated in step S32 to the LLM 15A. The error determination prompt is input to the LLM 15A in the LLM server 15. In this way, the processor 20 inputs the error determination prompt to the LLM 15A of the LLM server 15.

[0062] In step S34, the processor 20 executes error determination result acquisition processing. In this processing, the processor 20 receives the error determination result from the LLM server 15. Thereby, the processor 20 acquires the error determination result. That is, the processor 20 acquires the translation error corresponding to the error classification by inputting the error determination prompt to the LLM 15A.

[0063] In particular, the processor 20 acquires, as an error determination result, the error content of at least one translation error included in the translated text associated with the error classification. The processor 20 may acquire, as an error determination result, the error content of a plurality of duplicate errors included in the translated text associated with the error classification.

[0064] In step S35, the processor 20 executes error determination result registration processing. In this processing, the processor 20 registers the error determination result acquired in step S34 in the error management database 30 so as to correspond to the original text identifier and the translation identifier.

[0065] Specifically, the processor 20 generates an error identifier corresponding to at least one translation error included in the translated text. The processor 20 registers the error content and error classification in the error management database 30 so as to correspond to the original text identifier, translation identifier, and error identifier.

[0066] In particular, even when there are overlapping error contents among a plurality of error contents, the processor 20 registers the error content and error classification in the error management database 30 so as to correspond to the original text identifier, translation identifier, and error identifier respectively.

[0067] In step S36, the processor 20 performs an error grouping process. In this process, when a plurality of error identifiers correspond to the translated text identifier determined as the analysis target, the processor 20 groups the translation errors.

[0068] When there are overlapping error contents among a plurality of error contents corresponding to a plurality of error identifiers respectively, the processor 20 updates the error management database 30 so as to associate the same group with the plurality of error identifiers having overlapping error contents. In this way, when the processor 20 obtains a plurality of translation errors from the original text, the overlapping translation errors among the plurality of translation errors are grouped.

[0069] In step S37, the processor 20 performs an error appropriateness prompt generation process. In this process, the processor 20 generates an error appropriateness prompt. Specifically, the processor 20 generates an error appropriateness prompt including the original text obtained in step S30, the translated text obtained in step S31, the error determination result obtained in step S34, and an instruction to determine the appropriateness of the error determination. The error appropriateness prompt may include an error identifier corresponding to the error determination result.

[0070] In particular, for a plurality of error determination results grouped as the same group, the processor 20 generates an error appropriateness prompt that includes one error determination result and does not include the remaining error determination results. That is, the processor 20 generates an error appropriateness prompt that includes the grouped error determination results.

[0071] The instruction to determine the appropriateness of the error determination includes re-determining whether the error determination result determined to be included in the translated text for the original text is appropriate. The instruction to determine the appropriateness of the error determination includes, as a result of the re-determination, replying whether the error determination result is appropriate.

[0072] In this way, the error appropriateness prompt includes the original text, the translated text, the translation error, and an instruction to determine the appropriateness of the translation error. The error appropriateness prompt corresponds to an example of the second prompt.

[0073] In step S38, the processor 20 executes error appropriateness prompt input processing. In this processing, the processor 20 transmits the error appropriateness prompt generated in step S37 to the LLM 15A. The error appropriateness prompt is input to the LLM 15A in the LLM server 15. In this way, the processor 20 inputs the error appropriateness prompt to the LLM 15A of the LLM server 15.

[0074] In step S39, the processor 20 executes error appropriateness result acquisition processing. In this processing, the processor 20 receives the error appropriateness result from the LLM server 15. Thereby, the processor 20 acquires the error appropriateness result. That is, the processor 20 acquires the appropriateness of the translation error by inputting the error appropriateness prompt to the LLM 15A.

[0075] In step S40, the processor 20 executes error applicability result registration processing. In this processing, the processor 20 registers the error applicability result obtained in step S39 in the error management database 30 so as to correspond to the original text identifier, the translated text identifier, and the error identifier. In particular, the processor 20 registers the same error applicability result in the error management database 30 so as to correspond to a plurality of error identifiers grouped as the same group.

[0076] <Error classification processing> As shown in FIG. 5, when the error determination processing ends, in step S22, the processor 20 executes error classification processing. In this processing, the processor 20 performs error classification of translation errors.

[0077] As shown in FIG. 7, in the error classification processing, in step S50, the processor 20 executes error classification prompt generation processing. In this processing, the processor 20 generates an error classification prompt.

[0078] Specifically, the processor 20 acquires from the error management database 30 the error identifier corresponding to the error applicability result for which the error determination has been appropriately performed. The processor 20 generates an error classification prompt including the error content and error classification corresponding to the error identifier, and an instruction to select an appropriate main error classification from the error classification.

[0079] In this case, the processor 20 may acquire from the error management database 30 the error identifiers corresponding to a plurality of error classifications among the error identifiers corresponding to the error applicability results for which the error determination is appropriate. The processor 20 does not necessarily have to acquire from the error management database 30 the error identifier corresponding to one error classification among the error identifiers corresponding to the error applicability results for which the error determination is appropriate. That is, the processor 20 may generate an error classification prompt when a plurality of error classifications are associated with the translation error.

[0080] In particular, for error classifications corresponding to a plurality of error identifiers grouped as the same group, the processor 20 generates an error classification prompt including a plurality of error classifications corresponding to the grouped plurality of error identifiers. That is, the processor 20 generates an error classification prompt including the grouped error classifications.

[0081] Thus, the error classification prompt includes a translation error, a plurality of error classifications, and an instruction to select an appropriate main error classification from the plurality of error classifications. The error classification prompt corresponds to an example of the third prompt.

[0082] In step S51, the processor 20 executes error classification prompt input processing. In this processing, the processor 20 transmits the error classification prompt generated in step S50 to the LLM 15A. The error classification prompt is input to the LLM 15A in the LLM server 15. Thus, the processor 20 inputs the error classification prompt to the LLM 15A of the LLM server 15.

[0083] In step S52, the processor 20 executes classification selection result acquisition processing. In this processing, the processor 20 receives the main error classification from the LLM server 15. Thereby, the processor 20 acquires the main error classification. That is, the processor 20 acquires the main error classification corresponding to the translation error by inputting the error classification prompt to the LLM 15A.

[0084] In step S53, the processor 20 executes classification selection result registration processing. In this processing, the processor 20 registers the acquired main error classification in the error management database 30 so as to correspond to the error identifier. In particular, the processor 20 registers the same main error classification in the error management database 30 so as to correspond to a plurality of error identifiers grouped as the same group.

[0085] <Severity determination processing> As shown in FIG. 5, when the error classification process ends, in step S23, the processor 20 executes a severity determination process. In this process, the processor 20 determines the severity of the translation error.

[0086] As shown in FIG. 8, in the severity determination process, in step S60, the processor 20 executes a severity prompt generation process. In this process, the processor 20 generates a severity prompt.

[0087] Specifically, the processor 20 obtains from the error management database 30 an error identifier corresponding to the error appropriateness result for which the error determination has been appropriately made. The processor 20 generates a severity prompt including the error content corresponding to the error identifier, the main error classification, and an instruction to determine the severity of the translation error.

[0088] In this way, the severity prompt includes the translation error, the main error classification, and an instruction to determine the severity of the translation error. The translation error may include the error content. The severity prompt corresponds to an example of the fourth prompt.

[0089] The severity prompt may include severity guideline data for determining the severity of the translation error. The severity guideline data is data indicating guidelines for determining the severity of the translation error. For example, the severity guideline data may include guidelines for determining that a translation error that has a great impact on the user is severe. For example, the severity guideline data may include guidelines for determining that a translation error that does not have a great impact on the user but has a different meaning is medium severity. For example, the severity guideline data may include guidelines for determining that a translation error that does not have a different meaning but has a degraded quality of the sentence expression is low severity.

[0090] In step S61, the processor 20 executes severity prompt input processing. In this process, the processor 20 sends the severity prompt generated in step S60 to the LLM 15A. The severity prompt is input to the LLM 15A in the LLM server 15. In this way, the processor 20 inputs the severity prompt to the LLM 15A of the LLM server 15.

[0091] In step S62, the processor 20 executes severity acquisition processing. In this process, the processor 20 receives the severity from the LLM server 15. Thereby, the processor 20 acquires the severity. That is, the processor 20 acquires the severity of the translation error by inputting the severity prompt to the LLM 15A.

[0092] In step S63, the processor 20 executes severity registration processing. In this process, the processor 20 registers the acquired severity in the error management database 30 so as to correspond to the error identifier. In particular, the processor 20 registers the same severity in the error management database 30 so as to correspond to a plurality of error identifiers grouped as the same group.

[0093] <Evaluation control process> As shown in FIG. 5, when the severity determination process ends, in step S24, the processor 20 executes evaluation control processing. In this process, the processor 20 evaluates the translated text. That is, the processor 20 evaluates the translation server 12. It can also be said that the processor 20 evaluates the translation language model.

[0094] As shown in FIG. 9, in the evaluation control process, in step S70, the processor 20 executes character count acquisition processing. In this process, the processor 20 counts the number of characters in the translated text to be analyzed. In this way, the processor 20 acquires the number of characters in the translated text.

[0095] In step S71, the processor 20 executes an evaluation value calculation process. In this process, the processor 20 calculates an evaluation value of the translation text to be analyzed. The processor 20 may calculate the evaluation value of the translation text to be analyzed by any operation.

[0096] Specifically, the processor 20 refers to the error management database 30 and calculates the number of translation errors corresponding to the combination of the main error classification and severity for each error identifier corresponding to the translation text identifier to be analyzed.

[0097] The processor 20 refers to the coefficient database 31 and reads out the coefficients corresponding to the combination of the main error classification and severity from the memory 21. The processor 20 multiplies the number of translation errors by the coefficient for each combination of the main error classification and severity.

[0098] The processor 20 adds up the multiplication results for each combination of the main error classification and severity. The processor 20 calculates the result of dividing the addition result by the number of characters of the translation text as the evaluation value.

[0099] In this way, the processor 20 calculates an evaluation value for the translation text based on the translation error, the main error classification, and the severity. In particular, the processor 20 calculates an evaluation value for the first translation text and an evaluation value for the second translation text, respectively.

[0100] Specifically, the processor 20 calculates the evaluation value based on the calculation result of the number of translation errors included in the translation text and the coefficient corresponding to the main error classification and severity. The processor 20 calculates the evaluation value based on the number of characters of the translation text.

[0101] In step S72, the processor 20 executes an evaluation value registration process. In this process, the processor 20 registers the calculated evaluation value in the error management database 30 so as to correspond to the translation text identifier.

[0102] As shown in FIG. 5, when the evaluation control process ends, in step S25, the processor 20 determines whether all the translation texts to be analyzed have been analyzed. If the processor 20 determines that not all the translation texts to be analyzed have been analyzed, the process proceeds to step S20. If the processor 20 determines that all the translation texts to be analyzed have been analyzed, the translation text analysis process ends. In this way, the processor 20 repeatedly executes steps S20 to S24 until all the translation texts to be analyzed have been analyzed.

[0103] <Display control process> Next, the display control process will be described with reference to FIG. 10. The display control process is executed by the processor 20 at a predetermined cycle.

[0104] In step S80, the processor 2 determines whether there is a display instruction. If the processor 20 determines that there is no display instruction, the display control process ends. If the processor 20 determines that there is a display instruction, the process proceeds to step S81.

[0105] The display instruction includes the translation text that is the display target for displaying the analysis result. That is, the display instruction includes the translation server that is the display target. It can also be said that the display instruction includes the translation language model that is the display target.

[0106] The display instruction includes the display mode of the analysis result. The display mode of the analysis result is a mode in which the evaluation value is displayed for each translation text that is the display target. The display mode of the analysis result includes a first display mode and a second display mode.

[0107] The first display mode is a mode in which the number of translation errors corresponding to the severity is displayed for each translation text that is the display target. The second display mode is a mode in which the number of translation errors corresponding to the main error classification is displayed for each translation text that is the display target.

[0108] In step S81, the processor 20 executes analysis result display processing. In this processing, the processor 20 causes the display device 24 to display an image related to a translation error so as to correspond to the display instruction.

[0109] Specifically, the processor 20 refers to the error management database 30 corresponding to the translation text to be displayed included in the display instruction. Thereby, the processor 20 acquires data related to the translation error corresponding to the translation text to be displayed included in the display instruction.

[0110] The processor 20 causes the display device 24 to display an image related to a translation error so as to correspond to the display mode of the analysis result included in the display instruction. When the first display mode is included in the display instruction, for each translation text to be displayed, the processor 20 causes the display device 24 to display an image related to the evaluation value and an image related to the number of translation errors corresponding to the severity. When the second display mode is included in the display instruction, for each translation text to be displayed, the processor 20 causes the display device 24 to display an image related to the evaluation value and an image related to the number of translation errors corresponding to the main error classification.

[0111] As shown in FIG. 11, when the first display mode is included in the display instruction, a first image 24A is displayed on the display device 24. In the first image 24A, an image indicating the evaluation value for each translation text is displayed. In the first image 24A, for each translation text, an image indicating the number of errors with a low severity, the number of errors with a medium severity, and the number of errors with a high severity is displayed.

[0112] When the second display mode is included in the display instruction, a second image 24B is displayed on the display device 24. In the second image 24B, an image indicating the evaluation value for each translation text is displayed. In the second image 24B, for each translation text, an image indicating the number of errors whose main error classification is the first error classification, the number of errors whose main error classification is the second error classification, and the number of errors whose main error classification is the third error classification is displayed. An example of the first error classification may be accuracy, an example of the second error classification may be fluency, and an example of the third error classification may be style.

[0113] In this way, the processor 20 causes the display device 24 to display an image related to a translation error. In particular, the processor 20 causes the display device 24 to display an image related to an evaluation value corresponding to the translated text. The processor 20 causes the display device 24 to display an image related to a translation error associated with the main error classification. The processor 20 causes the display device 24 to display an image related to a translation error associated with the severity.

[0114] <Actions and Effects of the First Embodiment> The actions and effects of the first embodiment will be described. (1) The processor 20 generates an error determination prompt including the original text, the translated text, and an instruction to associate a translation error in the translated text with an error classification. By inputting the error determination prompt into the LLM 15A, the processor 20 obtains a translation error corresponding to the error classification. According to this configuration, the translation error can be obtained from the LLM 15A corresponding to the error classification. Thereby, the opportunity for human judgment regarding whether it is a translation error and the error classification of the translation error can be reduced. Therefore, the manual effort can be reduced. Thus, the convenience regarding the evaluation of translation accuracy can be improved.

[0115] In addition to this, the accuracy of the evaluation can be improved as compared with the opportunity for human judgment regarding whether it is a translation error and the error classification of the translation error. Thus, the convenience regarding the evaluation of translation accuracy can be improved.

[0116] (2) The processor 20 generates an error suitability prompt that includes a translation error and an instruction to determine the suitability of the translation error. The processor 20 obtains the suitability of the translation error by inputting the error suitability prompt to the LLM 15A. According to this configuration, after determining that it is a translation error by the LLM 15A, the suitability of the translation error can be determined again by the LLM 15A. Thereby, the opportunity for human judgment regarding whether it is a translation error can be reduced, and the determination accuracy regarding whether it is a translation error can be improved. For this reason, the manual labor can be reduced. Therefore, the convenience regarding the evaluation of translation accuracy can be improved.

[0117] (3) When the processor 20 obtains a plurality of translation errors from the original text, it executes grouping the duplicate translation errors among the plurality of translation errors. According to this configuration, by grouping the duplicate translation errors, the control load regarding the duplicate translation errors can be reduced. Therefore, the convenience regarding the evaluation of translation accuracy can be improved.

[0118] (4) When a plurality of error classifications are associated with a translation error, the processor 20 generates an error classification prompt that includes the translation error, the plurality of error classifications, and an instruction to select an appropriate main error classification from the plurality of error classifications. The processor 20 obtains an appropriate main error classification corresponding to the translation error by inputting the error classification prompt to the LLM 15A. According to this configuration, even when a plurality of error classifications are associated with a translation error, an appropriate main error classification corresponding to the translation error can be obtained from the LLM 15A. Thereby, the opportunity for human judgment regarding an appropriate main error classification corresponding to the translation error can be reduced. For this reason, the manual labor can be reduced. Therefore, the convenience regarding the evaluation of translation accuracy can be improved.

[0119] (5) The processor 20 generates a severity prompt that includes a translation error, a main error classification, and an instruction to determine the severity of the translation error. The processor 20 obtains the severity of the translation error by inputting the severity prompt into the LLM 15A. According to this configuration, the severity of the translation error can be obtained from the LLM 15A. Thereby, the opportunity for human judgment regarding the severity of the translation error can be reduced. For this reason, the manual effort can be reduced. Therefore, the convenience regarding the evaluation of translation accuracy can be improved.

[0120] (6) The severity prompt includes guidelines for determining the severity of the translation error. According to this configuration, by inputting the guidelines for determining the severity of the translation error into the LLM 15A, the severity of the translation error can be determined as intended by the evaluator. Therefore, the convenience regarding the evaluation of translation accuracy can be improved.

[0121] (7) The processor 20 calculates an evaluation value for the translated text based on the translation error, the main error classification, and the severity. According to this configuration, an objective evaluation can be performed based on the evaluation value for the translated text. Therefore, the convenience regarding the evaluation of translation accuracy can be improved.

[0122] (8) The processor 20 calculates the evaluation value based on the calculation result of the number of translation errors included in the translated text and the coefficient corresponding to the main error classification and the severity. According to this configuration, based on the number of translation errors included in the translated text, an objective evaluation can be performed using the evaluation value for the translated text. In addition to this, the evaluation value for the translated text can be weighted by the coefficient corresponding to the main error classification and the severity. Therefore, the convenience regarding the evaluation of translation accuracy can be improved.

[0123] (9) The processor 20 calculates the evaluation value based on the number of characters in the translated text. According to this configuration, the evaluation value for the translated text can be calculated considering the number of characters in the translated text. Therefore, the convenience regarding the evaluation of translation accuracy can be improved.

[0124] (10) The processor 20 obtains a first translated sentence by inputting the original text into the first translation language model 13A. The processor 20 obtains a second translated sentence by inputting the original text into the second translation language model 14A. The processor 20 calculates an evaluation value for the first translated sentence and an evaluation value for the second translated sentence, respectively. According to this configuration, the first translated sentence and the second translated sentence can be obtained by inputting the same original text into both the first translation language model 13A and the second translation language model 14A. In addition to this, the evaluation value of the first translated sentence and the evaluation value of the second translated sentence can be compared and objectively evaluated. Therefore, the convenience regarding the evaluation of translation accuracy can be improved.

[0125] (11) The processor 20 causes the display device 24 to display an image regarding a translation error associated with the main error classification. According to this configuration, the evaluator can be made to recognize the analysis result regarding the translation error associated with the main error classification. Therefore, the convenience regarding the evaluation of translation accuracy can be improved.

[0126] (12) The processor 20 causes the display device 24 to display an image regarding a translation error associated with the severity. According to this configuration, the evaluator can be made to recognize the analysis result regarding the translation error associated with the severity. Therefore, the convenience regarding the evaluation of translation accuracy can be improved.

[0127] (13) The processor 20 causes the display device 24 to display an image regarding the evaluation value. According to this configuration, the evaluator can be made to recognize the analysis result regarding the evaluation value for the translated sentence. Therefore, the convenience regarding the evaluation of translation accuracy can be improved.

[0128] [Modification Example] This embodiment can be implemented with the following modifications. This embodiment and the following modification examples can be implemented in combination with each other within a technically non - conflicting range.

[0129] · The processor 20 may generate a grouping prompt including the error content of the translation error and an instruction to group the translation errors. The processor 20 may obtain the result of grouping the translation errors from the LLM15A by inputting the grouping prompt into the LLM15A.

[0130] · The processor 20 may analyze each of the translation errors with duplicate error content without grouping the translation errors. The processor 20 may evaluate each of the translation errors with duplicate error content without grouping the translation errors.

[0131] · The processor 20 may not re-determine the appropriateness of the translation error after it is determined as a translation error from the translated text. The processor 20 may determine the severity of the translation error based on the result of determining the translation error from the translated text. The processor 20 may calculate an evaluation value based on the result of determining the translation error from the translated text.

[0132] · When one type of error classification corresponds to the error identifier, the processor 20 may not execute step S22 in FIG. 5. That is, when a plurality of error classifications correspond to the error identifier, the processor 20 may at least generate an error classification prompt and obtain the main error classification corresponding to the translation error by inputting the error classification prompt into the LLM15A.

[0133] · Even if the error identifier corresponds to multiple error classifications, the processor 20 may not execute step S22 in FIG. 5. In this case, the processor 20 may analyze the translated text based on at least one error classification corresponding to the error identifier instead of the main error classification. For example, the processor 20 may determine the severity of the translation error based on at least one error classification corresponding to the error identifier instead of the main error classification. For example, the processor 20 may calculate an evaluation value of the translation error based on at least one error classification corresponding to the error identifier instead of the main error classification. When a plurality of error classifications correspond to one error identifier, the processor 20 may calculate an evaluation value of the translation error for each of the plurality of error classifications and calculate the average of the evaluation values as the evaluation value corresponding to one error identifier. For example, the processor 20 may cause the analysis result of the translation error to be displayed on the display device 24 based on at least one error classification corresponding to the error identifier instead of the main error classification.

[0134] · The processor 20 may calculate an evaluation value of the translated text regardless of the main error classification. In this case, the processor 20 may not obtain the main error classification. The processor 20 may calculate an evaluation value of the translated text regardless of the severity of the translation error. In this case, the processor 20 may not obtain the severity of the translation error.

[0135] · The coefficient database 31 may include main error classifications with the same coefficient. The coefficient database 31 may include severities with the same coefficient. The coefficient database 31 may include a coefficient based on either the main error classification or the severity. The coefficient database 31 may include another element other than the main error classification and the severity. For example, the other element may be the number of characters in the translated text.

[0136] · The processor 20 may calculate the evaluation value of the translated text regardless of the coefficient. In this case, the coefficient database 31 may not be stored in the memory 21. The processor 20 may calculate the evaluation value of the translated text based on the number of characters in the original text. The processor 20 may calculate the evaluation value of the translated text regardless of the number of characters in the translated text.

[0137] · The processor 20 may compare the evaluation values for a plurality of translated texts. The processor 20 may cause the display device 24 to display the comparison result of the evaluation values for the plurality of translated texts. The processor 20 may cause the display device 24 to display the number of characters of the translated text for the plurality of translated texts. The processor 20 may compare the evaluation values for a plurality of translated texts and upload the translated text with a smaller evaluation value to a web server (not shown).

[0138] · The processor 20 may obtain the original text from a web server (not shown). The processor 20 may not execute the original text translation process of obtaining the translated text from the original text. In such a case, the processor 20 may obtain the translated text from another server.

[0139] · The processor 20 may divide one prompt into a plurality of prompts and input them to the LLM 15A. The processor 20 may integrate a plurality of prompts into one prompt and input them to the LLM 15A.

[0140] · The information processing system 10 may not use the translated text obtained by translating the original text as an analysis target. The information processing system 10 may not use the summarized text obtained by summarizing the original text as an analysis target. That is, the information processing system 10 may not use the edited text obtained by editing the original text as an analysis target.

[0141] In such a case, the translation server 12 may be an editing server, the first translation server 13 may be a first editing server, and the second translation server 14 may be a second editing server. The first translation language model 13A may be the first editing language model, and the second translation language model 14A may be the first editing language model.

[0142] The processor 20 may obtain an edited text by inputting a prompt including the original text, editing pointer data which is a pointer for editing the original text, and an instruction to edit the original text, into an editing language model. The processor 20 may obtain a determination result of an editing error by inputting a prompt including the original text, the edited text, the editing pointer data, and an instruction to determine the editing error, into the LLM 15A. In this case, the processor 20 may obtain a determination result of an error classification and a corresponding editing error by inputting a prompt including an instruction to determine the error classification and the corresponding editing error, into the LLM 15A.

[0143] · The translation server 12 may be provided with the LLM 15A. In this case, the destination of a prompt other than the translation prompt from the information processing apparatus 11 becomes the translation server 12. The processor 20 may input a prompt regarding the first translated text into the LLM 15A of the second translation server 14, and input a prompt regarding the second translated text into the LLM 15A of the first translation server 13.

[0144] · The information processing apparatus 11 may be provided with the LLM 15A. Thus, in the information processing apparatus 11, the processor 20 may input a prompt into the LLM 15A. In such cases, the information processing system 10 may not be provided with the LLM server 15.

[0145] · In the information processing system 10, the processor 20 may input a prompt into a small-scale language model whose computational amount, data amount, and number of model parameters are smaller than those of the LLM 15A, instead of the LLM 15A. That is, in the information processing system 10, the processor 20 may input a prompt into a language model. The language model may be a learning model using deep learning.

[0146] ·The information processing apparatus 11 may not include at least either the input device 23 or the display device 24. The information processing apparatus 11 may be communicably connected to a terminal device (not shown) via the communication interface 22. The terminal device may include at least either the input device 23 or the display device 24. In this way, the processor 20 may cause various images to be displayed on the display device 24 of the terminal device via the communication interface 22. The processor 20 may input various instructions from the input device 23 of the terminal device via the communication interface 22.

[0147] ·The information processing apparatus 11 may include at least one processor 20 and at least one memory 21. The information processing system 10 may include at least one processor 20 and at least one memory 21.

[0148] ·The information processing apparatus 11 may be composed of a plurality of servers. In this case, the plurality of servers are communicably connected. The plurality of servers may each include a divided function of the information processing apparatus 11.

[0149] ·The information processing system 10 may include, in addition to the translation server 12 and the LLM server 15, another server as long as it includes at least the information processing apparatus 11. The information processing system 10 may not include at least any of the servers such as the translation server 12 and the LLM server 15 as long as it includes at least the information processing apparatus 11.

[0150] ·The expression "at least either" used in this specification means one or more of the desired options. As an example, the expression "at least either" used in this specification means only one option or both options if the number of options is two. As another example, the expression "at least either" used in this specification means only one option or any combination of two or more options if the number of options is three or more.

[0151] [Appendix] The following describes the technical idea grasped from the above-described embodiments and modification examples. [1] The information processing apparatus includes at least one memory configured to store a program, and at least one processor configured to execute processing based on the program. The at least one processor generates a first prompt including the original text, an edited text obtained by editing the original text, and an instruction to determine an editing error in the edited text in association with an error classification, and inputs the first prompt into a language model to obtain the editing error corresponding to the error classification.

[0152] [2] The information processing apparatus according to [1], wherein the at least one processor generates a second prompt including the editing error and an instruction to determine the appropriateness of the editing error, and inputs the second prompt into a language model to obtain the appropriateness of the editing error.

[0153] [3] The information processing apparatus according to [1] or [2], wherein when the at least one processor obtains a plurality of the editing errors from the original text, the at least one processor groups the overlapping editing errors among the plurality of editing errors.

[0154] [4] The information processing apparatus according to any one of [1] to [3], wherein when a plurality of the error classifications are associated with the editing error, the at least one processor generates a third prompt including the editing error, the plurality of error classifications, and an instruction to select an appropriate error classification from the plurality of error classifications, and inputs the third prompt into a language model to obtain an appropriate error classification corresponding to the editing error.

[0155] The information processing apparatus according to any one of [1] to [4], wherein the at least one processor generates a fourth prompt including the editing error, the error classification, and an instruction to determine the severity of the editing error, and obtains the severity of the editing error by inputting the fourth prompt into a language model.

[0156] The information processing apparatus according to [5] of [6], wherein the fourth prompt includes a guideline for determining the severity of the editing error. The information processing apparatus according to [5] or [6] of [7], wherein the at least one processor calculates an evaluation value for the edited text based on the editing error, the error classification, and the severity.

[0157] The information processing apparatus according to [7] of [8], wherein calculating the evaluation value includes calculating the evaluation value based on the result of an operation between the number of editing errors included in the edited text and a coefficient corresponding to the error classification and the severity.

[0158] The information processing apparatus according to [8] of [9], wherein calculating the evaluation value includes calculating the evaluation value based on the number of characters in the edited text. The information processing apparatus according to any one of [7] to [9] of

[10] , wherein the at least one processor obtains a first translated text as the edited text by inputting the original text into a first machine translation model, and obtains a second translated text as the edited text by inputting the original text into a second machine translation model, and calculating the evaluation value includes calculating the evaluation value for the first translated text and the evaluation value for the second translated text, respectively.

[0159] The information processing apparatus according to any one of [1] to

[10] of

[11] , wherein the at least one processor causes a display device to display an image related to the editing error associated with the error classification.

[0160] An information processing apparatus according to any one of

[12] [5] to

[10] , wherein the at least one processor causes a display device to display an image related to the editing error associated with the severity.

[0161] An information processing apparatus according to any one of

[13] [7] to

[10] , wherein the at least one processor causes a display device to display an image related to the evaluation value. An information processing apparatus according to any one of

[14] [1] to

[13] , wherein the edited text is a translated text obtained by translating the original text.

[0162]

[15] An information processing system includes at least one memory configured to store a program, and at least one processor configured to execute processing based on the program. The at least one processor generates a first prompt including an original text, an edited text obtained by editing the original text, and an instruction to determine an editing error in the edited text in association with an error classification, and obtains the editing error corresponding to the error classification by inputting the first prompt to a language model.

[0163]

[16] An information processing method includes: generating, by at least one processor, a first prompt including an original text, an edited text obtained by editing the original text, and an instruction to determine an editing error in the edited text in association with an error classification; and obtaining the editing error corresponding to the error classification by inputting the first prompt to a language model.

[0164]

[17] A program causes at least one processor to generate a first prompt including an original text, an edited text obtained by editing the original text, and an instruction to determine an editing error in the edited text in association with an error classification, and obtain the editing error corresponding to the error classification by inputting the first prompt to a language model.

Description of Reference Numerals

[0165] 10... Information processing system, 11... Information processing device, 12... Translation server, 13... First translation server, 13A... First translation language model, 14... Second translation server, 14A... Second translation language model, 15... Large language model server, 15A... Large language model, 20... Processor, 21... Memory, 24... Display device, 24A... First image, 24B... Second image, 26... Program, 27... Database, 30... Error management database, 31... Coefficient database.

Claims

1. At least one memory configured to store a program, At least one processor configured to execute processing based on the program, and The at least one processor Generates a first prompt including the original text, a translation of the original text, error pointer data for determining an error classification of a translation error, and an instruction to associate the translation error in the translation with the error classification; By inputting the first prompt into a language model having a function of detecting the translation error based on the original text and the translation, obtaining the translation error corresponding to the error classification; When a plurality of the translation errors are obtained from the original text, grouping the overlapping translation errors among the plurality of translation errors; When a plurality of the error classifications are associated with the grouped translation errors, generating a third prompt including the translation error, the plurality of error classifications, and an instruction to select an appropriate error classification from the plurality of error classifications; By inputting the third prompt into the language model, obtaining an appropriate error classification corresponding to the translation error; Performs An information processing apparatus.

2. The information processing apparatus according to claim 1, wherein The at least one processor Generates a second prompt including the translation error and an instruction to determine the propriety of the translation error; By inputting the second prompt into the language model, obtaining the propriety of the translation error; Performs An information processing apparatus.

3. The information processing apparatus according to claim 1, wherein The at least one processor Generates a fourth prompt including the translation error, the error classification, and an instruction to determine the severity of the translation error; By inputting the fourth prompt into the language model, obtaining the severity of the translation error; Performs An information processing apparatus.

4. The information processing apparatus according to claim 3, wherein The fourth prompt includes a guideline for determining the severity of the translation error, An information processing apparatus.

5. The information processing apparatus according to claim 3, wherein The at least one processor executes calculating an evaluation value for the translated sentence based on the translation error, the error classification, and the severity. An information processing apparatus. **Claim 6** The information processing apparatus according to claim 5, calculating the evaluation value includes calculating the evaluation value based on a calculation result of the number of the translation errors included in the translated sentence and a coefficient corresponding to the error classification and the severity. An information processing apparatus. **Claim 7** The information processing apparatus according to claim 6, calculating the evaluation value includes calculating the evaluation value based on the number of characters of the translated sentence. An information processing apparatus. **Claim 8** The information processing apparatus according to claim 5, the at least one processor obtains a first translated sentence as the translated sentence by inputting the original text into a first machine translation model, and obtains a second translated sentence as the translated sentence by inputting the original text into a second machine translation model, and executes calculating the evaluation value includes calculating an evaluation value for the first translated sentence and an evaluation value for the second translated sentence, respectively. An information processing apparatus. **Claim 9** The information processing apparatus according to claim 1, the at least one processor executes causing a display device to display an image related to the translation error associated with the error classification. An information processing apparatus. **Claim 10** The information processing apparatus according to any one of claims 3 to 8, the at least one processor executes causing a display device to display an image related to the translation error associated with the severity. An information processing apparatus. **Claim 11** The information processing apparatus according to any one of claims 5 to 8, the at least one processor executes causing a display device to display an image related to the evaluation value. An information processing apparatus. **Claim 12** [[ID=3l]]at least one memory configured to store a program, and at least one processor configured to execute processing based on the program, the at least one processor generates a first prompt including an original text, a translated sentence obtained by translating the original text, error pointer data for determining an error classification of a translation error, and an instruction for associating and determining the translation error in the translated sentence with the error classification. Inputting the first prompt into a language model having a function of detecting the translation error based on the original text and the translated text to obtain the translation error corresponding to the error classification; When obtaining a plurality of the translation errors from the original text, grouping the overlapping translation errors among the plurality of translation errors; When a plurality of the error classifications are associated with the grouped translation errors, generating a third prompt including the translation error, the plurality of error classifications, and an instruction to select an appropriate error classification from the plurality of error classifications; Inputting the third prompt into the language model to obtain an appropriate error classification corresponding to the translation error; Executing; An information processing system.

13. At least one processor: Generating a first prompt including an original text, a translated text obtained by translating the original text, error pointer data for determining an error classification of a translation error, and an instruction to associate and determine the translation error in the translated text with the error classification; Inputting the first prompt into a language model having a function of detecting the translation error based on the original text and the translated text to obtain the translation error corresponding to the error classification; When obtaining a plurality of the translation errors from the original text, grouping the overlapping translation errors among the plurality of translation errors; When a plurality of the error classifications are associated with the grouped translation errors, generating a third prompt including the translation error, the plurality of error classifications, and an instruction to select an appropriate error classification from the plurality of error classifications; Inputting the third prompt into the language model to obtain an appropriate error classification corresponding to the translation error; Executing; An information processing method.

14. To at least one processor: Generating a first prompt including an original text, a translated text obtained by translating the original text, error pointer data for determining an error classification of a translation error, and an instruction to associate and determine the translation error in the translated text with the error classification; By inputting the first prompt into a language model having a function of detecting the translation error based on the original text and the translated text, obtaining the translation error so as to correspond to the error classification; When obtaining a plurality of the translation errors from the original text, grouping the overlapping translation errors among the plurality of translation errors; When a plurality of the error classifications are associated with the grouped translation errors, generating a third prompt including the translation error, the plurality of error classifications, and an instruction to select an appropriate error classification from the plurality of error classifications; By inputting the third prompt into the language model, obtaining an appropriate error classification corresponding to the translation error; To execute, Program.

Images