Information processing system, information processing method, and program

Abstract

To provide an information processing system, information processing method and program.SOLUTION: An information processing system is configured to execute: extracting one or more pre-translation terms relating to one or more original texts and one or more post-translation terms respectively corresponding to the one or more pre-translation terms from a bilingual dictionary including a plurality of pre-translation terms written in a first language and a plurality of post-translation terms written in a second language respectively corresponding to the plurality of pre-translation terms (S103); inputting a translating prompt into a language model (S105), where the translating prompt includes the one or more original texts, the one or more pre-translation terms, the one or more post-translation terms, and a translation instruction, the translating instruction including an instruction for causing the language model to translate the one or more pre-translation terms into the one or more post-translation terms; and acquiring, from the language model, a translated text obtained by translating the one or more original texts into the second language according to the translation instruction (S106).SELECTED DRAWING: Figure 5

Description

[Technical Field]

[0001] The present disclosure relates to an information processing system, an information processing method, and a program. [Background technology]

[0002] In recent years, machine translation systems using language models (LM) have been provided. A language model is a language model specialized for natural language processing (NLP) and is a type of generative AI (artificial intelligence). Generally, when a language model receives an instruction sentence called a prompt as input, it outputs a sentence called a completion that is generated based on the instruction.

[0003] Since such language models do not take into account the meaning of the entire source text when translating, the same word in the source text may be translated into multiple different words. For example, Patent Document 1 discloses a technology that uses a bilingual dictionary that associates pre-translation terms with post-translation terms in order to unify translations. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2009-026100 Summary of the Invention [Problem to be solved by the invention]

[0005] However, some pre-translation terms contained in a bilingual dictionary are not included in the source text to be translated. Pre-translation terms not included in a source text are not used in the machine translation of that source text. Therefore, if a prompt includes all bilingual terms, more data than necessary will be input to the language model. As a result, unnecessary calculations will be performed when the language model translates.

[0006] An object of the present disclosure is to provide an information processing system, an information processing method, and a program that can appropriately input a bilingual dictionary into a language model when performing machine translation using the language model. [Means for solving the problem]

[0007] An information processing system according to one aspect of the present disclosure is an information processing system configured to translate one or more original sentences written in a first language into a second language using a language model, the information processing system including one or more memories configured to store a program and one or more processors configured to execute processing based on the program, wherein the one or more processors select one or more terms related to the one or more original sentences from a bilingual dictionary including a plurality of pre-translation terms written in the first language and a plurality of translated terms written in the second language corresponding to the plurality of pre-translation terms, respectively. The system is configured to: extract the above pre-translation terms and one or more translated terms corresponding to the one or more pre-translation terms; input a translation prompt to the language model, wherein the translation prompt includes the one or more source sentences, the one or more pre-translation terms, the one or more translated terms, and a translation instruction, and the translation instruction includes an instruction for causing the language model to translate the one or more pre-translation terms into the one or more translated terms; and obtain from the language model a translation of the one or more source sentences into the second language in accordance with the translation instruction.

[0008] An information processing method according to one embodiment of the present disclosure is an information processing method configured to translate an original text written in a first language into a second language using a language model, the information processing method including causing one or more processors to execute the following operations: extracting one or more pre-translation terms related to the one or more original texts and one or more post-translation terms corresponding to the one or more pre-translation terms from a bilingual dictionary containing a plurality of pre-translation terms written in the first language and a plurality of post-translation terms written in the second language corresponding to the one or more pre-translation terms; inputting a translation prompt into the language model, the translation prompt including the one or more original texts, the one or more pre-translation terms, the one or more post-translation terms, and a translation instruction, the translation instruction including an instruction to cause the language model to translate the one or more pre-translation terms into the one or more post-translation terms; and obtaining from the language model a translation of the one or more original texts into the second language in accordance with the translation instruction.

[0009] A program according to one embodiment of the present disclosure is a program for translating one or more original sentences written in a first language into a second language using a language model, the program including: program code for extracting, from a bilingual dictionary containing a plurality of pre-translation terms written in the first language and a plurality of post-translation terms written in the second language corresponding to the plurality of pre-translation terms, one or more pre-translation terms related to the one or more original sentences and one or more post-translation terms corresponding to the one or more pre-translation terms; program code for inputting a translation prompt into the language model, the translation prompt including the one or more original sentences, the one or more pre-translation terms, the one or more post-translation terms, and a translation instruction, the translation instruction including an instruction to cause the language model to translate the one or more pre-translation terms into the one or more post-translation terms; and program code for obtaining, from the language model, a translation of the one or more original sentences into the second language in accordance with the translation instruction. [Brief explanation of the drawings]

[0010] [Figure 1]FIG. 1 is a schematic diagram of an information processing system according to an embodiment. [Figure 2] FIG. 2 is a diagram illustrating the configuration of the bilingual dictionary included in the information processing system of FIG. [Figure 3] FIG. 3 is a diagram illustrating the configuration of the terminal device and the information processing device included in the information processing system of FIG. [Figure 4] FIG. 4 is a diagram illustrating an information processing method performed by the information processing system of FIG. [Figure 5] FIG. 5 is a diagram illustrating an information processing method performed by the information processing system of FIG. [Figure 6] FIG. 6 is a diagram showing a method for outputting a translation by the information processing system of FIG. [Figure 7] FIG. 7 is a diagram illustrating another configuration of the terminal device and the information processing device included in the information processing system of FIG. [Figure 8] FIG. 8 is a diagram showing another information processing method by the information processing system of FIG. DETAILED DESCRIPTION OF THE INVENTION

[0011] Examples of the information processing system 11, information processing method, and program of the present disclosure will be described with reference to Figures 1 to 6. The present invention is not limited to these examples, but is defined by the claims, and is intended to include all modifications within the meaning and scope equivalent to the claims.

[0012] [Overall configuration of information processing system] As shown in FIG. 1, an information processing system 11 is configured to translate one or more original texts written in a first language into a second language using a language model. In the present disclosure, the first language is Japanese and the second language is English. The combination of the first language and the second language is not limited to Japanese and English, and may be any other language combination. The first language and the second language may have a relationship that has differences depending on the region in which they are used, such as English (US) and English (UK).

[0013] The information processing system 11 includes one or more information processing devices 20. The information processing device 20 may be realized as a computer including, for example, one or more processors 21, one or more memories 22, and a communication interface (IF) 23. The configurations of the multiple information processing devices 20 may be partially or entirely the same, or may be different from each other.

[0014] The one or more memories 22 store programs 25 and data for realizing various functions. The programs 25 include applications and an operating system. The one or more processors 21 execute processes based on the programs 25 to realize various functions. The communication interface (IF) 23 enables communication with other devices via the network 12. The network 12 includes, for example, the Internet, a wide area network (WAN), a local area network (LAN), a provider terminal, a wireless communication network, a wireless base station, a dedicated line, etc.

[0015] The one or more information processing devices 20 may be capable of communicating with one or more terminal devices 30 via the network 12. The one or more terminal devices 30 may be realized as a computer including one or more processors 31, one or more memories 32, and a communication interface (IF) 33. The one or more processors 31 realize various functions by executing processing based on a program 35. The communication interface 33 enables communication with other devices via the network 12. The one or more terminal devices 30 may include a monitor 34. In the information processing system 11, the monitor 34 is configured to display information generated by the information processing system 11.

[0016] The processors 21 and 31 are, for example, a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor unit (MPU), a field-programmable gate array (FPGA), or other arithmetic devices. The processors 21 and 31 are, for example, processing circuits configured to execute various software processes. The processing circuits may include dedicated hardware circuits (e.g., ASICs) that process at least part of the software processes. In other words, the software processes may be executed by processing circuitry that includes at least one or more software processing circuits and one or more dedicated hardware circuits.

[0017] The memories 22 and 32 may be non-transitory computer-readable media. The memories 22 and 32 may include, for example, random access memory (RAM) or other volatile memory. The memories 22 and 32 may be configured to temporarily store the programs 25 and 35 and data, respectively. The memories 22 and 32 may also include storage for permanently storing data including the programs 25 and 35, respectively. The storage may be, for example, read-only memory (ROM), a hard disk drive, flash memory, or other non-volatile storage device. The storage may also be a removable storage device such as a memory card. The communication interfaces 23 and 33 may be realized, for example, as a LAN or other wired communication IF.

[0018] The information processing system 11 may include a language model 14. Alternatively, the information processing device 20 may use a language model 14 not included in the information processing system 11 via the network 12. The language model 14 may be a large-scale language model for natural language processing trained using a large amount of text data. The language model 14 may be a general-purpose language model applicable to various natural language processing tasks, such as translation, information extraction, text summarization, text generation, or question and answering. The language model 14 may be built in the information processing device 20 or in the terminal device 30.

[0019] The language model 14 is configured to, when a prompt 60 (see FIG. 3) including some instruction is input, create a sentence according to the instruction and output it as a completion 61 (see FIG. 3). In the present disclosure, the information processing device 20 generates the prompt 60 and inputs it to the language model 14, and then the information processing device 20 edits the completion 61 of the language model 14 as necessary and outputs it.

[0020] In this disclosure, prompt 60 may include a translation prompt 60. Translation prompt 60 includes instructions for translating one or more source sentences written in a first language into a second language. In this disclosure, completion 61 is a translation completion 61. Translation completion 61 may include a translation that is the result of translating one or more source sentences written in the first language into the second language.

[0021] [Electronic document] The one or more electronic documents 27 may be stored in memory 22 of information processing device 20. The one or more electronic documents 27 may be electronic documents 37 that are uploaded from memory 32 of terminal device 30 to information processing device 20 and then stored in memory 22.

[0022] Each of the one or more electronic documents 27, 37 is, for example, an HTML, PDF, or plain text file containing content created for internal or external viewing, although the file format of the electronic document 37 is not limited to these. Each of the one or more electronic documents 27, 37 may include at least one of a table, a figure, or text data.

[0023] In the present disclosure, each of the electronic documents 27, 37 may include one or more original texts written in a first language. The one or more original texts may include, for example, one or more financial sentences and / or one or more financial terms. That is, the one or more electronic documents 27, 37 may be financial documents. The one or more original texts may include a date and a fiscal year. The one or more original texts may include financial terms. The one or more original texts may include numbers written with units of measure.

[0024] The units may include units that indicate the amount of something and that are different from the International System of Units and are specific to a particular region. The particular region may be, for example, a region where a first language is used and a region where a second language is used. The units may include, for example, a currency unit stated together with an amount. The currency unit may differ depending on the region where each language is used. The amount can be converted using an exchange rate.

[0025] [Bilingual dictionary] The one or more bilingual dictionaries 26 may be stored in the memory 22 of the information processing device 20. The one or more bilingual dictionaries 26 may be bilingual dictionaries 36 that are uploaded from the memory 32 of the terminal device 30 to the information processing device 20 and then stored in the memory 22.

[0026] 2, each of the bilingual dictionaries 26, 36 may include a plurality of pre-translation terms written in a first language and a plurality of translated terms written in a second language corresponding to the plurality of pre-translation terms. Each of the bilingual dictionaries 26, 36 may include a plurality of search terms corresponding to the plurality of pre-translation terms.

[0027] In this disclosure, a set of corresponding pre-translation terms, search terms, and translated terms is referred to as a term set, and a set of pre-translation terms and corresponding translated terms is referred to as a bilingual set. That is, each of the bilingual dictionaries 26 and 36 includes multiple term sets and multiple bilingual sets.

[0028] In the present disclosure, the plurality of pre-translation terms may include one or more financial terms written in a first language. In the present disclosure, the plurality of translated terms may include one or more financial terms written in a second language. Each financial term may consist of one word or may be configured as a phrase including multiple words. The plurality of search terms may be the same term as the corresponding pre-translation term. The plurality of search terms may include a regular expression of the corresponding pre-translation term.

[0029] For example, if the first language is "Japanese" and the second language is "English (US)," one term set includes the original term "earnings or loss per share," the search term "earnings or loss per share," and the translated term "earnings (losses) per share." In this example, the original term and the search term are the same term. Another term set includes the original term "as of December 31, 2023" (any date, e.g., the last day of a fiscal year), the search term "as of December 31, 2023 (0?[1-9]|1[0-2])(0?[1-9]|[1-2][0-9]|3[0-1]) of 20[0-9]{2}," and the translated term "As of December 31, 2023." Note that the regular expressions included as search terms in this term set are merely examples of regular expressions for matching any "date." The regular expressions included in the term set as search terms are written according to the pre-translation terms to be matched. In this example, the pre-translation terms and the search terms are different terms. The search terms may include one or more regular expressions. The search terms may include regular expressions that indicate a year or a date.

[0030] [Information processing device] 3, the information processing device 20 operates as a device including functional units 201 to 211 by the processor 21 executing processing based on the program 25. The functional units 201 to 211 may be program code.

[0031] The functional unit 201 is a bilingual dictionary acquisition unit 201. The bilingual dictionary acquisition unit 201 is configured to acquire one or more bilingual dictionaries 36 from the memory 32 of the terminal device 30. The one or more bilingual dictionaries 36 include one or more pre-translation terms, search terms associated with the one or more pre-translation terms, and translated terms associated with the one or more pre-translation terms. The bilingual dictionary acquisition unit 201 is configured to store the one or more bilingual dictionaries 36 acquired from the terminal device 30 in the memory 22 as a bilingual dictionary 26.

[0032] The functional unit 202 is an electronic document acquisition unit 202. The electronic document acquisition unit 202 is configured to acquire one or more electronic documents 37 from the memory 32 of the terminal device 30. The one or more electronic documents 37 include one or more original documents. The one or more electronic documents 37 are, for example, financial documents. The electronic document acquisition unit 202 is configured to store the one or more electronic documents 37 acquired from the terminal device 30 in the memory 22 as electronic documents 27.

[0033] The functional unit 203 is a splitter 203. The splitter 203 is configured to split a plurality of sentences included in the electronic document 27, 37 into a plurality of chunks 52 (see FIG. 5 ). Each of the plurality of chunks 52 includes one or more original sentences written in a first language.

[0034] The functional unit 204 is a term search unit 204. The term search unit 204 is configured to search for one or more source sentences using a plurality of search terms included in the bilingual dictionary 26. The term search unit 204 searches each of the plurality of chunks 52 using the plurality of search terms included in the bilingual dictionary 26.

[0035] The functional unit 205 is a term extraction unit 205. The term extraction unit 205 is configured to extract one or more pre-translation terms related to one or more original sentences and one or more translated terms corresponding to the one or more pre-translation terms, respectively, from the bilingual dictionary 26. The term search unit 204 is configured to extract, for example, one or more pre-translation terms corresponding to one or more search terms hit in a search executed by the term search unit 204, and translated terms associated with the pre-translation terms.

[0036] A set of pre-translation terms included in a source text to be translated and the corresponding post-translation terms is called a related bilingual set. That is, the term search unit 204 is configured to extract one or more related bilingual sets related to the source text to be translated from the multiple bilingual sets in the bilingual dictionary 26.

[0037] The functional unit 206 is a translation condition acquisition unit 206. The translation condition acquisition unit 206 is configured to acquire translation conditions. The translation condition acquisition unit 206 is configured to acquire, for example, the current fiscal year or a designated fiscal year input by the user. The translation conditions may be, for example, a designated fiscal year designated in response to an input operation by the user via the terminal device 30. The translation conditions may be the current fiscal year. The translation condition acquisition unit 206 may be configured to acquire, for example, the current exchange rate. The translation conditions may be the current exchange rate.

[0038] The functional unit 207 is a translation prompt generation unit 207. The translation prompt generation unit 207 is configured to generate a translation prompt 60 to be input to the language model 14. The translation prompt 60 may be generated corresponding to each of the multiple chunks 52. The translation prompt 60 may include one or more source sentences to be translated. The translation prompt 60 may include a related bilingual set extracted by the term extraction unit 205. One or more pre-translation terms included in the related bilingual set may be extracted according to the results of searching one chunk 52 using one or more search terms.

[0039] The translation prompt 60 may include translation instructions, such as instructions to cause the language model 14 to translate one or more pre-translation terms into one or more corresponding post-translation terms, or instructions to cause the language model 14 to generate a translation by replacing year numbers in one or more source sentences with the current year or a specified year.

[0040] The translation instructions may include instructions to have the language model 14 translate one or more terms other than one or more pre-translation terms contained in one or more source sentences using financial terms. The instructions may be a set sentence to have the user respond as an expert in financial information, such as, for example, "You are an expert in translating financial documents from Japanese to English using related bilingual translation sets. You will consistently use the most standard financial terms and financial phrases except when following the translations in the related bilingual translation sets."

[0041] The translation prompt 60 may further include a unit conversion instruction. When the source text contains one or more numerical values ​​expressed in units specific to the region in which the first language is used, the unit conversion instruction may cause the language model 14 to convert the one or more numerical values ​​into units specific to the region in which the second language is used. For example, the instruction may convert units based on the Japanese shakkanho system into units based on the imperial system.

[0042] The translation prompt 60 may further include a currency conversion instruction. The currency conversion instruction may include an instruction to cause the language model 14 to convert an amount expressed in a currency unit of a region in which a first language is used, if the amount is included in one or more source sentences, into a currency unit of a region in which a second language is used. For example, this instruction may convert Japanese yen into U.S. dollars. The currency conversion instruction may include an instruction to cause the language model 14 to convert the amount based on an exchange rate at the time of translation.

[0043] The functional unit 208 is a translation prompt input unit 208. The translation prompt input unit 208 is configured to input a translation prompt 60 generated by the translation prompt generation unit 207 to the language model 14. The functional unit 209 is an answer acquisition unit 209. The answer acquisition unit 209 is configured to acquire, as an answer, one or more completions 61 output by the language model 14 based on the translation prompt 60. The one or more completions 61 include, for example, translations of one or more original sentences included in the translation prompt 60.

[0044] The functional unit 210 is an answer generation unit 210. For example, when the answer generation unit 210 acquires a plurality of completions 61 from the language model 14, the answer generation unit 210 adds pre-translation terms and post-translation terms included in the translation prompt 60 input to acquire the completions 61. The answer generation unit 210 is configured, for example, to concatenate a plurality of completions 61 to generate an answer sentence 53. The answer generation unit 210 may, for example, add or edit the answer sentence 53. The functional unit 210 is an answer output unit 211. The answer output unit 211 is configured to output the answer sentence 53 generated by the answer generation unit 210 to the answer window 46 (see FIG. 4 ).

[0045] [Terminal Device] 1, 3, and 4, the terminal device 30 operates as a device including functional units 301 to 304 by the processor 31 executing processing based on the program 35. The functional units 301 to 304 may be program code.

[0046] The functional unit 301 is a bilingual dictionary input unit 301. The bilingual dictionary input unit 301 is configured to input one or more bilingual dictionaries 36 stored in the memory 32 of the terminal device 30 to the information processing device 20. The bilingual dictionary input unit 301 is configured to identify one or more bilingual dictionaries 36 to be input to the information processing device 20, for example, in response to an input operation by a user via the terminal device 30. This input operation to the terminal device 30 may be, for example, dragging and dropping an icon representing the bilingual dictionary 36 into an operation area 44 included in the setting window 41.

[0047] The functional unit 302 is an electronic document input unit 302. The electronic document input unit 302 is configured to input one or more electronic documents 37 stored in the memory 32 of the terminal device 30 to the information processing device 20. The electronic document input unit 302 is configured, for example, to identify one or more electronic documents 37 to be input to the information processing device 20 in response to an input operation by a user via the terminal device 30. This input operation to the terminal device 30 may be, for example, dragging and dropping an icon representing the electronic document 37 into an operation area 45 included in the setting window 41.

[0048] The functional unit 303 is a translation condition input unit 303. The translation condition input unit 303 is configured to input translation conditions to the information processing device 20. The translation condition input unit 303 is configured, for example, to identify translation conditions to be input to the information processing device 20 in response to an input operation by a user via the terminal device 30. This input operation to the terminal device 30 may be, for example, inputting text data indicating a designated year into an operation area 43 included in the setting window 41.

[0049] The functional unit 304 is a response display unit 304. The response display unit 304 is configured to display, for example, a response sentence 53 output from the response output unit 211 in a response window 46 (see FIG. 4).

[0050] [Settings window] As shown in FIG. 1, the monitor 34 may display a setting window 41. The setting window 41 may have operation areas 42 to 45. The operation area 42 is operated, for example, when downloading the bilingual dictionary 26 from the information processing device 20. The operation area 43 is operated, for example, when inputting translation conditions. The operation area 43 is operated, for example, when inputting a specified year as text data. The operation area 44 is operated, for example, when inputting a new bilingual dictionary 36 into the information processing device 20. The operation area 45 is operated, for example, when inputting an electronic document 37.

[0051] [Response window] As shown in Fig. 4, the monitor 34 may display an answer window 46. An answer sentence 53 is displayed in the answer window 46. The answer window 46 has one or more display areas 47. Each of the one or more display areas 47 is displayed corresponding to one chunk 52. Each of the one or more display areas 47 has a display area 54 that displays the original text, a display area 55 that displays pre-translation terms and translated terms, and a display area 56 that displays the translation. The display area 55 displays the related bilingual sets used to translate each chunk 52.

[0052] [Information processing method] 5 and 6, an information processing method in which the information processing system 11 translates into a second language will be described using an example of a financial document whose original text is written in a first language.

[0053] As a premise, when an icon indicating a bilingual dictionary 36 is dragged and dropped into the operation area 44, the processor 21 operating as the bilingual dictionary acquisition unit 201 acquires one or more bilingual dictionaries 36 from the memory 32 of the terminal device 30. The processor 21 operating as the bilingual dictionary acquisition unit 201 may store the bilingual dictionary 36 in the memory 22 as the bilingual dictionary 26. As a premise, when text data indicating a specified year is input into the operation area 43, the processor 21 operating as the translation condition acquisition unit 206 acquires the specified year as one of the translation conditions.

[0054] As shown in FIG. 5 , when an icon representing an electronic document 37 is dragged and dropped into the operation area 43, in step S100, the processor 21 operating as the electronic document acquisition unit 202 acquires one or more electronic documents 37 from the memory 32 of the terminal device 30. Since each of the one or more electronic documents 37 includes one or more original sentences, the one or more original sentences are acquired in step S100. The one or more original sentences may be written in, for example, a first language. The original sentences may include at least one of financial sentences and financial terms. The processor 21 operating as the electronic document acquisition unit 202 may store the electronic document 37 in the memory 22 as the electronic document 27.

[0055] In step S101, the processor 21 operating as the dividing unit 203 divides one or more original sentences included in the electronic document 27 into n chunks 52. The dividing unit 203 may extract only text data from the electronic document 27 and divide the text data. In this case, each chunk 52 does not include tables or figures, and includes only text data.

[0056] The dividing unit 203 may divide multiple original sentences into multiple chunks 52 at appropriate delimiters such as the end of a sentence, a line break, a paragraph, or a page break. The amount of data (e.g., the number of tokens) of each chunk 52 is set to be less than the amount of data that can be input to the language model 14 at one time.

[0057] In step S102, processor 21 operating as term search unit 204 searches for m-th chunk 52 using multiple search terms contained in bilingual dictionary 26. In the first iteration of step S102, m is 1, and in the second and subsequent iterations of step S102, m is the value updated in step S108.

[0058] In step S103, the processor 21 operating as the term extraction unit 205 extracts a related bilingual set from the bilingual dictionary 26. Note that if the original text to be translated does not contain a description that matches the search term, no related bilingual set is extracted.

[0059] In step S104, the processor 21 operating as the translation prompt generation unit 207 generates a translation prompt 60 to be input to the language model 14. The translation prompt 60 includes one or more original sentences included in the m-th chunk 52. The translation prompt 60 includes a related bilingual set extracted by the processor 21 operating as the term extraction unit 205. The translation prompt 60 includes a translation instruction. The translation prompt 60 may include a unit conversion instruction. The translation prompt 60 may include a currency conversion instruction.

[0060] In step S105, processor 21 operating as translation prompt input unit 208 inputs translation prompt 60 to language model 14. In step S106, processor 21 operating as answer acquisition unit 209 acquires one or more completions 61 output by language model 14 as answers. Each of one or more completions 61 includes, for example, translations of one or more original sentences included in one or more chunks 52.

[0061] 6, in step S107, the processor 21 operating as another functional unit determines whether the current chunk 52 is the n-th chunk 52. If the current chunk 52 is not the n-th chunk 52, in step S108, the processor 21 operating as another functional unit updates m to m+1. Thereafter, the processor 21 returns to step S102 (see FIG. 5).

[0062] If the current chunk 52 is the n-th chunk 52, in step S109, the processor 21 operating as the answer generation unit 210 concatenates a plurality of completions 61 to generate an answer sentence 53. In step S110, the processor 21 operating as the answer output unit 211 outputs the final answer sentence 53 to the answer window 46.

[0063] [Effects of the present disclosure] Because the language model 14 does not understand the meaning of the entire source text and produce a consistent translation, it may translate the same word in the source text into multiple different terms. This kind of meaningless variation in translation is not appropriate as a translation.

[0064] Therefore, it is conceivable to input all bilingual translation sets related to a financial document together with an instruction sentence into the language model 14 as a prompt, and output a response based on the translation instruction. However, there is generally a limit to the token length that can be input to the language model 14 as a prompt. Therefore, if all bilingual translation sets are included in the prompt, the amount of data that can be allocated to the source text will be reduced. As a result, the number of chunks will increase, which will increase the processing time and the computational load on the language model 14.

[0065] Therefore, the information processing system 11 of the present disclosure includes a related bilingual translation set and translation instructions in the translation prompt 60 in addition to the original text obtained from the electronic document 37. The translation instructions include an instruction to use the related bilingual translation set when translating the original text written in the first language in the language model 14 into the second language. Therefore, by unifying the translation terms using the related bilingual translation set, it becomes possible to obtain a more appropriate translation.

[0066] Because there is a limit to the token length that can be input as a prompt, it is not realistic to include it in all bilingual sets in the bilingual dictionary 26. Therefore, the translation prompt 60 is configured to include a related bilingual set that is always used to translate the original text. This related bilingual set includes one or more pre-translation terms that are actually included in the original text to be translated, and one or more post-translation terms that each correspond to one or more pre-translation terms. Therefore, only the related bilingual sets that the language model 14 actually uses for translation are included in the translation prompt 60. This makes it possible to reduce the amount of information input to the language model 14. This in turn makes it possible to increase the amount of other information that can be included in the translation prompt 60.

[0067] Extraction of pre-translation terms is performed by searching the original text using search terms corresponding to the pre-translation terms, rather than the pre-translation terms themselves. When pre-translation terms include variables such as dates and years, it is possible to search the original text using multiple search terms set to cover all of the variables. However, such a method requires as many search terms as there are possible combinations of values ​​for the variables, and it is possible that a search term may be omitted due to user inattention. Therefore, in the information processing system 11 disclosed herein, search terms include one or more regular expressions. This allows the original text to be searched thoroughly using search terms.

[0068] The bilingual dictionary 26 does not necessarily include all terms contained in one or more source sentences. Therefore, the translation instructions included in the translation prompt 60 include instructions to have the language model 14 translate one or more terms other than one or more pre-translation terms contained in one or more source sentences using financial terms. Therefore, even terms not included in the bilingual dictionary 26 can be translated appropriately.

[0069] One or more original sentences may include a fiscal year. It may not be appropriate to include the fiscal year included in one or more original sentences directly in the translation. Therefore, the information processing system 11 of the present disclosure includes in the translation prompt 60 a translation instruction for the language model 14 to generate the translation by replacing the fiscal year number included in one or more original sentences with the current fiscal year or a specified fiscal year. This allows the fiscal year included in the translation to be written appropriately.

[0070] One or more original sentences may contain numerical values ​​expressed in units specific to the region where the first language is used. If a translation written in a second language contains numerical values ​​expressed in units specific to the region where the first language is used, it is difficult for a user reading the translation to visualize the magnitude of the numerical values. Therefore, the information processing system 11 disclosed herein includes in the translation prompt 60 a unit conversion instruction for converting one or more numerical values ​​into units used in the region where the second language is used in the language model 14. Therefore, in the translation sentence, one or more numerical values ​​expressed in units specific to the region where the first language is used are converted into numerical values ​​expressed in units used in the region where the second language is used.

[0071] One or more original sentences may include an amount expressed in the currency unit of the region in which the first language is used. If a translation written in a second language includes an amount expressed in the currency unit of the region in which the first language is used, it is difficult for a user reading the translation to understand the amount. Therefore, the information processing system 11 disclosed herein includes a currency conversion instruction in the translation prompt 60 for converting the amount into the currency unit of the region in which the second language is used in the language model 14. Therefore, in the translation, the amount expressed in the currency unit of the region in which the first language is used is converted into the amount expressed in the currency unit of the region in which the second language is used. Furthermore, using the current exchange rate to convert the currency unit allows for a more appropriate amount to be converted.

[0072] [Effects of this disclosure] According to the present disclosure, the following effects can be achieved. (1) The translation prompt 60 includes related bilingual sets extracted from the bilingual dictionary 26. Therefore, the amount of information input to the language model 14 can be reduced compared to when all bilingual sets in the bilingual dictionary 26 are included in the translation prompt 60. This can shorten the processing time and reduce the amount of calculations required by the language model 14.

[0073] (2) When searching one or more source sentences, search terms corresponding to the pre-translation terms are used, rather than the pre-translation terms themselves. The translation prompt 60 includes one or more pre-translation terms corresponding to one or more search terms found in the search. This allows for optimization of the search, making it easier to find terms that would be difficult to find using the pre-translation terms.

[0074] (3) The search term contains one or more regular expressions, so pre-translation terms that match the regular expressions can be extracted comprehensively. (4) The translation instructions include instructions to the language model 14 to generate a translation by replacing the year numbers included in one or more source sentences with the current year or a specified year. Therefore, a translation can be generated in which the year numbers included in one or more source sentences are appropriately replaced.

[0075] (5) The translation instructions include instructions to have the language model 14 translate one or more terms other than one or more pre-translation terms contained in one or more source sentences using financial terms. Therefore, even terms not included in the bilingual dictionary 26 can be translated into general financial terms.

[0076] (6) When the original text contains one or more numerical values ​​expressed in units specific to the region where the first language is used, the translation prompt 60 includes a unit conversion instruction that causes the language model 14 to convert the one or more numerical values ​​into units used in the region where the second language is used. Therefore, the translated text shows the numerical values ​​converted into units used in the region where the second language is used. This allows the information processing system 11 to output a more appropriate translation.

[0077] (7) When one or more original sentences contain an amount expressed in the currency unit of the region in which the first language is used, the translation prompt 60 includes a currency conversion instruction that causes the language model 14 to convert the amount into the currency unit of the region in which the second language is used. Therefore, the translated sentence shows the amount converted into the currency unit of the region in which the second language is used. This allows the information processing system 11 to output an appropriate translation according to the region.

[0078] (8) The currency conversion instruction includes an instruction to have the language model 14 convert the amount based on the current exchange rate, so that the amount according to the current exchange rate can be reflected in the translation.

[0079] [Example of change] This embodiment can be modified as follows: This embodiment and the following modifications can be combined and implemented within the scope of technical compatibility.

[0080] [Change Example 1] 7, the information processing system 11 may be configured to acquire one or more new search terms from the language model 14. The functional units 202 to 211 and 301 to 304 have the same configuration as in the present embodiment, and therefore detailed description thereof will be omitted.

[0081] The information processing device 20 includes functional units 221 to 224 in addition to the functional units 202 to 211. The information processing device 20 may include the functional unit 201. The functional unit 221 is a term example acquisition unit 221. The term example acquisition unit 221 is configured to acquire, as term samples, at least some of the pre-translation terms included in the bilingual dictionary 26 or the bilingual dictionary 36 and one or more search terms corresponding to at least some of the pre-translation terms.

[0082] The term samples acquired by the term example acquisition unit 221 are used as correct answer data indicating the relationship of search terms to pre-translation terms. The term example acquisition unit 221 may be configured to acquire, for example, one or more pre-translation terms specified in response to an input operation by the user via the terminal device 30, and one or more search terms corresponding to the one or more pre-translation terms.

[0083] The functional unit 222 is a new term acquisition unit 222. The new term acquisition unit 222 is configured to acquire, for example, one or more new terms. The new terms are terms written in the first language that are not included in the bilingual dictionary 26. The new term acquisition unit 222 may be configured to acquire, for example, one or more new terms specified in response to an input operation by the user via the terminal device 30.

[0084] The functional unit 223 is a completion prompt generation unit 223. The completion prompt generation unit 223 is configured to generate, for example, a completion prompt 70. The completion prompt 70 includes, for example, a term sample extracted from the bilingual dictionary 26, one or more new terms, and a completion instruction. The completion instruction includes, for example, an instruction to cause the language model 14 to output one or more new search terms corresponding to the one or more new terms, respectively, in accordance with the relationship of the search terms to the pre-translation terms in the term sample.

[0085] The functional unit 224 is a completion prompt input unit 224. The completion prompt input unit 224 is configured to input the completion prompt 70 generated by the completion prompt generation unit 223 to the language model 14, for example.

[0086] The functional unit 225 is a search term acquisition unit 225. The search term acquisition unit 225 is configured to acquire one or more new search terms from the language model 14. The search term acquisition unit 225 is configured to acquire, for example, one or more completions 71 output by the language model 14 based on the completion prompt 70 as answers. The one or more completions 71 include one or more search terms corresponding to the one or more new terms.

[0087] The search term acquisition unit 225 may associate one or more new terms with one or more search terms acquired from the language model 14 and add them to the bilingual dictionary 26. The search term acquisition unit 225 may add one or more translated terms to the bilingual dictionary 26, associating them with one or more new terms. The one or more translated terms associated with the one or more new terms may be terms specified in response to an input operation by the user via the terminal device 30. The one or more translated terms associated with the one or more new terms may be terms generated by the language model 14 together with one or more search terms. The term search unit 204 may search the original text using the search terms acquired by the search term acquisition unit 225.

[0088] The terminal device 30 includes a functional unit 321 in addition to the functional units 301 to 304. The functional unit 321 is a new term input unit 321. The new term input unit 321 is configured to input one or more new terms to the information processing device 20. The new term input unit 321 is configured, for example, to specify one or more new terms to be input to the information processing device 20 in response to an input operation by a user via the terminal device 30. The input operation may be, for example, a user inputting text data indicating one or more new terms into a specific area included in the setting window 41. Alternatively, the input operation may be a user dragging and dropping an icon indicating an electronic document including one or more new terms into a specific area included in the setting window 41.

[0089] An information processing method for acquiring one or more search terms corresponding to one or more new terms will be described with reference to Fig. 8. In step S201, processor 21 operating as term example acquisition unit 221 acquires term samples (one or more pre-translation terms and one or more search terms corresponding to the one or more pre-translation terms) from bilingual dictionary 26 or bilingual dictionary 36.

[0090] When one or more new terms are input in response to a user's input operation via the terminal device 30, the processor 21 operating as the new term acquisition unit 222 acquires the one or more new terms in step S202. In step S203, the processor 21 operating as the completion prompt generation unit 223 generates a completion prompt 70 including the one or more new terms acquired by the new term acquisition unit 222 and the term sample acquired by the term example acquisition unit 221.

[0091] In step S204, the processor 21 operating as the completion prompt input unit 224 inputs the completion prompt 70 generated in step S203 into the language model 14. In step S205, the processor 21 operating as the search term acquisition unit 225 acquires, as answers, one or more completions 71 output by the language model 14 based on the completion prompt 70.

[0092] One or more search terms may include regular expressions. Regular expressions are useful in that they can comprehensively match pre-translation terms that correspond to a search term. However, it is difficult for users who are not familiar with regular expressions to express search terms in regular expressions. Therefore, the information processing system 11 of this modified example generates a completion prompt 70 and acquires one or more search terms output by the language model 14 based on the completion prompt 70. Therefore, even users who are not familiar with regular expressions can easily acquire search terms written in regular expressions.

[0093] [Change Example 2] The unit conversion instructions may include instructions for causing language model 14 to convert, when one or more original texts contain one or more numerical values ​​expressed in units specific to the region in which the first language is used, each of the one or more numerical values ​​into a corresponding unit included in the International System of Units. According to this modified example, an appropriate translation can be obtained regardless of the type of second language and the region in which the second language is used.

[0094] [Change Example 3] The functional units 201-211 and 221-225 may be implemented by one information processing device 20, or by three or more information processing devices 20. The functional units 201-211 and 221-225 may be implemented by one terminal device 30, or by three or more terminal devices 30. Furthermore, steps S100-S110 and S201-S205 may be implemented by different functional units (program codes). The functional units 301-304 and 321 may be implemented by one terminal device 30, or by three or more terminal devices 30. The functional units 301-304 and 321 may be implemented by one information processing device 20, or by three or more information processing devices 20. In this way, the information processing system 11 is not limited to being configured to include one or more information processing devices 20 and one or more terminal devices 30. The information processing system 11 may be configured without including a terminal device 30. The information processing system 11 may be configured without including the information processing device 20.

[0095] [Change Example 4] The electronic documents 27 and 37 may be documents other than financial documents, such as papers, reports, educational textbooks, academic conference materials, business documents, patent documents, or instruction manuals in a particular technical field.

[0096] [Change Example 5] The language model for inputting the completion prompt 70 to obtain the completion 71 and the language model for inputting the translation prompt 60 to obtain the translation completion 61 may be the same language model or different language models.

[0097] [Change Example 6] The information processing system 11 may be realized by one or more information processing devices 20 that realize various functions of the present disclosure using an external language model 14. The information processing system 11 may be realized as a translation system. The information processing method of the present disclosure may be realized as a translation method. The program of the present disclosure may be realized as a translation program.

[0098] The following are some aspects that can be understood based on the above-described embodiment and modifications. [1] An information processing system configured to translate one or more source texts written in a first language into a second language using a language model, the information processing system comprising: one or more memories configured to store a program; one or more processors configured to execute processing based on the program; the one or more processors: extracting one or more pre-translation terms related to the one or more original sentences and one or more translated terms corresponding to the one or more pre-translation terms from a bilingual dictionary including a plurality of pre-translation terms written in the first language and a plurality of translated terms written in the second language corresponding to the plurality of pre-translation terms; inputting a translation prompt into the language model, the translation prompt including the one or more source sentences, the one or more pre-translation terms, the one or more translated terms, and a translation instruction, the translation instruction including an instruction for causing the language model to translate the one or more pre-translation terms into the one or more translated terms; obtaining, from the language model, a translation of the one or more original sentences into the second language in accordance with the translation instruction; An information processing system configured to execute the

[0099] [2] The bilingual dictionary includes a plurality of search terms corresponding to the plurality of pre-translation terms, the one or more processors: searching the one or more source texts utilizing the plurality of search terms; acquiring the one or more pre-translation terms corresponding to the one or more search terms found in the search, [1] The information processing system according to [1].

[0100] [3] the search term includes one or more regular expressions; [2] The information processing system according to [2]. [4] the one or more regular expressions include a regular expression that represents a year; the one or more processors are further configured to obtain a current year or a designated year input by a user; the translation instructions include instructions to the language model to generate the translation by replacing a year number included in the one or more source sentences with the current year or the specified year; [3] The information processing system according to [3].

[0101] [5] The language model is a general-purpose language model capable of multiple language processes including translation and input completion; the one or more processors: acquiring one or more new terms, the one or more new terms being terms not included in the bilingual dictionary written in the first language; inputting a completion prompt into the language model, the completion prompt including at least some pre-translation terms included in the bilingual dictionary, one or more search terms corresponding to the at least some pre-translation terms, the one or more new terms, and a completion instruction, the completion instruction including an instruction to cause the language model to output one or more new search terms corresponding to the one or more new terms, in accordance with the relationship of the one or more search terms to the at least some pre-translation terms; obtaining the one or more new search terms from the language model; and configured to perform The information processing system according to any one of [2] to [4].

[0102] [6] the one or more original documents include one or more financial documents; the plurality of pre-translation terms includes one or more financial terms; the translation instructions include instructions to cause the language model to translate one or more terms other than the one or more pre-translation terms included in the one or more source sentences using financial terms; The information processing system according to any one of [1] to [4].

[0103] [7] the translation prompt further includes unit conversion instructions; the unit conversion instructions include, when the original text contains one or more numerical values ​​expressed in units specific to a region in which the first language is used, instructions for causing the language model to convert the one or more numerical values ​​into units used in a region in which the second language is used; The information processing system according to any one of [1] to [4].

[0104] [8] the translation prompt further includes unit conversion instructions; the unit conversion instructions include instructions for causing the language model to convert, when the one or more source sentences contain one or more numerical values ​​expressed in units specific to a region in which the first language is used, each of the one or more numerical values ​​into a corresponding unit included in the International System of Units; 5. The information processing system according to claim 1.

[0105] [9] the translation prompt further includes currency conversion instructions; the currency conversion instructions include instructions for causing the language model to convert, when the one or more source sentences contain an amount indicated in a currency unit of a region in which the first language is used, the amount into a currency unit of a region in which the second language is used; The information processing system according to any one of [1] to [4].

[0106]

[10] the one or more processors are further configured to obtain a current exchange rate; the currency conversion instructions include instructions for causing the language model to convert the amount based on the exchange rate. [9] The information processing system according to [9].

[0107]

[11] An information processing method configured to translate an original text written in a first language into a second language using a language model, comprising: One or more processors extracting one or more pre-translation terms related to the one or more original sentences and one or more translated terms corresponding to the one or more pre-translation terms from a bilingual dictionary including a plurality of pre-translation terms written in the first language and a plurality of translated terms written in the second language corresponding to the plurality of pre-translation terms; inputting a translation prompt into the language model, the translation prompt including the one or more source sentences, the one or more pre-translation terms, the one or more translated terms, and a translation instruction, the translation instruction including an instruction for causing the language model to translate the one or more pre-translation terms into the one or more translated terms; obtaining, from the language model, a translation of the one or more original sentences into the second language in accordance with the translation instruction; An information processing method, including causing a computer to execute the above.

[0108]

[12] A program for translating one or more source texts written in a first language into a second language using a language model, comprising: a program code for extracting, from a bilingual dictionary including a plurality of pre-translation terms written in the first language and a plurality of post-translation terms written in the second language corresponding to the plurality of pre-translation terms, one or more pre-translation terms related to the one or more original sentences and one or more post-translation terms corresponding to the one or more pre-translation terms; program code for inputting a translation prompt to the language model, the translation prompt including the one or more source sentences, the one or more pre-translation terms, the one or more translated terms, and a translation instruction, the translation instruction including an instruction for causing the language model to translate the one or more pre-translation terms into the one or more translated terms; program code for obtaining, from the language model, a translation of the one or more source sentences into the second language in accordance with the translation instruction; Including, the program. [Explanation of symbols]

[0109] 11...information processing system, 12...network, 13...vector database, 14...language model, 15...language model, 20...information processing device, 21...processor, 22...memory, 23...communication interface, 25...program, 26...bilingual dictionary, 27...electronic document, 30...terminal device, 31...processor, 32...memory, 33...communication interface, 34...monitor, 35...program, 36...bilingual dictionary, 37...electronic document, 41...setting window, 42...operation area, 43...operation area, 44...operation area, 45...operation area, 46...answer window, 47...display area, 52...chunk, 53...answer sentence, 54...display area, 55...display area, 56...display area, 60...translation prompt, 60...prompt prompt, 61...translation completion, 61...completion, 70...completion prompt, 71...completion completion, 201...bilingual dictionary acquisition unit, 202...electronic document acquisition unit, 203...division unit, 204...term search unit, 205...term extraction unit, 206...translation condition acquisition unit, 207...translation prompt generation unit, 208...translation prompt input unit, 209...answer acquisition unit, 210...answer generation unit, 211...answer output unit, 221...term example acquisition unit, 222...new term acquisition unit, 223...completion prompt generation unit, 224...completion prompt input unit, 225...search term acquisition unit, 301...bilingual dictionary input unit, 302...electronic document input unit, 303...translation condition input unit, 304...answer display unit, 321...new term input unit.

Claims

1. 1. An information processing system configured to translate one or more source sentences written in a first language into a second language using a language model, the information processing system comprising: one or more memories configured to store a program; one or more processors configured to execute processing based on the program; the one or more processors: searching the one or more original texts using a plurality of search terms contained in a bilingual dictionary including a plurality of pre-translation terms written in the first language, a plurality of search terms respectively corresponding to the plurality of pre-translation terms, and a plurality of translated terms written in the second language respectively corresponding to the plurality of pre-translation terms, and identifying one or more pre-translation terms among the plurality of pre-translation terms contained in the bilingual dictionary that correspond to one or more of the search terms found in the search, wherein each of the plurality of search terms is the same as the one or more corresponding pre-translation terms or a regular expression of the one or more corresponding pre-translation terms, and the plurality of search terms includes one or more of the regular expressions; extracting, from the bilingual dictionary, the one or more pre-translation terms identified in identifying the one or more pre-translation terms and included in the one or more source sentences, and one or more translated terms corresponding to the one or more pre-translation terms; inputting a translation prompt into the language model, the translation prompt including the one or more original sentences, the one or more pre-translation terms, the one or more translated terms, and a translation instruction for causing the language model to translate the one or more original sentences, the translation instruction including an instruction for translating the one or more pre-translation terms included in the one or more original sentences into the one or more translated terms included in the translation prompt when translating the one or more original sentences; obtaining, from the language model, a translation of the one or more original sentences into the second language in accordance with the translation instruction; Run the one or more processors perform dividing the one or more source sentences into a plurality of chunks; the one or more processors perform, for each chunk, identifying the one or more pre-translation terms, extracting them from the bilingual dictionary, inputting the translation prompt into the language model, and obtaining the translation; The one or more processors are configured to execute the steps of: concatenating the translations of each of the chunks to generate an answer sentence.

2. the one or more processors display one or more sets of the one or more pre-translation terms and the one or more post-translation terms used to obtain the translation of each of the chunks; The information processing system according to claim 1 .

3. 1. An information processing method configured to translate one or more source texts written in a first language into a second language using a language model, comprising: one or more processors, searching the one or more original texts using a plurality of search terms contained in a bilingual dictionary including a plurality of pre-translation terms written in the first language, a plurality of search terms respectively corresponding to the plurality of pre-translation terms, and a plurality of translated terms written in the second language respectively corresponding to the plurality of pre-translation terms, and identifying one or more pre-translation terms among the plurality of pre-translation terms contained in the bilingual dictionary that correspond to one or more of the search terms found in the search, wherein each of the plurality of search terms is the same as the one or more corresponding pre-translation terms or a regular expression of the one or more corresponding pre-translation terms, and the plurality of search terms includes one or more of the regular expressions; extracting, from the bilingual dictionary, the one or more pre-translation terms identified in identifying the one or more pre-translation terms and included in the one or more source sentences, and one or more translated terms corresponding to the one or more pre-translation terms; inputting a translation prompt into the language model, the translation prompt including the one or more original sentences, the one or more pre-translation terms, the one or more translated terms, and a translation instruction for causing the language model to translate the one or more original sentences, the translation instruction including an instruction for translating the one or more pre-translation terms included in the one or more original sentences into the one or more translated terms included in the translation prompt when translating the one or more original sentences; obtaining, from the language model, a translation of the one or more original sentences into the second language in accordance with the translation instruction; Execute causing the one or more processors to divide the one or more source sentences into a plurality of chunks; causing the one or more processors to, for each chunk, identify the one or more pre-translation terms, extract them from the bilingual dictionary, input the translation prompt to the language model, and obtain the translation; causing the one or more processors to generate an answer sentence by concatenating the translations of each of the chunks.

4. 1. A program for translating one or more source sentences written in a first language into a second language using a language model, comprising: a program code for searching the one or more original texts using a plurality of search terms contained in a bilingual dictionary including a plurality of pre-translation terms written in the first language, a plurality of search terms respectively corresponding to the plurality of pre-translation terms, and a plurality of translated terms written in the second language respectively corresponding to the plurality of pre-translation terms, and for identifying one or more pre-translation terms among the plurality of pre-translation terms contained in the bilingual dictionary that correspond to one or more of the search terms found in the search, wherein each of the plurality of search terms is the same as the one or more corresponding pre-translation terms, or a regular expression of the one or more corresponding pre-translation terms, and the plurality of search terms includes one or more of the regular expressions; a program code for extracting, from the bilingual dictionary, the one or more pre-translation terms identified in identifying the one or more pre-translation terms and included in the one or more source sentences, and one or more translated terms corresponding to the one or more pre-translation terms; program code for inputting a translation prompt to the language model, the translation prompt including the one or more original sentences, the one or more pre-translation terms, the one or more translated terms, and a translation instruction for causing the language model to translate the one or more original sentences, the translation instruction including an instruction for translating the one or more pre-translation terms included in the one or more original sentences into the one or more translated terms included in the translation prompt when translating the one or more original sentences; program code for obtaining, from the language model, a translation obtained by translating the one or more source sentences into the second language in accordance with the translation instruction; Including, program code for dividing the one or more source sentences into a plurality of chunks; program code for performing, for each of the chunks, identifying the one or more pre-translation terms, extracting them from the bilingual dictionary, inputting the translation prompt into the language model, and obtaining the translation; and program code for generating an answer sentence by concatenating the translations of each of the chunks.

Images