Information processing system and information processing method

Abstract

The present invention provides an information processing system that can prevent the leakage of confidential information. The information processing system includes an accepting unit and a processing unit. The accepting unit has a function of accepting first text data and at least one first string. The processing unit is configured to execute: processing to convert the at least one first string to a second string associated with the first string in a 1:1 manner, thereby converting the first text data to second text data containing at least one second string; processing to transmit the second text data to a language model over a network, thereby acquiring third text data containing at least one second string; and processing to convert the at least one second string to the first string associated with the second string in a 1:1 manner, thereby converting the third text data to fourth text data. The second string contains a symbol.

Description

Information processing system and information processing method

[0001] TECHNICAL FIELD One aspect of the present invention relates to an information processing system and an information processing method, and more particularly, to an information processing system and an information processing method that can utilize processing using a language model.

[0002] One embodiment of the present invention is not limited to the above technical field, and examples of the technical field of one embodiment of the present invention include semiconductor devices, display devices, light-emitting devices, power storage devices, memory devices, electronic devices, lighting devices, input devices (e.g., touch sensors), input / output devices (e.g., touch panels), driving methods thereof, and manufacturing methods thereof.

[0003] In recent years, the development of language models using neural networks has been actively carried out, and large-scale language models (LLMs) in particular have attracted attention. A large-scale language model is a natural language processing model trained using a large amount of data. A large-scale language model can realize, for example, a dialogue model that responds to user instructions. Non-Patent Document 1 discloses GPT-4 (Generative Pre-trained Transformer 4) as a large-scale language model and ChatGPT as a dialogue model.

[0004] The use of large-scale language models has significantly increased the capabilities of natural language processing models. However, as language models become larger, it is difficult to incorporate and operate language models in-house due to the equipment and cost involved. Therefore, one way to use language models is to use external services that provide language models.

[0005] When using external services, there is a concern that confidential information included in data sent to a language model may be leaked due to unexpected events such as hacking. Patent Document 1 discloses an information processing device and an information processing method that generate a document image that can be made public without using the confidential information, based on a document image that includes confidential information.

[0006] In addition, sentences input to a language model may be used for training. For example, if confidential information is input to a language model, this information may be collected and used for training the language model, which may lead to information leakage.

[0007] JP 2023-87840 A

[0008] Summary of ChatGPT / GPT-4 Research and Perspective Towards the Future of Large Language Models, Yiheng Liu et al. (Submitted on 4 Apr 2023, [online], Internet <URL: https: / / arxiv.org / abs / 2304.01852>

[0009] One method for preventing information leakage is a simple method such as deleting confidential information contained in the data. However, deleting only the confidential information from the data destroys the context, and if such data is sent to a language model, the performance of the language model may not be fully utilized. Another method is to replace the confidential information contained in the data with a different word. Even if such data is sent to a language model, the replaced word, which depends on the content learned by the language model, may affect the behavior of the language model and reduce the quality of the language model output. Furthermore, there is a possibility that the output of the language model may contain the same word as the replaced word, which is expected to make decoding difficult. Note that the same applies to personal information, and therefore confidential information described in this specification may also be referred to as personal information.

[0010] In view of the above, it is an object of one embodiment of the present invention to provide an information processing system or an information processing method that can prevent leakage of confidential information. Another object of one embodiment of the present invention is to provide an information processing system or an information processing method that generates text data that can prevent leakage of confidential information. Another object of one embodiment of the present invention is to provide an information processing system or an information processing method that can be used safely by a user.

[0011] Another object of one embodiment of the present invention is to provide a highly convenient information processing system or information processing method.Another object of one embodiment of the present invention is to provide a highly secure information processing system or information processing method.Another object of one embodiment of the present invention is to provide a novel information processing system or information processing method.

[0012] Note that the description of these problems does not preclude the existence of other problems. One embodiment of the present invention does not necessarily have to solve all of these problems. Problems other than these can be extracted from the description in the specification, drawings, and claims.

[0013] One aspect of the present invention is an information processing system including a receiving unit and a processing unit. The receiving unit has a function of receiving first text data and at least one first character string included in the first text data. The processing unit is configured to perform the following processes: converting the at least one first character string in the first text data into second text data including at least one second character string by converting the at least one first character string in the first text data into a second character string that has a one-to-one correspondence with the first character string; acquiring third text data including the at least one second character string by transmitting the second text data to a language model via a network; and converting the at least one second character string in the third text data into a first character string that has a one-to-one correspondence with the second character string by converting the third text data into fourth text data. The second character string includes a first symbol.

[0014] In the above information processing system, the first character string preferably includes confidential information.

[0015] In the information processing system, the second character string is preferably a character string made up of a first symbol and at least one selected from a symbol different from the first symbol, an alphabet, and a number.

[0016] In the information processing system, it is preferable that the number of characters in the second string matches the number of characters in the corresponding first string.

[0017] In the above information processing system, it is preferable that the second string has a configuration in which a sixth string is sandwiched between a pair of first symbols, and that the sixth string includes at least one selected from symbols different from the first symbols, alphabets, and numbers.

[0018] In the information processing system, it is preferable that the number of characters in the sixth character string matches the number of characters in the corresponding first character string.

[0019] In the above information processing system, the second text data is preferably a prompt.

[0020] In the information processing system, the processing unit is preferably configured to execute a process of creating a dictionary that defines a correspondence relationship between the first character string and the second character string, and the fourth text data is preferably converted from the third text data by referring to the dictionary.

[0021] The information processing system preferably includes a storage unit. The storage unit preferably includes a plurality of third character strings that are requested to be converted into second character strings. The processing unit is preferably configured to execute a process of registering at least one first character string in the storage unit as a third character string when at least one first character string is not included in the storage unit.

[0022] In the information processing system, the receiving unit preferably has a function of receiving a fourth character string that is one of at least one second character string. The processing unit is preferably configured to execute a process of updating the second text data by referring to the storage unit and restoring the fourth character string included in the second text data to a third character string that has a one-to-one correspondence with the fourth character string, and a process of recording in the storage unit a history of the fourth character string being restored to the third character string that has a one-to-one correspondence with the fourth character string.

[0023] In the information processing system, the accepting unit preferably has a function of accepting fifth text data. Furthermore, when the fifth text data includes a third character string that has a high historical frequency among the plurality of third character strings, the processing unit preferably does not convert the third character string into a fourth character string that has a one-to-one correspondence with the third character string.

[0024] In the information processing system, it is preferable that each of the plurality of third character strings is assigned a numerical value that is set so that the numerical value decreases as the character string becomes more likely to contain confidential information. Furthermore, in the process of converting the first text data into the second text data, it is preferable that a third character string having a smaller numerical value among at least one first character string is converted into a second character string in one-to-one correspondence.

[0025] In the information processing system, it is preferable that each of the plurality of third character strings is assigned a category extracted using named entity extraction. Furthermore, in the process of converting the first text data into the second text data, it is preferable that, among at least one first character string, a third character string assigned a specific category is converted into a second character string in one-to-one correspondence.

[0026] In the information processing system, the second text data is preferably converted from the first text data so that the proportion of the second character string in the second text data is 35% or more and 50% or less.

[0027] One aspect of the present invention is an information processing system including a receiving unit and a processing unit. The receiving unit has a function of receiving first text data and at least one first character string included in the first text data. The processing unit is configured to execute the following processes: converting the at least one first character string in the first text data into second text data including the at least one second character string by converting the at least one first character string in the first text data into a second character string that has a one-to-one correspondence with the first character string; acquiring third text data including the at least one second character string by transmitting the second text data to a language model via a network; and converting the at least one second character string in the third text data into a first character string that has a one-to-one correspondence with the second character string by converting the third text data into fourth text data. In the process of converting first text data into second text data, for each of at least one first character string, a plurality of third character strings that are not included in the first text data are selected, a verification process is performed for each of the plurality of third character strings, and a process is performed to select from the plurality of third character strings a second character string that is in one-to-one correspondence with the first character string by referring to the results of the verification process, thereby converting the first character string into a second character string.

[0028] In the above information processing system, the first character string preferably includes confidential information.

[0029] In the above information processing system, the verification process prepares a plurality of sentences consisting of a first sentence including a first character string and a plurality of second sentences in which the first character string included in the first sentence is converted into each of a plurality of third character strings, and performs natural language processing to judge the plurality of sentences, thereby calculating a score for the natural language processing.In the process performed with reference to the results of the verification process, it is preferable that the third character string included in the second sentence among the plurality of second sentences that has the score closest to that of the first sentence is set as the second character string that corresponds one-to-one to the first character string.

[0030] In the above information processing system, the second character string is preferably a hypernym, a synonym, or a hyponym of the corresponding first character string.

[0031] In the above information processing system, the second character string preferably has a configuration in which a hypernym, a synonym, or a hyponym of the corresponding first character string is sandwiched between a pair of symbols.

[0032] In the above information processing system, the second text data is preferably a prompt.

[0033] In the information processing system, the processing unit is preferably configured to execute a process of creating a dictionary that defines a correspondence relationship between the first character string and the second character string, and the fourth text data is preferably converted from the third text data by referring to the dictionary.

[0034] One aspect of the present invention is an information processing method that receives first text data and at least one first character string included in the first text data, converts the at least one first character string in the first text data into a second character string that has a one-to-one correspondence with the first character string, thereby converting the first text data into second text data that includes at least one second character string, transmits the second text data to a language model via a network to obtain third text data that includes the at least one second character string, and converts the at least one second character string in the third text data into a first character string that has a one-to-one correspondence with the second character string, thereby converting the third text data into fourth text data, where the second character string includes a first symbol.

[0035] In the above information processing method, it is preferable that the first character string includes confidential information.

[0036] In the above information processing method, the second character string is preferably a character string made up of a first symbol and at least one selected from a symbol different from the first symbol, an alphabet, and a number.

[0037] In the above information processing method, it is preferable that the number of characters in the second character string matches the number of characters in the corresponding first character string.

[0038] In the above information processing method, it is preferable that the second string has a configuration in which a sixth string is sandwiched between a pair of first symbols, and that the sixth string includes at least one selected from symbols different from the first symbols, alphabets, and numbers.

[0039] In the above information processing method, it is preferable that the number of characters in the sixth character string matches the number of characters in the corresponding first character string.

[0040] In the above information processing method, the second text data is preferably a prompt.

[0041] In the above information processing method, it is preferable that a dictionary defining the correspondence between the first character string and the second character string is created, and the fourth text data is converted from the third text data by referring to the dictionary.

[0042] One aspect of the present invention is an information processing method that receives first text data and at least one first character string included in the first text data, converts the at least one first character string in the first text data into a second character string that has a one-to-one correspondence with the first character string, thereby converting the first text data into second text data that includes at least one second character string, transmits the second text data to a language model via a network to obtain third text data that includes the at least one second character string, and converts the at least one second character string in the third text data into a first character string that has a one-to-one correspondence with the second character string, thereby converting the third text data into fourth text data. The second character string is a character string that is not included in the first text data. In converting the first text data into the second text data, for each of at least one first character string, a plurality of third character strings not included in the first text data are selected, a verification process is performed on each of the plurality of third character strings, and a second character string that is in one-to-one correspondence with the first character string is selected from the plurality of third character strings by referring to the results of the verification process, thereby converting the first character string into the second character string.

[0043] In the above information processing method, it is preferable that the first character string includes confidential information.

[0044] In the above information processing method, the verification process prepares a plurality of sentences consisting of a first sentence including a first character string and a plurality of second sentences in which the first character string included in the first sentence is converted into a plurality of third character strings, and performs natural language processing to judge the plurality of sentences, thereby calculating a score for the natural language processing. In the process performed with reference to the results of the verification process, it is preferable that the third character string included in the second sentence among the plurality of second sentences that has a score closest to that of the first sentence is set as the second character string that corresponds one-to-one to the first character string.

[0045] In the above information processing method, the second character string is preferably a hypernym, a synonym, or a hyponym of the corresponding first character string.

[0046] In the above information processing method, the second character string preferably has a configuration in which a hypernym, a synonym, or a hyponym of the corresponding first character string is sandwiched between a pair of symbols.

[0047] In the above information processing method, the second text data is preferably a prompt.

[0048] In the above information processing method, it is preferable that a dictionary defining the correspondence between the first character string and the second character string is created, and the fourth text data is converted from the third text data by referring to the dictionary.

[0049] One aspect of the present invention is a program having a function of causing a processor to execute any of the above information processing methods.

[0050] According to one aspect of the present invention, it is possible to provide an information processing system or an information processing method that can prevent leakage of confidential information. Also, according to one aspect of the present invention, it is possible to provide an information processing system or an information processing method that generates text data that can prevent leakage of confidential information. Also, according to one aspect of the present invention, it is possible to provide an information processing system or an information processing method that can be used safely by users.

[0051] According to one embodiment of the present invention, a highly secure information processing system or an information processing method can be provided. According to another embodiment of the present invention, a highly convenient information processing system or an information processing method can be provided. According to another embodiment of the present invention, a novel information processing system or an information processing method can be provided.

[0052] Note that the description of these effects does not preclude the existence of other effects. One embodiment of the present invention does not necessarily have all of these effects. Effects other than these can be extracted from the description in the specification, drawings, and claims.

[0053] FIG. 1 is a diagram illustrating an example of an information processing system. FIG. 2 is a diagram illustrating an example of an information processing system. FIG. 3 is a block diagram illustrating an example of an information processing system. FIG. 4 is a diagram illustrating an example of an information processing method. FIGS. 5A to 5D are diagrams illustrating an example of an information processing method. FIG. 6 is a diagram illustrating an example of an information processing method. FIGS. 7A to 7D are diagrams illustrating an example of an information processing method. FIG. 8 is a diagram illustrating an example of an information processing method. FIG. 9 is a diagram illustrating an example of an information processing method. FIGS. 10A to 10C are diagrams illustrating an example of an information processing method. FIG. 11 is a diagram illustrating an example of an information processing system. FIG. 12 is a block diagram illustrating an example of an information processing system. FIG. 13 is a diagram illustrating an example of an information processing system. FIG. 14 is a diagram illustrating an example of an information processing system.

[0054] The embodiments will be described in detail with reference to the drawings. However, the present invention is not limited to the following description, and it will be readily understood by those skilled in the art that various changes can be made in form and detail without departing from the spirit and scope of the present invention. Therefore, the present invention should not be interpreted as being limited to the description of the embodiments shown below.

[0055] In the configuration of the invention described below, the same parts or parts having similar functions are denoted by the same reference numerals in different drawings, and repeated explanations thereof will be omitted. Furthermore, when referring to similar functions, the same hatching pattern may be used and no particular reference numeral may be assigned.

[0056] Furthermore, for ease of understanding, the position, size, range, etc. of each component shown in the drawings may not represent the actual position, size, range, etc. Therefore, the disclosed invention is not necessarily limited to the position, size, range, etc. disclosed in the drawings.

[0057] In this specification, the ordinal numbers "first" and "second" are used for convenience and do not limit the number of components or the order of the components (for example, the order of processes or the order of stacking). Furthermore, the ordinal numbers assigned to components in one part of this specification may not match the ordinal numbers assigned to the same components in other parts of this specification or in the claims.

[0058] Embodiment In this embodiment, an information processing system according to one embodiment of the present invention will be described with reference to FIGS.

[0059] An information processing system according to one aspect of the present invention includes a receiving unit and a processing unit. The receiving unit has a function of receiving first text data and at least one first character string included in the first text data. If the first character string includes confidential information, the first text data includes confidential information. When using an external service, there is a concern that the confidential information may be leaked when the first text data is sent to a language model.

[0060] Therefore, the processing unit has a function of converting the first text data into second text data by converting at least one first character string in the first text data into a second character string that corresponds one-to-one to the first character string.

[0061] In order to hide or delete confidential information contained in the first string, the second string is configured so as not to include the meaning of the first string. For example, the second string preferably includes a symbol, and more preferably has a configuration in which any character string is sandwiched between a pair of symbols. Alternatively, in order to hide or delete confidential information contained in the first string, a character string different from the first string and having a small impact on the output content of the language model is used as the second string. For example, the second string is preferably a hypernym, antonym, or hyponym of the first string. Alternatively, the second string may include a word or an abbreviation indicating the category of the first string, such as “[CHEM_1].” This is expected to preserve as much of the context of the first string as possible while hiding confidential information.

[0062] By converting the first character string into the second character string, confidential information included in the first text data can be hidden or deleted. In this specification, the process of converting the first character string into the second character string may be referred to as a masking process. Furthermore, the second character string may be referred to as a mask character string.

[0063] The information processing system is capable of transmitting and receiving data via a network to and from an information processing device capable of performing processing using a language model. The processing unit has a function of acquiring third text data by transmitting second text data to the language model via the network. Since confidential information is hidden or deleted from the second text data, leakage of confidential information can be prevented. Furthermore, an answer based on the language model can be acquired as the second character string without being converted. Note that the third text data includes at least one second character string.

[0064] The processing unit has a function of converting at least one second character string in the acquired third text data into a first character string that corresponds one-to-one to the second character string. This converts the third text data into fourth text data. This makes it possible to decrypt hidden or deleted confidential information. Furthermore, the second character string may be a character string that is not included in the first text data. This facilitates the conversion of the second character string into the first character string. In this specification, the process of converting the second character string into the first character string may be referred to as mask removal.

[0065] The above-described configuration makes it possible to prevent leakage of confidential information and to realize an information processing system that can be used safely by users.

[0066] The second character string does not need to include symbols. In this case, it is preferable that the second character string is not included in the first text data and is a hypernym, homonym, or hyponym of the first character string. This reduces the impact on the output content of the language model. Furthermore, by replacing the second character string with a character string related to the first character string, it is possible to prevent the relationship between words and the contextual characteristics of the first text data from being lost in the second text data. Therefore, it is possible to obtain a desired answer from the language model while preventing the leakage of confidential information.

[0067] In this specification, symbols refer to punctuation marks such as parentheses and accents, mathematical symbols, logical symbols, conjugation marks, etc., and exclude numbers and characters used in the language included in the first text data. For example, if the language of the first text data is English, symbols in this specification do not include alphabetic characters.

[0068] Furthermore, if the information processing system has a storage unit, it is preferable that character strings that may be confidential information are registered in the storage unit. A user can refer to the character strings registered in the storage unit and appropriately select a first character string from the first text data. Furthermore, the processing unit can refer to the character strings registered in the storage unit and extract the first character string from the first text data. This eliminates the overlooking of confidential information and prevents the leakage of confidential information.

[0069] The information processing system can also be referred to as an information processing device that can execute processing using an information processing method according to an embodiment of the present invention. Specifically, one embodiment of the present invention is an information processing method that receives first text data and at least one first character string included in the first text data, converts the at least one first character string in the first text data into a second character string that has a one-to-one correspondence with the first character string, thereby converting the first text data into second text data that includes the at least one second character string, transmits the second text data to a language model via a network, obtains third text data that includes the at least one second character string, and converts the at least one second character string in the third text data into a first character string that has a one-to-one correspondence with the second character string, thereby converting the third text data into fourth text data.

[0070] <Configuration Example 1 of Information Processing System> Fig. 1 shows an example of an information processing system according to this embodiment. In the information processing system shown in Fig. 1, an information processing device 10 is connected to an information processing device 40 via a network 30. The information processing device 10 is also connected to a plurality of information terminals 20 via a network 31. Fig. 1 shows information terminal 20a, information terminal 20b, information terminal 20c, and information terminal 20d as examples of the information terminals 20.

[0071] A user of the information processing system can access the information processing device 10 from the information terminals 20a to 20d, etc. The user can then receive a service using the information processing method according to one aspect of the present invention.

[0072] A user can use the information processing device 10, for example, by using dedicated software or an application installed on the information terminal 20. Alternatively, a user can use the information processing device 10, for example, from a web browser on the information terminal 20.

[0073] An organization such as a company that provides a service using an information processing method according to one embodiment of the present invention (hereinafter also referred to as a service provider) can provide the service using the information processing device 10. The service provider does not need to prepare the information processing device 40 on its own, but can use the information processing device 40 via the network 30.

[0074] The information processing device 10 is a device that can execute processing using an information processing method according to one aspect of the present invention. In FIG. 1, a server computer is illustrated as an example.

[0075] The information processing device 10 can perform information processing such as calculations using data input from the information terminal 20a via the network 31. The information processing device 10 can transmit the results of the information processing to the information terminal 20a via the network 31. This can reduce the calculation load on the information terminal 20a. Here, the information terminal 20a has been described as an example, but the same can be said for the information terminals 20b to 20d.

[0076] The information processing device 10 can input data to the information processing device 40 via the network 30. The information processing device 40 can perform information processing such as calculations using the input data. The information processing device 40 can transmit the results of the information processing to the information processing device 10 via the network 30. This can reduce the calculation load on the information processing device 10. The information processing device 10 can then transmit the data received from the information processing device 40 or data generated based on the data to the information terminal 20a via the network 31.

[0077] The information terminals 20a to 20d are each an information terminal device such as a computer used by a user, and may also be referred to as a client computer. FIG. 1 illustrates, as an example, the information terminal 20a, which is a desktop computer; the information terminal 20b, which is a notebook computer; the information terminal 20c, which is a smartphone; and the information terminal 20d, which is a tablet computer. Note that the information terminal 20d can also be used as a notebook computer by connecting it to a housing 21 having a keyboard. The number of information terminals connected to the information processing device 10 is not particularly limited. While FIG. 1 shows four information terminals, the number of information terminals may be one, two, three, five, or more. Examples of such information terminals include desktop information terminals, notebook information terminals, tablet information terminals, and mobile information terminals such as smartphones.

[0078] The information processing device 10 is a large computer such as a server computer or a supercomputer. The information processing device 10 is a computer with higher processing power than the information terminal 20. The information processing device 10 can function as a parallel computer. By using the information processing device 10 as a parallel computer, it is possible to perform large-scale calculations required for learning and inference of artificial intelligence (AI), for example. The information processing device 10 may also be capable of performing processing using a natural language processing model using AI.

[0079] The information processing device 40 is a large computer such as a server computer or a supercomputer. The information processing device 40 is a computer with higher processing power than the information terminal 20. The information processing device 40 is also a computer with higher processing power than the information processing device 10. It is preferable that the information processing device 40 has a function as a parallel computer. By using the information processing device 40 as a parallel computer, it is possible to perform large-scale calculations required for AI learning and inference, for example. Note that when both the information processing device 10 and the information processing device 40 have a function as a parallel computer, the information processing device 40 has a higher processing power and can perform large-scale calculations than the information processing device 10.

[0080] The information processing device 40 can perform processing using a natural language processing model that uses AI. Examples of the natural language processing model that uses AI include BERT (Bidirectional Encoder Representations from Transformers) and T5 (Text-to-Text Transfer Transformer). The information processing device 40 can also perform processing using a model that utilizes a large-scale language model (such as a sentence generation model or a dialogue model). Examples of large-scale language models include GPT-3, GPT-3.5, GPT-4, LaMDA (Language Model for Dialogue Applications), PaLM (Pathways Language Model), and PaLM2, and it is preferable to use GPT-4.

[0081] The information processing device 40 can execute processing using a general-purpose AI model that can perform various tasks. When the information processing device 10 and the information processing device 40 can each execute processing using a model that uses a large-scale language model, the information processing device 40 can execute processing using a larger-scale AI model than the information processing device 10.

[0082] Network 31 is a computer network smaller than network 30. Typically, network 30 is a global network, and network 31 is a local network. It is preferable to use the Internet, which is the foundation of the World Wide Web (WWW), as network 30. It is preferable to use an intranet or an extranet as network 31. In this embodiment, an example will be mainly described in which the Internet is used as network 30 and an intranet is used as network 31.

[0083] Other computer networks that can be used as the network 31 include a PAN (Personal Area Network), a LAN (Local Area Network), a CAN (Campus Area Network), a MAN (Metropolitan Area Network), a WAN (Wide Area Network), and a GAN (Global Area Network).

[0084] Furthermore, when wireless communication is performed, communication standards such as the fourth generation mobile communication system (4G), fifth generation mobile communication system (5G), and sixth generation mobile communication system (6G), or specifications standardized by the IEEE such as Wi-Fi (registered trademark) and Bluetooth (registered trademark), can be used as communication protocols or communication technologies.

[0085] The information processing system shown in FIG. 1 is suitable for a case where a user who uses the information terminal 20 and a service provider belong to the same organization such as a company.

[0086] For example, it is preferable that data be transmitted and received between the information terminal 20 and the information processing device 10 using a network 31 established within an organization such as a company. It is also preferable that data be transmitted and received between the information processing device 10 and the information processing device 40 using the Internet. This allows applications, tools, services, etc. provided outside the organization to be used more safely than when the information terminal 20 is directly connected to the information processing device 40 via the Internet. Furthermore, since the above-described masking process can be performed within the organization on the information processing device 10, it is possible to assist users in entering data, improve user convenience, and prevent the leakage of confidential information within the organization.

[0087] The flow of data in the information processing system shown in Fig. 1 will be described with reference to Fig. 2. In Fig. 2 and other figures, the order of data transmission and reception is indicated by circled numbers.

[0088] First, the user inputs desired data DA1 into the information processing device 10 using the information terminal 20. The data DA1 is transmitted from the information terminal 20 to the information processing device 10 via the network 31.

[0089] The data DA1 corresponds to a target sentence in a prompt that the user wants to input to the language model, or a prompt that the user wants to input to the dialogue model. Note that a prompt corresponds to an input sentence that causes the language model to perform a desired operation. When a prompt is given, the language model generates a response content based on the prompt. The format of the data DA1 is not particularly important, but it is preferably text data. Details of the data DA1 will be explained later in <Information Processing Method 1>.

[0090] When the data DA1 includes data other than text data, it is preferable to convert the data DA1 into text data in the information processing device 10. When various data are converted into text data, it is preferable to use AI technology.

[0091] The information processing device 10 preferably has a function of generating text data using image data. For example, the information processing device 10 can generate text data using image data by performing processing using at least one AI technology, such as image recognition, handwritten character recognition, or caption generation. The information processing device 10 may also have an optical character recognition (OCR) function.

[0092] Next, the information processing device 10 converts the data DA1 into data DA2. The following description will be given taking as an example a case where the data DA1 is text data. Note that if the data DA1 contains data other than text data, the data DA1 can be interpreted as text data generated using the data DA1. Details of the data DA2 will be described later in <Information Processing Method 1>.

[0093] The information processing device 10 can add text data to the data DA2. The text data corresponds to an instruction sentence in a prompt to be input to the language model. Here, the instruction sentence is part of the prompt to be input to the language model, and can be rephrased as a question sentence, an imperative sentence, or the like.

[0094] The text data added to data DA2 may be a system prompt. In this specification, a system prompt refers to an instruction in a prompt that is preset by a developer or a service provider. Presetting a system prompt makes it easier to obtain a desired answer from a language model.

[0095] Examples of text data to be added to data DA2 include instruction sentences such as "Please proofread target sentence A" and "Please translate target sentence A." Note that target sentence A corresponds to data DA2 before the text data is added.

[0096] The data DA2 obtained through such processing is transmitted from the information processing device 10 to the information processing device 40 via the network 30.

[0097] Next, data DB1 is transmitted from information processing device 40 to information processing device 10 via network 30. That is, information processing device 10 acquires data DB1 from information processing device 40 via network 30. Data DB1 is text data equivalent to a response sentence corresponding to data DA2 input to information processing device 40. Details of data DB1 will be explained later in <Information Processing Method 1>.

[0098] Next, the information processing device 10 converts the data DB1 into the data DB2. Details of the data DB2 will be explained later in the section <Information processing method 1>.

[0099] Next, the information processing device 10 outputs the data DB2 to the information terminal 20 via the network 31. In this way, the user can obtain the data DB2 corresponding to the response sentence to the data DA1.

[0100] For example, data DB2 may be converted into data other than text data and then output to information terminal 20. For example, if data DA1 is data other than text data, data DB2 is preferably data in the same format as data DA1. It is preferable that information processing device 10 has a function of generating image data using text data, for example.

[0101] Fig. 3 is a block diagram of the information processing system shown in Fig. 1. In Fig. 3, an example of the configuration of the information processing device 10 will be mainly described.

[0102] The information processing device 10 includes a receiving unit 110 , a storage unit 120 , a processing unit 130 , an output unit 140 , and a transmission path 150 .

[0103] In the drawings accompanying this specification, the components are classified by function and shown as independent blocks in the block diagrams. However, in reality, it is difficult to completely separate the components by function, and one component may be involved in multiple functions. For example, part of the processing unit 130 may function as the reception unit 110. Also, one function may be involved in multiple components. For example, the processing performed by the processing unit 130 may be executed by different servers depending on the processing.

[0104] [Reception Unit 110] The reception unit 110 can receive data from outside the information processing device 10. For example, the reception unit 110 receives data from the information terminal 20 via the network 31. At this time, examples of the data include first text data and at least one character string included in the first text data. The first text data corresponds to the above-mentioned data DA1. Furthermore, for example, the reception unit 110 receives data from the information processing device 40 via the network 30. At this time, examples of the data include the above-mentioned data DB1.

[0105] As described above, it is preferable that the data received by the receiving unit 110 is text data. Furthermore, if the data is data other than text data, it is preferable that the processing unit 130 generates text data using the data received by the receiving unit 110.

[0106] The data supplied to the reception unit 110 is supplied to one or both of the storage unit 120 and the processing unit 130 via the transmission path 150 .

[0107] [Storage Unit 120] The storage unit 120 has a function of storing a program executed by the processing unit 130. The storage unit 120 may also have a function of storing data generated by the processing unit 130 (e.g., calculation results, analysis results, inference results), data input to the reception unit 110, etc.

[0108] The storage unit 120 preferably has a function of storing character strings that may be confidential information. The storage unit 120 also preferably has a function of recording one or more types of information for the character strings selected from history, numerical values, categories, etc. Details of history, numerical values, categories, etc. for the character strings will be described later in <Modification of information processing method>.

[0109] The storage unit 120 may have a database. Furthermore, the information processing device 10 may have a database separate from the storage unit 120. The information processing device 10 may have a function to retrieve data from a database that exists outside the storage unit 120, outside the information processing device 10, or outside the information processing system. Furthermore, the information processing device 10 may have a function to retrieve data from both its own database and an external database.

[0110] A file server may be used instead of the database. For example, when using files stored in a file server, it is preferable that the database has paths to files stored in the file server.

[0111] The storage unit 120 includes at least one of a volatile memory and a non-volatile memory. Examples of the volatile memory include a dynamic random access memory (DRAM) and a static random access memory (SRAM). Examples of the non-volatile memory include a resistive random access memory (ReRAM), a phase change random access memory (PRAM), a ferroelectric random access memory (FeRAM), a magnetoresistive random access memory (MRAM), and a flash memory. The storage unit 120 may also include at least one of NOSRAM (registered trademark) and DOSRAM (registered trademark). The storage unit 120 may also include a recording media drive. Examples of recording media drives include a hard disk drive (HDD) and a solid state drive (SSD).

[0112] NOSRAM is an abbreviation for "Nonvolatile Oxide Semiconductor Random Access Memory (RAM)." NOSRAM refers to a memory in which memory cells are two-transistor (2T) or three-transistor (3T) gain cells and transistors (also called OS transistors) that use metal oxide in their channel formation regions. OS transistors have an extremely small leakage current, i.e., a current that flows between the source and drain in an off state. NOSRAM can be used as a nonvolatile memory by retaining a charge corresponding to data in the memory cell using its extremely small leakage current characteristic. In particular, NOSRAM can read stored data without destroying it (nondestructive readout), making it suitable for arithmetic processing in which only data read operations are repeated a large number of times. NOSRAM can increase its data capacity by stacking layers, and therefore can be used as a large-scale cache memory, main memory, or storage memory to improve the performance of semiconductor devices.

[0113] DOSRAM is an abbreviation for "Dynamic Oxide Semiconductor RAM" and refers to a RAM having 1T (transistor) 1C (capacitor) type memory cells. DOSRAM is a DRAM formed using OS transistors, and is a memory that temporarily stores information sent from an external device. DOSRAM is a memory that takes advantage of the low off-state current of OS transistors.

[0114] In this specification and the like, a metal oxide refers to an oxide of a metal in a broad sense. Metal oxides are classified into oxide insulators, oxide conductors (including transparent oxide conductors), oxide semiconductors (also referred to as oxide semiconductors or simply as OSs), and the like. For example, when a metal oxide is used for a semiconductor layer of a transistor, the metal oxide may be referred to as an oxide semiconductor.

[0115] The metal oxide included in the channel formation region preferably contains indium (In). When the metal oxide included in the channel formation region contains indium, the carrier mobility (electron mobility) of the OS transistor is increased. Furthermore, the metal oxide included in the channel formation region is preferably an oxide semiconductor containing element M. The element M is preferably at least one of aluminum (Al), gallium (Ga), and tin (Sn). Other elements applicable to element M include boron (B), silicon (Si), titanium (Ti), iron (Fe), nickel (Ni), germanium (Ge), yttrium (Y), zirconium (Zr), molybdenum (Mo), lanthanum (La), cerium (Ce), neodymium (Nd), hafnium (Hf), tantalum (Ta), and tungsten (W). However, a combination of two or more of the above elements may be used as element M. The element M is, for example, an element having a high bond energy with oxygen. For example, it is an element having a higher bond energy with oxygen than indium. The metal oxide contained in the channel formation region is preferably a metal oxide containing zinc (Zn), since zinc-containing metal oxides may be easily crystallized.

[0116] The metal oxide contained in the channel formation region is not limited to a metal oxide containing indium, but may be, for example, a metal oxide containing zinc but not indium, such as zinc tin oxide or gallium tin oxide, a metal oxide containing gallium, or a metal oxide containing tin.

[0117] [Processing Unit 130] The processing unit 130 has a function of performing processes such as calculation, analysis, and inference using data supplied from one or both of the receiving unit 110 and the storage unit 120. The processing unit 130 can supply the generated data (e.g., calculation results, analysis results, and inference results) to one or both of the storage unit 120 and the output unit 140.

[0118] The processing unit 130 has a function of acquiring data from the storage unit 120. The processing unit 130 may also have a function of recording or registering data in the storage unit 120.

[0119] The processing unit 130 has a function of converting the first text data into the second text data. The conversion from the first text data to the second text data is performed by converting a first character string included in the first text data into a second character string that is in one-to-one correspondence with the first character string. In other words, the second text data includes at least one second character string. The second text data corresponds to the data DA2 described above.

[0120] The processing unit 130 preferably has a function of selecting a second character string that corresponds one-to-one to the first character string. The second character string and a method for selecting the second character string will be described later in the information processing method.

[0121] The processing unit 130 preferably has a function of creating a dictionary that defines the correspondence between first character strings and second character strings.

[0122] The processing unit 130 may have a function of adding data to the second text data. For example, the data to be added to the second text data may be a part of a prompt. When the second text data is a target sentence in the prompt, the processing unit 130 may add the instruction sentence in the prompt to the second text data.

[0123] The processing unit 130 has a function of acquiring third text data by transmitting second text data to the language model via the network 30. The third text data includes at least one second character string. The third text data corresponds to the data DB1 described above.

[0124] The processing unit 130 has a function of converting the third text data into fourth text data. The conversion from the third text data to the fourth text data is performed by converting the second character string included in the third text data into a first character string that corresponds one-to-one. That is, the fourth text data includes at least one first character string. Furthermore, if a dictionary that defines the correspondence between the first character string and the second character string has been created, the processing unit 130 may have a function of converting the third text data into the fourth text data by referring to the dictionary. The fourth text data corresponds to the data DB 2 described above.

[0125] As described above, the processing unit 130 has the function of performing mask processing and mask removal.

[0126] The storage unit 120 preferably has a function for storing a plurality of third character strings. The third character strings are character strings that may contain confidential information. Furthermore, the third character strings are character strings that require conversion into second character strings.

[0127] It is preferable that the processing unit 130 has a function of registering at least one first character string in the storage unit 120 as a third character string when at least one first character string is not included in the storage unit 120.

[0128] When the receiving unit 110 receives a fourth character string that is one of at least one second character string, the processing unit 130 preferably has a function of updating the second text data by restoring the fourth character string included in the second text data to a third character string that corresponds one-to-one to the fourth character string. Furthermore, when the fourth character string is restored to the third character string, the processing unit 130 preferably has a function of recording in the storage unit 120 a history of the second character string being restored to the first character string that corresponds one-to-one to the fourth character string.

[0129] The processing unit 130 may have a function of converting a first character string included in the first text data into a second character string that is in one-to-one correspondence with the plurality of third character strings, by referring to one or more types selected from history, numerical values, categories, etc., for the plurality of third character strings recorded in the storage unit 120. The history, numerical values, categories, etc. for the plurality of third character strings will be described later in the information processing method.

[0130] The processing unit 130 may preset a target value for the range of the proportion of the second character string in the second text data, and may warn the user of the information processing system if the proportion of the second character string is outside that range. The greater the proportion of the second character string, the greater the loss of original information and the higher the security, but the greater the possibility that the context will be lost and the performance of the language model will not be fully utilized. The target range depends on the application or the required security level, but may be, for example, 25% to 75% or 35% to 50%.

[0131] The processing unit 130 may include, for example, an arithmetic circuit, a central processing unit (CPU), and a graphics processing unit (GPU).

[0132] The processing unit 130 may have a microprocessor such as a DSP (Digital Signal Processor). The microprocessor may be implemented by a PLD (Programmable Logic Device) such as an FPGA (Field Programmable Gate Array) or an FPAA (Field Programmable Analog Array). The processing unit 130 may also have a quantum processor. The processing unit 130 can perform various data processing and program control by interpreting and executing instructions from various programs using the processor. Programs that can be executed by the processor are stored in at least one of the memory area of ​​the processor and the storage unit 120.

[0133] The processing unit 130 may include a main memory. The main memory may include at least one of a volatile memory such as a random access memory (RAM) and a non-volatile memory such as a read-only memory (ROM). The main memory may also include at least one of the above-mentioned NOSRAM and DOSRAM.

[0134] The RAM may be, for example, a DRAM or an SRAM, and a virtual memory space is allocated and used as a working space for the processing unit 130. The operating system, application programs, program modules, program data, lookup tables, and the like stored in the storage unit 120 are loaded into the RAM for execution. The data, programs, and program modules loaded into the RAM are each directly accessed and operated by the processing unit 130.

[0135] The ROM can store a BIOS (Basic Input / Output System), firmware, etc., which do not require rewriting. Examples of ROM include mask ROM, OTPROM (One Time Programmable Read Only Memory), and EPROM (Erasable Programmable Read Only Memory). Examples of EPROMs include UV-EPROMs (Ultra-Violet Erasable Programmable Read Only Memories), which allow stored data to be erased by exposure to ultraviolet light, EEPROMs (Electrically Erasable Programmable Read Only Memories), and flash memories.

[0136] The processing section 130 can include one or both of an OS transistor and a transistor having silicon in a channel formation region (a Si transistor).

[0137] The processing unit 130 preferably includes an OS transistor. Because an OS transistor has an extremely small off-state current, using the OS transistor as a switch for retaining charge (data) flowing into a capacitor functioning as a memory element can ensure a long data retention period. By using this characteristic in at least one of the register and cache memory of the processing unit, the processing unit can be operated only when necessary, and can be turned off in other cases by saving information from the previous processing in the memory element. In other words, normally-off computing is possible, and the power consumption of the information processing system can be reduced.

[0138] It is preferable that the information processing device 10 uses AI for at least some of its processing.

[0139] In particular, it is preferable that the information processing device 10 uses an artificial neural network (ANN, hereinafter also simply referred to as a neural network). A neural network is realized by a circuit (hardware) or a program (software).

[0140] In this specification, a neural network refers to a general model that mimics the neural circuit network of a living organism, determines the connection strength between neurons through learning, and has problem-solving capabilities. A neural network has an input layer, an intermediate layer (hidden layer), and an output layer.

[0141] In this specification and the like, when discussing neural networks, determining the connection strengths (also called weighting coefficients) between neurons from existing information may be referred to as "learning."

[0142] In this specification and the like, the act of constructing a neural network using connection strengths obtained by learning and deriving a new conclusion from it may be referred to as "inference."

[0143] [Output Unit 140] The output unit 140 outputs data based on the processing result of the processing unit 130. The output unit 140 can supply at least one of the calculation result, analysis result, and inference result of the processing unit 130 to an external device of the information processing device 10.

[0144] The output unit 140 can output data to the outside of the information processing device 10. For example, the output unit 140 outputs data to the information terminal 20 via the network 31. The output unit 140 can output data converted by the processing unit 130 based on data received from the information processing device 40 to the information terminal 20. Alternatively, the output unit 140 can output the data itself received from the information processing device 40 to the information terminal 20.

[0145] The data can be presented, for example, by displaying it on the display screen of the information terminal 20 and / or by outputting a file in CSV format or the like.

[0146] Furthermore, the output unit 140 outputs data to the information processing device 40 via the network 30. The output unit 140 outputs data converted by the processing unit 130 based on the data received from the information terminal 20 to the information processing device 40.

[0147] [Transmission Path 150] The transmission path 150 has a function of transmitting data. Data can be transmitted and received between the reception unit 110, the storage unit 120, the processing unit 130, and the output unit 140 via the transmission path 150.

[0148] <Information Processing Method 1> Information processing method 1 of this embodiment includes the processes of steps S11 to S14 shown in Fig. 4. Fig. 5A to Fig. 5D are diagrams explaining information processing method 1. Fig. 5A to Fig. 5D show an example of a case where a language model is instructed to proofread a sentence.

[0149] [Step S11] In step S11, the receiving unit 110 receives first text data and at least one first character string. The at least one first character string is a character string included in the first text data. Specifically, the receiving unit 110 receives the first text data and the at least one first character string from the user via the information terminal 20. The first text data corresponds to the above-mentioned data DA1.

[0150] For example, a user inputs first text data to the information processing device 10 using the information terminal 20. The first text data is output from the information processing device 10 to a display unit of the information terminal 20. The user selects at least one first character string from the first text data output to the display unit of the information terminal 20. The accepting unit 110 accepts the selected at least one first character string.

[0151] In the information processing method 1 of the present embodiment, a case will be described in which the receiving unit 110 receives the text data 101 shown in Fig. 5A. In Fig. 5A, the text data 101 includes sentences 101a and 101b. Note that the text data 101 may include sentences other than sentences 101a and 101b.

[0152] 5A, the receiving unit 110 receives a character string 111a and a character string 111b as at least one first character string. Here, the receiving unit 110 receives "silicon oxide film" included in the sentence 101a as the character string 111a, and receives "aluminum" included in the sentence 101b as the character string 111b. Note that the sentence 101b shown in FIG. 5A includes the character string 111a, "silicon oxide film."

[0153] [Step S12] In step S12, the processing unit 130 converts the first text data into second text data. For example, in the first text data, at least one first character string is converted into a second character string that is in one-to-one correspondence with the first character string. As a result, the second text data includes at least one second character string.

[0154] The second character string is a character string including a first symbol. For example, the second character string is preferably a character string including the first symbol and at least one selected from a symbol different from the first symbol, an alphabet, and a number, and more preferably a character string including the first symbol and at least one selected from a symbol different from the first symbol, an alphabet, and a number.

[0155] In particular, it is preferable that the second character string be sandwiched between a pair of first symbols. Here, the pair of first symbols refers to a pair of a start symbol and a terminal symbol. In other words, it is preferable that symbols applicable as the pair of first symbols are punctuation marks. Examples of punctuation marks include parentheses, square brackets, curly brackets, tortoiseshell brackets, angle brackets, single quotes, and double quotes. An example of a pair of first symbols is a pair of an opening parenthesis and a closing parenthesis.

[0156] By replacing the second character string with a character string that is less related to the first character string, the possibility of confidential information being leaked can be significantly reduced. In addition, since the second character string can be prepared without using a concept dictionary or the like, the method for determining the second character string can be simplified. In addition, accurate decryption becomes easier.

[0157] The character string sandwiched between the pair of first symbols is referred to as a sixth character string. There are no particular limitations on the sixth character string. The sixth character string can include at least one selected from symbols different from the pair of first symbols, alphabets, and numbers. In this case, the second character string has a configuration in which the sixth character string is sandwiched between the pair of first symbols.

[0158] Furthermore, the sixth character string may include a word or an abbreviation thereof that indicates a category of the first character string. When a specific example of a compound is selected as the first character string, for example, "CHEM_1" or "COMPOUND1" may be used as the sixth character string. This is expected to preserve as much of the original context as possible while concealing confidential information.

[0159] In FIG. 5B, a character string 112a, "[MASK1]", is prepared as a second character string that corresponds one-to-one to character string 111a, and a character string 112b, "[MASK2]", is prepared as a second character string that corresponds one-to-one to character string 111b.

[0160] 5A and 5B, the processing unit 130 converts the character string 111a into the character string 112a, thereby converting the sentence 101a into the sentence 102a. The processing unit 130 also converts the character strings 111a and 111b into the character strings 112a and 112b, respectively, thereby converting the sentence 101b into the sentence 102b. As described above, the processing unit 130 converts the text data 101 into text data 102 that includes the sentences 102a and 102b. In other words, the processing unit 130 converts the text data 101 into text data 102 that includes at least one second character string.

[0161] Furthermore, it is preferable that the number of characters in the second string be the same as the number of characters in the corresponding first string. Alternatively, when the second string has a configuration in which a sixth string is sandwiched between a pair of first symbols, it is preferable that the number of characters in the sixth string be the same as the number of characters in the corresponding first string. Note that even when the second string has a configuration in which the sixth string is sandwiched between a pair of first symbols, the number of characters in the second string may be the same as the number of characters in the corresponding first string. By adopting such a configuration, it is possible to minimize the impact on input and output to the language model.

[0162] The second character string described above does not include the meaning of the first character string that corresponds one-to-one to the second character string. This allows the answer from the language model to be obtained in the form of the second character string itself. This also makes it easier to convert the second character string into the first character string (also called decoding), as described below.

[0163] Here, the processing unit 130 may generate a dictionary that defines the correspondence between the first character string and the second character string. For example, as shown in Fig. 6, it is preferable to create a dictionary that defines the correspondence between a character string 111 (the first character string accepted by the accepting unit 110) and a character string 112 (the second character string). By creating this dictionary, it is possible to convert the third text data into the fourth text data, which will be described later, in a short time.

[0164] By performing step S12, confidential information included in the first text data can be hidden or deleted, and text data that can prevent the confidential information from being leaked can be created.

[0165] The second text data corresponds to a prompt to be input to the language model. When the second text data corresponds to a target sentence in the prompt, it is preferable to add a part of the prompt to the second text data. When the second text data corresponds to a target sentence in the prompt, it is preferable to add an instruction sentence in the prompt to the second text data. In this case, the processing unit 130 can add a part of the prompt to the second text data. Note that when the second text data includes a target sentence and an instruction sentence in the prompt, the second text data can be sent to the language model in its original form.

[0166] [Step S13] In step S13, the processing unit 130 acquires third text data by transmitting the second text data to the language model via the network 30. The second text data corresponds to the above-mentioned data DA2, and the third text data corresponds to the above-mentioned data DB1.

[0167] In the information processing method 1 of this embodiment, the instruction sentence in the prompt is prepared as "Please point out any unnatural parts in the text data 101." At this time, by performing processing using a language model, it is possible to acquire text data 103, which is third text data (see FIG. 5C ). The text data 103 includes sentence 103a and sentence 103b. The sentence 103a includes character string 112a, and the sentence 103b includes character string 112a and character string 112b. In other words, the text data 103 includes character string 112a and character string 112b, which are second character strings. In other words, the third text data includes at least one second character string.

[0168] [Step S14] In step S14, the processing unit 130 converts the third text data into fourth text data. For example, in the third text data, at least one second character string is converted into a first character string that corresponds one-to-one to the second character string. As a result, the fourth text data includes at least one first character string. The fourth text data corresponds to the data DB2 described above.

[0169] 5C and 5D, the processing unit 130 converts the character string 112a into the character string 111a, thereby converting the sentence 103a into the sentence 104a. The processing unit 130 also converts the character strings 112a and 112b into the character strings 111a and 111b, respectively, thereby converting the sentence 103b into the sentence 104b. As described above, the processing unit 130 converts the text data 103 into text data 104 including the sentences 104a and 104b. In other words, the processing unit 130 converts the text data 103 into text data 104 including at least one first character string.

[0170] By performing step S14, the confidential information that was hidden or deleted by the processing of step S12 can be decrypted.

[0171] When a dictionary is created in step S12, the third text data can be converted into the fourth text data in a short time by referring to the dictionary.

[0172] As a result, when an external service that provides a language model is used, it is possible to prevent confidential information from being leaked.

[0173] This concludes the description of information processing method 1.

[0174] <Information Processing Method 2> Information processing method 2 of the present embodiment, like information processing method 1 described above, includes the processes of steps S11 to S14 shown in Fig. 4. The main difference between information processing method 2 and information processing method 1 described above is the configuration of the second character string, etc. Hereinafter, differences from information processing method 1 described above will be mainly described, and overlapping parts will not be described.

[0175] In step S12 of information processing method 2, first to third processes are performed on each of at least one first character string.

[0176] As a first process, the processing unit 130 selects a plurality of fourth character strings for a first character string. The plurality of fourth character strings are candidates for a second character string corresponding to the first character string. Since the plurality of fourth character strings are candidates for a second character string, it is preferable that the fourth character strings are character strings that are not included in the first text data. This facilitates conversion from the third text data to the fourth text data.

[0177] Furthermore, the fourth character string is preferably a hypernym, a synonym, or a hyponym of the first character string. In other words, the fourth character string does not need to include a symbol. In this case, the second character string is also a hypernym, a synonym, or a hyponym of the first character string. This reduces the impact on the output content of the language model. Furthermore, by replacing the first character string with a character string related to the first character string, it is possible to prevent the relationship between words and the contextual characteristics of the first text data from being impaired in the second text data. Therefore, it is possible to obtain a desired answer from the language model while preventing the leakage of confidential information.

[0178] Note that hypernyms, appositives, or hyponyms may be found by referring to a concept dictionary or the like. A concept dictionary is a list of words with classifications, relationships with other words, and the like. The concept dictionary may be an existing concept dictionary. Alternatively, a user may create a concept dictionary specialized for a document field. Alternatively, a user may add words that are likely to be used in a document field to a general-purpose concept dictionary.

[0179] 7B , by performing the first process, character strings 113a_1, 113a_2, and 113a_3 are selected as multiple fourth character strings for character string 111a. Note that the number of selected fourth character strings is not limited to three, and may be two, four, or more.

[0180] As the second process, the processing unit 130 performs a verification process for each of the plurality of fourth character strings. Note that as the verification process, it is preferable to perform, for example, the following process.

[0181] First, a plurality of sentences are prepared. One of the plurality of sentences is a first sentence containing a first character string. The remaining plurality of sentences are a plurality of second sentences obtained by converting the first character string contained in the first sentence into each of a plurality of fourth character strings. In other words, the plurality of sentences are composed of the first sentence and a plurality of second sentences.

[0182] 7B , by performing the second process, a sentence 114a is prepared as a first sentence, and sentences 114a_1 to 114a_3 are prepared as a plurality of second sentences. Here, the sentence 114a includes the character string 111a, and the sentences 114a_1 to 114a_3 include the character strings 113a_1 to 113a_3, respectively.

[0183] Next, natural language processing is performed to determine the plurality of sentences, and a score of the natural language processing is calculated. Evaluation criteria such as GLUE (General Language Understanding Evaluation) may be used for the natural language processing. Scores may be calculated according to each criterion, or accuracy rate, precision, reproducibility, F1 score, BLEU (BiLingual Evaluation Understudy), etc. may be used.

[0184] The above is an example of the verification process.

[0185] As a third process, the processing unit 130 refers to the result of the verification process and selects one of the plurality of fourth character strings. The processing unit 130 sets the selected fourth character string as the second character string corresponding to the first character string.

[0186] In addition, when the score of the natural language processing is referred to in the third process, in the third process, a fourth character string contained in a second sentence among a plurality of second sentences that has a score closest to that of the first sentence can be set as the second character string that corresponds one-to-one to the first character string.

[0187] As shown in Fig. 7C, the score calculated by performing the natural language processing may be output together with the sentence. In the example shown in Fig. 7C, it can be seen that the sentence 114a_1 has the closest score to the sentence 114a. Therefore, the character string 113a_1 included in the sentence 114a_1 can be set as a second character string that is in one-to-one correspondence with the character string 111a.

[0188] In addition, the second character string that is associated one-to-one with the character string 111b can be selected in the same manner.

[0189] By performing the first process to the third process on each of at least one first character string, it is possible to prepare at least one second character string that corresponds one-to-one to at least one first character string. In information processing method 2 of this embodiment, the second character string corresponding to character string 111a is set to character string 113a, and the second character string corresponding to character string 111b is set to character string 113b.

[0190] In FIG. 7D , "silicon compound" is prepared as the character string 113a, and "metal" is prepared as the character string 113b. Here, the character string 113a is a hypernym of the character string 111a, and the character string 113b is a hypernym of the character string 111b. At this time, the processing unit 130 converts the character string 111a into the character string 113a, thereby converting the sentence 101a into the sentence 102a. Furthermore, by converting the character strings 111a and 111b into the character strings 113a and 113b, respectively, the processing unit 130 converts the sentence 101b into the sentence 102b. As described above, the processing unit 130 converts the text data 101 into text data 102 including the sentences 102a and 102b. In other words, the processing unit 130 converts the text data 101 into text data 102 including at least one second character string.

[0191] The fourth character string may include a symbol. For example, the fourth character string may have a configuration in which a hypernym, a synonym, or a hyponym of the first character string is sandwiched between a pair of the symbols. In this case, the fourth character string may also have a configuration in which a hypernym, a synonym, or a hyponym of the first character string is sandwiched between a pair of the symbols.

[0192] As a result, when an external service that provides a language model is used, it is possible to prevent confidential information from being leaked.

[0193] This concludes the description of information processing method 2.

[0194] In the information processing method according to one aspect of the present invention, it is preferable that the first character string and the second character string have a one-to-one correspondence. With this configuration, accurate decoding is possible.

[0195] Although the information processing method according to one embodiment of the present invention illustrates a configuration in which a first character string and a second character string are associated one-to-one, the present invention is not limited to this. For example, two or more second character strings may be associated with one first character string. Specifically, in the third process described above, two or more fourth character strings may be selected and used as the second character strings corresponding to the first character string. This configuration can eliminate the possibility of confidential information being read from the context. Therefore, leakage of confidential information can be prevented.

[0196] In the information processing method according to one aspect of the present invention, a third character string is preferably registered in the storage unit 120. For example, the third character string may be a character string that has been converted into a second character string in the past. In other words, the third character string may be a character string that may contain confidential information. The number of third character strings registered in the storage unit 120 may be one or more.

[0197] As shown in FIG. 8, the information processing method according to one embodiment of the present invention preferably includes step S16 in addition to steps S11 to S14.

[0198] [Step S16] In step S16, it is determined whether or not the at least one first character string received by the receiving unit 110 in step S11 is registered in the storage unit 120. If it is determined that the at least one first character string is not registered in the storage unit 120, the processing unit 130 registers the character string that is not registered in the storage unit 120, among the at least one first character string, in the storage unit 120 as a third character string.

[0199] By performing the process of step S16, character strings that may contain confidential information can be stored in the storage unit 120. Therefore, by referring to the third character strings registered in the storage unit 120, the user can ascertain character strings that have been previously treated as character strings that may contain confidential information. In other words, the user can determine whether or not confidential information is included in text data to be transmitted to an external service. Therefore, leakage of confidential information can be prevented. Furthermore, character strings that may contain confidential information can be shared within an organization, thereby preventing leakage of confidential information throughout the organization.

[0200] 8 shows a configuration in which the process of step S16 is performed between the process of step S11 and the process of step S12, but the present invention is not limited to this. For example, the process of step S16 may be performed between the process of step S12 and the process of step S13.

[0201] <Modification of Information Processing Method> The following describes a modification of the information processing method that differs in part from the above. In the modification of the information processing method, it is assumed that a third character string is registered in the storage unit 120. In the following, parts that overlap with the above will be referred to, and descriptions thereof will be omitted.

[0202] [Variation 1] Variation 1 of the information processing method includes processes of steps S17 to S19 between the process of step S12 and the process of step S13, as shown in Fig. 9. Figs. 10A to 10C are diagrams illustrating Variation 1 of the information processing method.

[0203] In the first modification of the information processing method, it is assumed that the history for the third character string is recorded in the storage unit 120 .

[0204] In the information processing method variation 1, a case will be described in which the receiving unit 110 receives text data 101 shown in Fig. 10A in step S11. In Fig. 10A, the text data 101 includes sentences 101a to 101c. For sentences 101a and 101b, the description of the information processing method 1 above can be referred to. Sentence 101c includes character string 111a.

[0205] In step S12, the processing unit 130 converts the first text data into second text data. As shown in Figures 10A and 10B, the processing unit 130 converts the character string 111a into the character string 112a, thereby converting the sentence 101c into the sentence 102c. As described above, the processing unit 130 converts the text data 101 into text data 102 including the sentences 102a to 102c.

[0206] [Step S17] In step S17, the receiving unit 110 receives a fifth character string. The fifth character string is one of the at least one second character string prepared in step S12. Note that in step S17, the receiving unit 110 may also receive a first character string that corresponds one-to-one to the fifth character string.

[0207] 10B , character string 112a included in sentence 102c is selected. At this time, the receiving unit 110 receives character string 112a included in sentence 102c as a fifth character string. The first character string that corresponds one-to-one to the fifth character string is, for example, a character string that is determined to be unlikely to be confidential information. For example, the user determines, from the context of sentence 101c, that character string 111a in sentence 101c is unlikely to be confidential information, and selects character string 112a included in sentence 102c.

[0208] [Step S18] In step S18, the processing unit 130 updates the second text data by restoring the fifth character string included in the second text data to the third character string that has a one-to-one correspondence with the fifth character string. The third character string that has a one-to-one correspondence with the fifth character string can also be considered as the first character string that has a one-to-one correspondence with the fifth character string.

[0209] In FIG. 10C, the processing unit 130 restores the character string 112a included in the sentence 102c to the character string 111a, and the sentence 102c is decoded into the sentence 101c, thereby updating the text data 102.

[0210] [Step S19] In step S19, the processing unit 130 records in the storage unit 120 the history of the fifth character string being converted back into the third character string. The processing unit 130 adds this history to the history for the third character string stored in the storage unit 120. In other words, among the third character strings recorded in the storage unit 120, a third character string that appears frequently in the history is estimated to be a character string that is unlikely to contain confidential information. The user can select a first character string by referring to the history recorded in the storage unit 120. This prevents the number of character strings converted into second character strings from increasing more than necessary, thereby enabling the desired answer to be obtained from the language model. Furthermore, by referring to the history recorded in the storage unit 120, the possibility that each character string contains confidential information can be shared within the organization, thereby preventing the leakage of confidential information throughout the organization.

[0211] Note that the processing unit 130 may record a sentence including a third character string that has a one-to-one correspondence with the fifth character string in the storage unit 120. With this configuration, when the first text data includes a plurality of third character strings, the user can easily determine whether to select each of the plurality of third character strings as the first character string by referring to the sentences recorded in the storage unit 120.

[0212] The above is the description of the first modification.

[0213] [Modification 2] Modification 2 of the information processing method differs from the above-described information processing method in that storage unit 120 is referenced to determine whether or not to convert a first character string into a second character string.

[0214] For example, as described in the first modification example above, when a history for a third character string is recorded in the storage unit 120, the processing unit 130 can refer to the storage unit 120 and determine whether or not to convert at least one first character string received by the receiving unit 110 into a second character string. For example, the processing unit 130 may not need to convert a character string with a high frequency of occurrence in the history among the at least one character string into a second character string. This configuration can prevent the number of character strings converted into second character strings from increasing more than necessary. Therefore, a desired answer can be obtained from the language model.

[0215] Furthermore, for example, a numerical value can be assigned to the third character string registered in the storage unit 120. The numerical value for the third character string may be set to be smaller, for example, as the character string becomes more likely to contain confidential information.

[0216] For example, a past text corpus may be used to calculate the frequency of character strings (mainly noun phrases) included in the first text data, and the calculated frequency may be used as a numerical value assigned to the character string. Examples of the text corpus include text data managed within an organization to which a user of an information processing system according to an embodiment of the present invention belongs, or text data disclosed by the organization (e.g., patent application documents). Character strings with high frequencies are common and can be assumed to be safe to disclose. On the other hand, character strings with low frequencies are rare and can be assumed to be potentially confidential information.

[0217] When a numerical value is assigned to the third character string, the processing unit 130 can refer to the storage unit 120 and determine whether or not to convert at least one first character string received by the receiving unit 110 into a second character string. For example, the processing unit 130 may not convert a character string having a numerical value greater than a specific value into a second character string among the at least one first character string. Alternatively, if a character string having a numerical value less than a specific value has not been accepted as a first character string, the processing unit 130 may convert the character string into a second character string. This configuration can prevent the number of character strings converted into second character strings from increasing more than necessary. Therefore, a desired answer can be obtained from the language model.

[0218] Furthermore, for example, a category can be assigned to the third character string registered in the storage unit 120. The category for the third character string can be extracted using, for example, named entity recognition (NER). Named entity recognition refers to a technique for extracting proper nouns such as people's names or place names, expressions related to dates, times, etc. from character strings.

[0219] When a category is assigned to the third character string, the processing unit 130 can refer to the storage unit 120 and determine whether to convert a first character string assigned a specific category, among at least one first character string received by the receiving unit 110, into a second character string. For example, the processing unit 130 may not perform processing to convert the at least one first character string assigned a specific category into a second character string. Alternatively, the processing unit 130 may perform processing to convert the first character string into a second character string. This configuration allows human readability to be maintained to a certain extent in the text data (second text data) after the masking process. Furthermore, by including categorization in the prompt of the language model, it is possible to reduce the impact on the output content of the language model.

[0220] Furthermore, for example, the processing unit 130 can calculate the proportion of the second character string in the second text data. In this case, the processing unit 130 may convert the first text data to the second text data so that the proportion is 50% or less. This configuration can prevent the number of character strings converted to the second character string from increasing more than necessary. Therefore, a desired answer can be obtained from the language model.

[0221] The above is the explanation of the second modification.

[0222] [Modification 3] Modification 3 of the information processing method differs from the above-described information processing method in that processing unit 130 refers to storage unit 120 to extract the first character string.

[0223] If a third character string is registered in the storage unit 120, in step S11, the receiving unit 110 receives first text data, and the processing unit 130 can extract the third character string included in the first text data by referring to the storage unit 120. At this time, the extracted third character string can be treated as the first character string described above. Note that the number of extracted third character strings may be one or more. In this way, at least one first character string can be extracted from the first text data without user specification. In step S12, the processing unit 130 can convert the extracted first character string into a second character string, thereby converting the first text data into second text data.

[0224] As described in the above-mentioned variants 1 and 2, when a history for the third character string is recorded in the memory unit 120, the processing unit 130 may perform a process of converting the third character string included in the first text data and having a low historical frequency into a second character string that corresponds one-to-one.

[0225] Furthermore, as described in the above-described second modification, when a category for a third character string is recorded in storage unit 120, processing unit 130 may perform processing to convert a third character string that is included in the first text data and that includes a specific category into a second character string that corresponds one-to-one. Alternatively, processing unit 130 may perform processing to convert a third character string that is included in the first text data and that does not include a specific category into a second character string that corresponds one-to-one.

[0226] Furthermore, as described in the above-described second modification, when a numerical value for a third character string is recorded in storage unit 120, processing unit 130 may convert a third character string included in the first text data whose numerical value is smaller than a specific value into a second character string having a one-to-one correspondence. Alternatively, processing unit 130 may convert the third character string included in the first text data into the second character string in ascending numerical order so that the proportion of the second character string in the second text data is, for example, 10%.

[0227] With the above configuration, the processing unit 130 can select character strings that have been previously treated as potentially confidential information without the user having to specify them, thereby eliminating the overlooking of confidential information and preventing the leakage of confidential information.

[0228] <Configuration Example 2 of Information Processing System> Fig. 11 shows another example of the information processing system of this embodiment. In the information processing system shown in Fig. 11, an information processing device 11 is connected to an information processing device 40 via a network 30. The information processing device 11 is also connected to an information processing device 15 and a plurality of information terminals 20 via a network 31. In Fig. 11, information terminals 20a and 20b are shown as examples of the information terminals 20.

[0229] 11 is mainly used to relay transmission and reception of data between the information terminal 20 and the information processing device 40. Then, in the information processing device 15, data DA2 is generated using data DA1.

[0230] First, the user inputs desired data DA1 into the information processing device 11 using the information terminal 20. The data DA1 is transmitted from the information terminal 20 to the information processing device 11 via the network 31.

[0231] Next, the information processing device 11 transmits the data DA1 to the information processing device 15 via the network 31.

[0232] Next, the information processing device 15 generates data DA2 using the data DA1. The information processing device 15 is configured to execute the various processes performed by the information processing device 10 described above.

[0233] Next, the information processing device 15 transmits the data DA2 to the information processing device 11 via the network 31.

[0234] Next, the information processing device 11 transmits the data DA2 to the information processing device 40 via the network 30.

[0235] Next, the data DB1 is transmitted from the information processing device 40 to the information processing device 11 via the network 30. The data DB1 is text data equivalent to a response sentence corresponding to the data DA2 input to the information processing device 40.

[0236] Next, the information processing device 11 transmits the data DB2 to the information terminal 20 via the network 31. The data DB2 is data based on the data DB1 input to the information processing device 11. As a result, the user can obtain the data DB2 corresponding to the response sentence to the data DA1.

[0237] The data DB2 can be generated by the information processing device 11 or the information processing device 15. For example, the information processing device 11 may input the data DB1 input from the information processing device 40 to the information processing device 15. Then, the information processing device 15 may generate the data DB2 based on the data DB1.

[0238] The information processing system shown in FIG. 11 is suitable for a case where a user who uses information terminal 20 and a service provider belong to the same organization such as a company, similar to the information processing system shown in FIG.

[0239] For example, it is preferable that data be transmitted and received between the information terminal 20 and the information processing device 11 using a network 31 established within an organization such as a company. It is also preferable that data be transmitted and received between the information processing device 15 and the information processing device 11 using a network 31 established within an organization such as a company. It is also preferable that data be transmitted and received between the information processing device 11 and the information processing device 40 using the Internet. This allows for more secure use of applications, tools, services, etc. provided outside the organization than when the information terminal 20 is directly connected to the information processing device 40 via the Internet. However, connecting the information processing device 15 to the Internet may raise the risk of external cyberattacks, etc. The information processing system shown in FIG. 11 is preferable from a security perspective because it does not require the information processing device 15 to be directly connected to the information processing device 40 via the Internet.

[0240] Fig. 12 is a block diagram of the information processing system shown in Fig. 11. In Fig. 12, an example of the configuration of the information processing device 11 and the information processing device 15 will be mainly described.

[0241] The information processing device 11 includes a receiving unit 110 , a storage unit 120 , a processing unit 130 , an output unit 140 , and a transmission path 150 .

[0242] For the reception unit 110, storage unit 120, output unit 140, and transmission path 150 of the information processing device 11, the descriptions of the information processing device 10 can be referred to. The following mainly describes the differences from the information processing device 10.

[0243] The information processing device 15 includes a receiving unit 115 , a storage unit 125 , a processing unit 135 , an output unit 145 , and a transmission path 155 .

[0244] The receiving unit 110 receives data from the output unit 145 via the network 31 .

[0245] The receiving unit 115 receives data from the output unit 140 via the network 31 .

[0246] The information supplied to the reception unit 115 is supplied to one or both of the storage unit 125 and the processing unit 135 via the transmission path 155 .

[0247] The storage unit 125 has a function of storing a program executed by the processing unit 135. The storage unit 125 may also have a function of storing data generated by the processing unit 135 (e.g., calculation results, analysis results, inference results), data input to the reception unit 115, etc. The storage unit 125 may have various configurations applicable to the storage unit 120.

[0248] The processing unit 135 has a function of performing processes such as calculation, analysis, and inference using data supplied from one or both of the reception unit 115 and the storage unit 125. The processing unit 135 can supply generated data (e.g., calculation results, analysis results, and inference results) to one or both of the storage unit 125 and the output unit 145. The processing unit 135 can use various configurations applicable to the processing unit 130.

[0249] The processing unit 135 has a function of acquiring data from the storage unit 125 .

[0250] The processing unit 135 uses the data accepted by the accepting unit 115 to convert the first text data into second text data.

[0251] The information processing device 15 preferably uses AI for at least a part of the processing, and preferably uses a neural network.

[0252] The output unit 140 outputs the data to the reception unit 115 via the network 31 .

[0253] The output unit 145 outputs information based on the processing result of the processing unit 135. The output unit 145 can supply at least one of the calculation result, analysis result, and inference result of the processing unit 135 to an external device of the information processing device 15.

[0254] The output unit 145 outputs data to the reception unit 110 of the information processing device 10 via the network 31 .

[0255] The transmission path 155 has a function of transmitting data. Data can be transmitted and received between the reception unit 115, the storage unit 125, the processing unit 135, and the output unit 145 via the transmission path 155.

[0256] The processing section 130 and the processing section 135 can each include one or both of a transistor having a metal oxide in a channel formation region (OS transistor) and a transistor having silicon in a channel formation region (Si transistor).

[0257] <Configuration Example 3 of Information Processing System> Fig. 13 shows an example of an information processing system according to this embodiment. In the information processing system shown in Fig. 13, an information processing device 10 is connected to an information processing device 40 and a plurality of information terminals 20 via a network 30. In Fig. 13, the above-mentioned information terminals 20a to 20d are shown as examples of the information terminals 20.

[0258] The flow of data in the information processing system shown in FIG. 13 can be understood by referring to the description of FIG. 2 by replacing the network 31 with the network 30.

[0259] A user of the information processing system can access the information processing device 10 from an information terminal 20. The user can then receive a service using the information processing method according to one aspect of the present invention.

[0260] The information processing system shown in FIG. 13 is suitable for a case where a user who uses the information terminal 20 and a service provider belong to different organizations such as companies.

[0261] The information processing device 10 is connected to an information terminal 20 used by a user who receives a service via a network 30 (typically, the Internet). A service provider can provide the service using the information processing device 10. Data is also transmitted and received between the information processing device 10 and the information processing device 40 using the network 30 (typically, the Internet). This allows applications, tools, services, etc. using the information processing device 40 to be used more safely than when the information terminal 20 is directly connected to the information processing device 40 via the Internet.

[0262] <Configuration Example 4 of Information Processing System> Fig. 14 shows an example of an information processing system according to this embodiment. In the information processing system shown in Fig. 14, an information processing device 11 is connected to an information processing device 40 and a plurality of information terminals 20 via a network 30. In Fig. 14, the above-mentioned information terminals 20a to 20d are shown as examples of the information terminals 20. Furthermore, the information processing device 11 is connected to an information processing device 15 via a network 31.

[0263] A user of the information processing system can access the information processing device 11 from an information terminal 20. The user can then receive services using the information processing method according to one aspect of the present invention.

[0264] The information processing system shown in FIG. 14, like the information system shown in FIG. 13, is suitable for a case where a user who uses information terminal 20 and a service provider belong to different organizations such as companies.

[0265] The information processing device 11 is connected to an information terminal 20 used by a user receiving a service via the Internet. Preferably, data is transmitted and received between the information processing device 11 and the information processing device 15 using a network 31 established within an organization such as a company that provides the service. The service provider can provide the service using the information processing device 11 and the information processing device 15. Data is also transmitted and received between the information processing device 11 and the information processing device 40 using the Internet. This allows applications, tools, services, etc. using the information processing device 40 to be used more safely than when the information terminal 20 is directly connected to the information processing device 40 via the Internet. Furthermore, the information processing device 15 does not need to be directly connected to the information processing device 40 via the Internet, which is preferable from the perspective of security.

[0266] According to one aspect of the present invention, it is possible to provide an information processing system or an information processing method that can prevent leakage of confidential information. Also, according to one aspect of the present invention, it is possible to provide an information processing system or an information processing method that generates text data that can prevent leakage of confidential information. Also, according to one aspect of the present invention, it is possible to provide an information processing system or an information processing method that can be used safely by users.

[0267] A plurality of configuration examples shown in this embodiment mode can be combined as appropriate.

[0268] 10: Information processing device, 11: Information processing device, 15: Information processing device, 20a: Information terminal, 20b: Information terminal, 20c: Information terminal, 20d: Information terminal, 20: Information terminal, 21: Housing, 30: Network, 31: Network, 40: Information processing device, 101a: Sentence, 101b: Sentence, 101c: Sentence, 101: Text data, 102a: Sentence, 102b: Sentence, 102c: Sentence, 102: Text data, 103a: Sentence, 103b: Sentence, 103: Text data, 104a: Sentence, 10 4b: sentence, 104: text data, 110: reception unit, 111a: character string, 111b: character string, 112a: character string, 112b: character string, 113a: character string, 113a_1: character string, 113a_2: character string, 113a_3: character string, 113b: character string, 114a: sentence, 114a_1: sentence, 114a_3: sentence, 115: reception unit, 120: storage unit, 125: storage unit, 130: processing unit, 135: processing unit, 140: output unit, 145: output unit, 150: transmission path, 155: transmission path

Claims

A reception unit and a processing unit are included. the receiving unit has a function of receiving first text data and at least one first character string included in the first text data; The processing unit includes: a process of converting at least one of the first character strings in the first text data into a second character string in one-to-one correspondence with the first character string, thereby converting the first text data into second text data including at least one of the second character strings; transmitting the second text data to a language model via a network to obtain third text data including at least one of the second character strings; a process of converting at least one of the second character strings in the third text data into the first character string having one-to-one correspondence with the second character string, thereby converting the third text data into fourth text data; The second string includes symbols. 　 In claim 1, The first string includes confidential information. 　 In claim 1 or 2, An information processing system, wherein the second character string is a character string composed of the symbol and at least one selected from a symbol different from the symbol, an alphabet, and a number. 　 In claim 1 or 2, An information processing system, wherein the number of characters of the second string is equal to the number of characters of the corresponding first string. 　 In claim 1 or 2, the second character string has a configuration in which a sixth character string is sandwiched between a pair of the symbols, An information processing system, wherein the sixth character string includes at least one selected from a symbol different from the symbol, an alphabet, and a number. 　 In claim 5, An information processing system, wherein the number of characters of the sixth string matches the number of characters of the corresponding first string. 　 In claim 1 or 2, the processing unit is configured to execute a process of creating a dictionary that defines a correspondence relationship between the first character string and the second character string; The fourth text data is converted from the third text data by referring to the dictionary. 　 In claim 1 or 2, A storage unit is provided, the storage unit includes a plurality of third character strings that are requested to be converted into the second character strings; The processing unit is configured to execute a process of registering at least one of the first character strings in the memory unit as the third character string when at least one of the first character strings is not included in the memory unit. 　 In claim 8, the reception unit has a function of receiving a fourth character string that is one of at least one of the second character strings; The processing unit includes: updating the second text data by returning the fourth character string included in the second text data to the third character string that has a one-to-one correspondence with the fourth character string with reference to the storage unit; and recording in the storage unit a history of the fourth character string being returned to the third character string having a one-to-one correspondence therewith. 　 In claim 9, the receiving unit has a function of receiving fifth text data, When a third character string that has a high history frequency among the multiple third character strings is included in the fifth text data, the processing unit does not convert the third character string to the fourth character string that has a one-to-one correspondence with the third character string. 　 In claim 8, a numerical value is assigned to each of the plurality of third character strings so that the numerical value is set to be smaller as the character string becomes more likely to be confidential information; An information processing system, wherein in the process of converting the first text data into the second text data, a third string having a smaller numerical value among at least one of the first strings is converted into the second string with which it corresponds one-to-one. 　 In claim 8, a category extracted by using named entity extraction is assigned to each of the third character strings; An information processing system, wherein in the process of converting the first text data into the second text data, a third character string among at least one of the first character strings, which is assigned a specific category, is converted into the second character string with which it corresponds one-to-one. 　 In claim 1 or 2, an information processing method, wherein the second text data is converted from the first text data such that a proportion of the second character string in the second text data is 35% or more and 50% or less. 　 A reception unit and a processing unit are included. the receiving unit has a function of receiving first text data and at least one first character string included in the first text data; The processing unit includes: a process of converting at least one of the first character strings in the first text data into a second character string in one-to-one correspondence with the first character string, thereby converting the first text data into second text data including at least one of the second character strings; transmitting the second text data to a language model via a network to obtain third text data including at least one of the second character strings; a process of converting at least one of the second character strings in the third text data into the first character string that has a one-to-one correspondence with the second character string, thereby converting the third text data into fourth text data; an information processing system, in which, in the process of converting the first text data into the second text data, for each of at least one of the first strings, a plurality of third strings not included in the first text data are selected, a verification process is performed for each of the plurality of third strings, and a process of selecting the second string that has a one-to-one correspondence with the first string from among the plurality of third strings by referring to a result of the verification process, thereby converting the first string into the second string. 　 In claim 14, The first string includes confidential information. 　 In claim 15, In the verification process, preparing a plurality of sentences including a first sentence including the first character string and a plurality of second sentences obtained by converting the first character string included in the first sentence into each of a plurality of third character strings; A process of calculating a score of the natural language processing by performing a natural language processing for determining the plurality of sentences; In the process performed by referring to the results of the verification process, a third string contained in a second sentence among the plurality of second sentences that has a score closest to the score of the first sentence is determined to be a second string that corresponds one-to-one to the first string. 　 In any one of claims 14 to 16, An information processing system, wherein the second character string is a hypernym, an appositive word, or a hyponym of the corresponding first character string. 　 In any one of claims 14 to 16, An information processing system, wherein the second character string has a configuration in which a hypernym, an homonym, or a hyponym of the corresponding first character string is sandwiched between a pair of symbols. 　 In any one of claims 14 to 16, the processing unit is configured to execute a process of creating a dictionary that defines a correspondence relationship between a first character string and a second character string; The fourth text data is converted from the third text data by referring to the dictionary. 　 receiving first text data and at least one first character string included in the first text data; converting the at least one first character string in the first text data into a second character string in one-to-one correspondence with the first character string, thereby converting the first text data into second text data including at least one second character string; transmitting the second text data to a language model via a network to obtain third text data including the at least one second character string; an information processing method for converting the third text data into fourth text data by converting the at least one second character string in the third text data into a first character string having one-to-one correspondence with the second character string, The information processing method, wherein the second string includes a symbol. 　 In claim 20, The first string includes confidential information. 　 In claim 20 or 21, An information processing method, wherein the second character string is a character string composed of the symbol and at least one selected from a symbol different from the symbol, an alphabet, and a number. 　 In claim 20 or 21, An information processing method, wherein the number of characters of the second string is equal to the number of characters of the corresponding first string. 　 In claim 20 or 21, the second character string has a configuration in which a sixth character string is sandwiched between a pair of the symbols, An information processing method, wherein the sixth character string includes at least one selected from a symbol different from the symbol, an alphabet, and a number.

25. In claim 24, An information processing method, wherein the number of characters of the sixth string matches the number of characters of the corresponding first string. 　 In claim 20 or 21, creating a dictionary that defines a correspondence between a first character string and a second character string; The information processing method, wherein the fourth text data is converted from the third text data by referring to the dictionary. 　 receiving first text data and at least one first character string included in the first text data; converting the at least one first character string in the first text data into a second character string in one-to-one correspondence with the first character string, thereby converting the first text data into second text data including at least one second character string; transmitting the second text data to a language model via a network to obtain third text data including the at least one second character string; an information processing method for converting the third text data into fourth text data by converting the at least one second character string in the third text data into a first character string having one-to-one correspondence with the second character string, the second character string is a character string not included in the first text data, an information processing method in which, in converting the first text data into the second text data, for each of the at least one first string, a plurality of third strings not included in the first text data are selected, a verification process is performed for each of the plurality of third strings, and a second string that has a one-to-one correspondence with the first string is selected from among the plurality of third strings by referring to a result of the verification process, thereby converting the first string into a second string.

28. In claim 27, The first string includes confidential information.

29. In claim 28, In the verification process, preparing a plurality of sentences including a first sentence including the first character string and a plurality of second sentences obtained by converting the first character string included in the first sentence into each of the plurality of third character strings; A process of calculating a score of the natural language processing by performing a natural language processing for determining the plurality of sentences; An information processing method in which, in the process performed with reference to the results of the verification process, a third character string contained in a second sentence among the plurality of second sentences that has a score closest to that of the first sentence is determined to be a second character string that corresponds one-to-one to the first character string. 　 In any one of claims 27 to 29, An information processing method, wherein the second character string is a hypernym, an appositive word, or a hyponym of the corresponding first character string. 　 In any one of claims 27 to 29, An information processing method, wherein the second character string has a configuration in which a hypernym, an homonym, or a hyponym of the corresponding first character string is sandwiched between a pair of symbols. 　 In any one of claims 27 to 29, creating a dictionary that defines a correspondence between a first character string and a second character string; The information processing method, wherein the fourth text data is converted from the third text data by referring to the dictionary.

Images