Information processing apparatus, information processing method and information processing program

Abstract

To provide an information processing apparatus for answering in a sentence to a question in a sentence, which can generate answers which refer to the latest information while reducing time and effort to relearn by a learning model.SOLUTION: An information processing apparatus is used for answering in a sentence to a question in a sentence, and comprises: a first storage unit for storing a learning model that has learned a correspondence between a question and an answer to the question; a second storage unit for storing original information as information on which the answer is based; a registration unit for registering additional information which could be the source of the answer by a user to a third storage unit; a reception unit for receiving a question from a user; a generation unit for generating an answer to the question using the learning model with reference to the original information and the additional information; and an output unit for outputting the answer generated by the generation unit.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an information processing device, an information processing method, and an information processing program for providing a written reply to a written question. [Background technology]

[0002] In recent years, there has been remarkable progress in what is called generative AI (Artificial Intelligence), which responds to questions or requests in textual form in the same way as a human would (for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2023-76413 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, since the answers of such generative AI are based on data that the AI ​​has already learned, there is a problem that the answers cannot be based on the latest information unless the learned knowledge data of the AI ​​is updated. However, even though the learned knowledge data is updated only with the latest information, the amount of information is large and diverse, so there is a problem that updating takes time.

[0005] Therefore, the present invention has been made in consideration of the above problems, and aims to provide an information processing device, a method and a program thereof that provide a generative AI that can easily generate answers desired by users without updating learned knowledge data. [Means for solving the problem]

[0006] An information processing device according to one embodiment of the present invention is an information processing device that provides a written answer to a written question, and includes a first memory unit that stores a learning model that has learned the correspondence between questions and answers to the questions, a second memory unit that stores original information that is the information that is the basis of the answer, a registration unit that allows a user to register additional information that may be the basis of the answer in a third memory unit, a reception unit that receives questions from a user, a generation unit that generates an answer to the question using the learning model and by referring to the original information and the additional information, and an output unit that outputs the answer generated by the generation unit.

[0007] In addition, the above information processing device may include a fourth memory unit that stores user account information including at least a user identifier that identifies the user, and an authentication unit that accepts input of user information including at least the user identifier and authenticates access from the user, wherein when the user is authenticated by the authentication unit, the reception unit accepts a question from the user, the registration unit registers the additional information in association with the user identifier indicating the user, and the generation unit generates an answer to the question using a learning model by referring to the original information and the additional information stored in association with the user identifier of the user who asked the question.

[0008] The information processing device may further include a deletion unit that deletes additional information designated by a user from the additional information registered by the user, from the third storage unit.

[0009] In addition, in the above information processing device, the registration unit may accept a permission input from a user permitting other users to refer to additional information added by the user, and store the additional information in a third memory unit in association with the permission information permitting other users to refer to the additional information, and the generation unit may generate an answer to a question using a learning model by referring to the original information, the additional information registered by the user who asked the question, and the additional information associated with the permission information.

[0010] In addition, in the above-mentioned information processing device, the account information for each user is associated with rank information indicating the rank of the user, and the generation unit may generate an answer to a question using a learning model by referring to the original information, additional information registered by the user who asked the question, and additional information registered by users with a rank equal to or lower than the user corresponding to the question received by the reception unit.

[0011] In addition, an information processing method according to one embodiment of the present invention executes the following steps: a first storage step in which the computer of an information processing device that provides written answers to questions in text stores a learning model that has learned the correspondence between questions and answers to the questions; a second storage step in which original information that is the information that serves as the basis for the answers is stored; a registration step in which a user registers additional information that may serve as the basis for the answers in a third storage unit; a reception step in which questions from users are received; a generation step in which an answer to the question is generated using the learning model by referring to the original information and the additional information; and an output step in which the answer generated in the generation step is output.

[0012] In addition, an information processing program according to one embodiment of the present invention causes a computer of an information processing device that provides written answers to questions in text to have the following functions: a first memory function that stores a learning model that has learned the correspondence between questions and answers to the questions; a second memory function that stores original information that is the information that serves as the basis for the answers; a registration function that allows a user to register additional information that may serve as the basis for the answers in a third memory unit; a reception function that accepts questions from a user; a generation function that generates answers to questions using the learning model and by referring to the original information and the additional information; and an output function that outputs the answers generated by the generation function. Effect of the Invention

[0013] The information processing device according to the present disclosure appropriately stores information that the AI ​​wants to learn as additional information and refers to it when answering questions, so that the AI's knowledge can be updated without re-learning data that has already been learned. As a result, the user can obtain the latest answer that he or she wants. [Brief description of the drawings]

[0014] [Figure 1] 1 is a block diagram showing an example of the configuration of an information processing device; [Diagram 2] 11 is a flowchart illustrating an example of the operation of the information processing device. [Diagram 3] 10 is a flowchart illustrating an example of an operation of the information processing device regarding adding and deleting additional information. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] An information processing device, an information processing method, and an information processing program according to the present invention will be described in detail below with reference to the drawings.

[0016] <Embodiment Mode> Hereinafter, an information processing apparatus according to the present invention will be described in detail with reference to the drawings.

[0017] FIG. 1 is a block diagram showing an example of the configuration of an information processing device 100. As shown in FIG.

[0018] The information processing device 100 is an information processing device (computer system) that receives a textual question from a user and outputs a textual answer. The information processing device 100 may be a web server type device that receives a textual question by accessing from an external communication terminal (e.g., a smartphone, a PC, etc.) of the user through communication and outputs an answer to the communication terminal, or may be a stand-alone type device that receives a question input directly from the user and outputs an answer. A textual question is a sentence that requests an answer, and may be a colloquial sentence, or may be a sentence that requests an answer or a sentence that requests the creation of some kind of deliverable. As an example, a textual question may be a question such as "What is generative AI?" or "What is Mr. A's great achievement?", or may be a request such as "Please explain what generative AI is" or "Please tell me about Company A's performance for the past five years." In other words, a textual question may be a question in natural language. A natural language may be a language that humans use on a daily basis for communication, transmission of information, recording, thinking, etc., such as, but not limited to, Japanese, English, Chinese, and Korean.

[0019] As shown in FIG. 1, the information processing device 100 includes a communication unit 110, an input unit 120, a control unit 130, a storage unit 140, and an output unit 150.

[0020] The communication unit 110 is a communication interface having a function of communicating with an external device. The communication unit 110 may communicate with an external device by wired or wireless means, as long as the communication unit 110 can communicate with the external device. For example, the communication unit 110 may receive question information indicating a question in text form from a user terminal of a user who uses the information processing device 100, and transmit the question to the control unit 130. The communication unit 110 may also transmit answer information indicating an answer to a question instructed by the control unit 130 to the user terminal that has transmitted the question information.

[0021] The input unit 120 accepts input from a user of the information processing device 100. The input unit 120 may accept input of a question or request in the form of text from the user. The input unit 120 transmits information indicating the content of the accepted question or request to the control unit 130. The input unit 120 may be realized by a keyboard, a touch panel, or the like provided in the information processing device 100, or may be realized by a microphone for accepting voice input. When accepting voice input, text information obtained by converting the input voice into a character string using existing voice analysis software or the like is transmitted to the control unit 130.

[0022] The control unit 130 is a processor having a function of controlling each unit of the information processing device 100. The control unit 130 executes various programs stored in the storage unit 140 to fulfill the functions of the information processing device 100. The control unit 130 executes the various programs to fulfill the functions of a reception unit 131, a generation unit 132, an output unit 133, a registration unit 134, a deletion unit 135, and an authentication unit 136.

[0023] The receiving unit 131 receives a question / request in the form of text via the communication unit 110 or the input unit 120. The receiving unit 131 transmits the content of the received question / request to the generating unit 132.

[0024] The generating unit 132 generates an answer to the question / request transmitted from the receiving unit 131. The generating unit 132 generates an answer to question information indicating a question / request in a sentence transmitted from the receiving unit 131 by using the learning model 141 and referring to the original information 142. That is, the generating unit 132 analyzes the question in a sentence transmitted from the receiving unit 131 by the learning model 141 (for example, syntactic analysis or word extraction), analyzes the interrelationship and context with words and character strings included in the question, and from the result, refers to the original information 142 to identify information required to generate an answer, and generates an answer from the identified information. At this time, when there is additional information 143 that is permitted to be used by the user who has asked the question, the generating unit 132 generates an answer by referring to the original information 142 and the additional information 143. The generating unit 132 transmits the generated answer information to the output unit 133.

[0025] The output unit 133 outputs the answer information generated by the generation unit 132. The output unit 133 may output the answer information by the output unit 150 of the information processing device 100, for example, or may output the answer information by transmitting the answer information to a communication terminal of a user accessing the information processing device 100 via the communication unit 110.

[0026] The registration unit 134 registers text information (knowledge information to be referred to when generating an answer) input by a user as additional information in the storage unit 140 based on an instruction from the user via the communication unit 110 or the input unit 120. The registration unit 134 may register the additional information added by the user in association with a user identifier indicating the user.

[0027] The deletion unit 135 deletes the additional information specified by the user from the storage unit 140 based on an instruction from the user via the communication unit 110 or the input unit 120. The deleted additional information will no longer be referenced by the generation unit 132 when generating an answer to a question.

[0028] When a user accesses the information processing device 100 via the user's communication terminal, the authentication unit 136 authenticates the user. The authentication unit 136 accepts input of a user identifier indicating the user and a password associated with the user via the communication unit 110 or the input unit 120. If the input password matches the password associated with the user identifier in the user information 144, the authentication unit 136 determines that the authentication is successful, and otherwise determines that the authentication is unsuccessful. If the authentication is successful, a question from the user is accepted, and if the authentication is unsuccessful, no question from the user is accepted.

[0029] The storage unit 140 is a storage medium having a function of storing various programs and data required for the operation of the information processing device 100. The storage unit 140 can be realized by, for example, an HDD, an SSD, a flash memory, or the like. The storage unit 140 may be realized by a plurality of physically different storage media, or one storage medium may store various information by making each storage area different. The storage unit 140 may store a learning model 141 that has learned the correspondence between a question and an answer to the question, original information 142 that is referred to when generating an answer to the question, additional information 143 that is added by a user and is referred to when generating an answer to the question, and user information 144 that indicates information about each user who uses the information processing device 100. In addition, the storage unit 140 may store a program for receiving a question from a user, using the learning model 141, and generating an answer to the question by referring to the original information 142 and the additional information 143. The storage unit 140 may be realized by a cloud storage that the information processing device 100 can access.

[0030] The learning model 141 may be, for example, a learning model that learns the correspondence between question information indicating a question or request in a sentence and answer information to be answered to the question or request as teacher data. That is, the learning model 141 may be a model that learns what kind of question should be answered and how to answer it, and may be a model that estimates, generates, and outputs an answer to a question sentence by inputting a question sentence. The learning model 141 may be, for example, a machine learning natural language processing model called LLM (Large Language Model), or may be similar to these. Therefore, the learning model 141 may be a model that learns the relationship between words contained in a huge amount of natural language data, and may be a model that generates an answer by inferring other words related to each word contained in the input natural language from the word.

[0031] The original information 142 is information to be referred to when generating an answer to a question, and is a knowledge database for the information processing device 100 that stores information that is the basis of the answer. The original information 142 may store various information as text information, and may be stored not only in text but also in the form of images or sounds. The original information 142 may be an accumulation of information collected from the web. In addition, the original information 142 is not limited to dictionary definitions of information, and may include information on ideas and methods for deriving various information. The ideas and methods may be information on a calculation formula for calculating information that is the subject of the desired answer or source code of a program. Therefore, for example, in order to respond to a question such as "Please tell me the average sales and deviation for the month," the original information 142 may include information on a calculation method for the average and information on a calculation method for the deviation, and each piece of information may be associated with information on the wording "average" and information on the wording "deviation." Also, for example, the original information 142 may include knowledge of SQL grammar (knowledge of programming, method of generating program code) so as to respond to a request such as "Please tell me SQL to display a list of names of users who purchased more than 30,000 yen in August." Also, for example, the original information 142 may include information on a method of aggregating sales by customer and information on a method of graphing the aggregated numerical values ​​so as to respond to a request such as "Graph sales by customer in one month in 2022." In this way, the original information 142 includes not only definitions of words as information that will be the answer to the question asked, but also information on the ideas and methods for deriving the answer, so that the generation unit 132 can generate an answer to the accepted question. The original information 142 does not include information after the creation of the original information 142. Therefore, depending on the timing of accepting the question, an answer that does not reflect the latest information may be output. Re-learning is required to reflect the latest information, but this requires a huge amount of time depending on the amount of data in the original information 142 used for learning. Note that the original information 142 is shown for comparison with the additional information 143, but is actually included in the learning model 141.The learning model 141 can also be said to be a huge set of parameters obtained by learning using the raw information 142.

[0032] The additional information 143 is information added by a user, and is information that is additionally referred to when the user asks the information processing device 100 a question and generates an answer to the question. The additional information may be text information, an image, or a voice. In the case of voice, the additional information 143 may be converted into text data of a character string by voice analysis. The additional information 143 is stored in association with a user identifier indicating the user who added it, and is basically referred to when generating an answer to a question from the user. The additional information 143 may be referred to when generating an answer to a question from a user other than the user who added it, if the user who added it has permission to refer to the additional information 143, or if the user has higher authority than the user who added it. It is effective to input information that is not included in the original information 142 as the additional information 143, and in particular, information that has been made public or generated after the timing at which the original information 142 was created is effective. For example, if a question is "I want to know the sales for the past 30 years up to 2023 on an annual basis," and the original information only includes sales information up to 2022, the generation unit 132 can generate an answer to the question by inputting sales information for 2023 as the additional information 143. In addition, even if the original information 142 of the A company does not seem to include information about the A company in response to a question such as "Please tell me the performance of the first development department," the generation unit 132 can generate an answer to the question by adding information about the A company as the additional information 143 by a person of the A company. In addition, if the source code of a newly developed program and information explaining the contents of the program are added as the additional information 143, the generation unit 132 can also generate source code to obtain an answer to the question by imitating the added new source code. The additional information 143 is used as the original information 142 included in the learning model 141 without being re-learned. In addition, when using additional information 143, when inputting a question or request, it is possible to specify which additional information 143 to refer to, and to specify identification information indicating that information to be used when generating an answer.For example, if A's sales data for 2023 has been added as additional information 143, the additional information to be referenced may be specified in the question (request) statement as follows: "Please output A's sales data from 2021 to 2023 in tabular format compared to the same month of the previous year. Please refer to A's sales data for 2023 as additional information."

[0033] Furthermore, when receiving a question or request from a user, the generating unit 132 may extract each word included in each additional information 143, identify which additional information 143 is related to the question sentence, and identify which additional information 143 to refer to. Furthermore, when adding the additional information 143, the generating unit 132 may not only simply store the text data input by the user as it is, but may also store the text data and a value obtained by converting the text data into a numerical vector format in association with each other. This process may be realized, for example, by a registration unit. The conversion of text data into a numerical vector format (also called natural language processing (embedding)) can be vectorized by using an algorithm such as word2vec, but is not limited to this. For example, Skip-Gram, Glove, fastText, or other algorithms may be used. Then, the text data shown in the question (request) sentence is similarly vectorized, and the distance between the vector of the text data of the question (request) sentence and the vector of the additional information 143 is calculated. If the distance is within a predetermined threshold, it may be identified as the additional information 143 to be used when generating the answer.

[0034] The learning model 141, the raw information 142, and the additional information 143 are referred to by the generation unit 132. The additional information 143 is added by the registration unit 134 and deleted by the deletion unit 135.

[0035] The user information 144 is information about a user who uses the information processing device 100, and is user account information. The user information 144 may be information in which at least a user identifier (login ID) and a password of the user are associated with each other, and may further be associated with the above-mentioned additional information 143. Furthermore, personal information of each user may be registered. The user information 144 is mainly referred to by the authentication unit 136.

[0036] The output unit 150 outputs information designated according to an instruction from the control unit 130. The output unit 150 may be realized by, for example, a monitor, a speaker, or the like of the information processing device 100. The output unit 150 outputs an answer generated in response to a question or request made by the user. The output unit 150 may transmit answer information to an information processing terminal such as a smartphone held by the user, and cause the information processing terminal to display the information.

[0037] In addition, since the user's communication terminal is similar to a general smartphone, mobile phone, tablet terminal, notebook PC, PC, etc., detailed explanation will be omitted.

[0038] <Operation> Fig. 2 is a flowchart showing an example of the operation of the information processing device 100. The flowchart shown in Fig. 2 is a flowchart showing the process from accepting a question from a user to outputting an answer.

[0039] As shown in FIG. 2, the authentication unit 136 of the control unit 130 accepts access to the information processing device 100 from a user (step S201). The access from the user includes information of the user's user identifier (login ID) and password. The authentication unit 136 authenticates whether or not the pair of the accepted user identifier and password is registered in the user information 144. If registered, the authentication unit 136 authenticates that the user is a legitimate user and executes the subsequent processing, and if not registered, it authenticates that the user is an unauthorized user and ends the processing.

[0040] If the user is authenticated as a legitimate user, the receiving unit 131 receives a question in the form of text from the user (step S203). The receiving unit 131 transmits the received question to the generating unit 132.

[0041] When the question is transmitted from the receiving unit 131, the generating unit 132 determines whether there is additional information that can be referred to (step S204). That is, it determines whether the additional information 143 associated with the user identifier indicating the user who asked the question is stored in the storage unit 140.

[0042] If there is additional information that can be referred to (YES in step S204), that is, if the user has registered additional information by himself in the past, the generation unit 132 refers to the original information 142 and the additional information 143 and generates an answer to the question using the learning model 141 (step S205). On the other hand, if there is no additional information that can be referred to (NO in step S204), that is, if the user has not registered additional information by himself in the past, the generation unit 132 refers to the original information 142 and generates an answer to the question using the learning model 141 (step S206). The generation unit 132 transmits the generated answer information to the output unit 133.

[0043] When the answer information is transmitted from the generation unit 132, the output unit 133 outputs the answer information via the communication unit 110 or the output unit 150 (step S207), and ends the process. This allows the information processing device 100 to function as an interactive and generative AI that returns a textual answer to a textual question from a user.

[0044] Fig. 3 is a flowchart showing an example of the operation of registration and deletion processing of additional information in the information processing device 100. Note that the processing shown in Fig. 3 is processing in a state where the user has been authenticated as a legitimate user.

[0045] 3, the communication unit 110 or the input unit 120 of the information processing device 100 receives an input from a user. The control unit 130 determines whether the information received from the user is additional information (step S301). If the information received from the user is additional information, the registration unit 134 associates the received additional information with a user identifier indicating the user, and registers (stores) the received additional information in the storage unit 140 as additional information 143 (step S302).

[0046] On the other hand, if the received information is not additional information (NO in step S301), it is determined whether a request to delete the additional information has been received from the user (step S303). If a request to delete the additional information has not been received (NO in step S303), the process ends.

[0047] If a request to delete additional information has been accepted (YES in step S303), the additional information 143 associated with the user and designated by the user is deleted from the storage unit 140 (step S304), and the process is terminated.

[0048] This allows the information processing device 100 to generate an answer that refers to new knowledge (additional information) without updating the original information 142. In addition, the added information can be easily deleted when it becomes unnecessary.

[0049] <Summary> According to the information processing device 100 of this embodiment, the user can register additional information that can be used when the information processing device 100 generates an answer. The registration of the additional information is simply storing text information, so it does not take time compared to the process of updating the original information 142, and the process is easy because it only requires registering small pieces of information that the user wants to add from time to time. In addition, this additional information can be easily deleted even if it becomes unnecessary or if incorrect information is registered, so it does not take time and the process content is simple compared to the process of updating the original information 142. Therefore, the information processing device 100 can output a wider variety of answers to the user using the additional information easily registered by the user, so it is possible to provide the information processing device 100 that is highly convenient for the user.

[0050] <Supplementary Information> The information processing device 100 according to the above embodiment is not limited to the aspects shown in the above embodiment. Various modifications will be described below.

[0051] (1) In the above embodiment, the information processing device 100 is enabled for use when the user is authenticated, but user authentication is not essential. In this case, registration of additional information may be enabled only when the user is authenticated, or may be enabled even when the user is not authenticated. Additional information registered without authentication may be referred to by all users.

[0052] (2) In the above embodiment, the additional information is referred to only when generating an answer to a question from the user who added the additional information, but this is not limited to the above. The additional information may be registered so that it can be referenced by other users if the user who added the additional information allows it. That is, each additional information may be associated with a permission flag indicating whether or not the registered user can refer to other users. If the permission flag is information indicating permission, the additional information is referred to when generating answers to questions from all users by the generating unit 132, and if the permission flag is information indicating no permission, the additional information is referred to when generating answers only to questions from the registered user. This permission flag may be set or changed after the registration of the additional information.

[0053] In order to realize this processing, the generation unit 132 of the control unit 130 may extract additional information 143 indicating that the permission flag associated with the additional information 143 allows reference in the processing of step S204 of the flowchart shown in FIG. 2, and combine the additional information 143 with the original information 142 to generate an answer to the user's question.

[0054] Also, the additional information may be switched between being referred to and not being referred to depending on the rank of the user. That is, the user information 144 may be set with a rank (authority) for using the information processing device 100 for each user. The higher the rank, the more information may be set to be referred to, and the generating unit 132 may be configured to generate an answer by also referring to additional information registered by a user with a relatively lower rank than the high-ranked user when generating an answer to a question from a high-ranked user. Note that, when generating an answer to a question from a low-ranked user, additional information registered by a user with a relatively higher rank than the low-ranked user is not referenced.

[0055] To realize this process, the generation unit 132 of the control unit 130 may identify the rank of the user who asked the question from the user information 144 in the process of step S204 of the flowchart shown in FIG. 2, compare the rank of each user who registered the additional information 143, extract the additional information 143 of the user whose rank is lower than the rank of the user who asked the question, and combine it with the original information 142 to generate an answer to the user's question. Therefore, for example, in a certain company, when the rank is determined by the position of the user, additional information that can be referred to by ordinary employees, additional information that can be referred to only by section managers or higher, additional information that can be referred to only by department managers or higher, etc. can be set. As a result, even if the same learning model 141 is used, the information processing device 100 may generate different answers depending on the authority to refer to the additional information 143. As a result, for example, it is possible to provide the information processing device 100 that can generate an accurate answer to users who are department managers or higher, even for information that ordinary employees cannot know.

[0056] (3) In the above embodiment, it is determined whether or not additional information has been registered by the user (see step S204). However, this determination is not necessary, and the generation unit 132 may be configured to search for the original information 142 and the additional information 143 associated with the user to generate an answer.

[0057] (4) When each functional unit of the information processing device 100 is realized by software, the information processing device 100 includes a CPU that executes instructions of a program that is software that realizes each function, a ROM (Read Only Memory) or a storage device (these are referred to as "recording media") in which the program and various information are recorded so as to be readable by a computer (or CPU), and a RAM (Random Access Memory) in which the program is expanded. The object of the present invention is achieved by the computer (or CPU) reading and executing the program from the recording medium. That is, the information processing device 100 according to the present invention functions as each of the above-mentioned components by the CPU executing the program loaded on the RAM. As the recording medium, a "non-transient tangible medium", for example, a semiconductor memory, a programmable logic circuit, etc. can be used. The program may also be supplied to the computer via any transmission medium (such as a communication network or a broadcast wave) capable of transmitting the program. The present invention may also be realized in the form of an information signal embedded in a carrier wave in which the program is embodied by electronic transmission.

[0058] The above program may be implemented using, for example, a scripting language such as ActionScript, JavaScript (registered trademark), Python, or Ruby; an object-oriented programming language such as C language, C++, C#, Objective-C, Swift, or Java (registered trademark); or a markup language such as HTML5. Furthermore, the term "section, module, or unit" in the claims may be read as "means" or "circuit." For example, a communication unit may be read as a communication means or a communication circuit. Furthermore, the virtual reality space and the program for information processing using the virtual reality space may be created using a Unity game engine.

[0059] Furthermore, the program of the present disclosure may be provided to the information processing device 100 via any transmission medium (such as a communication network or broadcast waves) capable of transmitting the program.

[0060] (5) The order and timing of each process described in the above embodiment may be changed as appropriate, provided that the same effects as those described in the above embodiment can be achieved. [Explanation of symbols]

[0061] 100 Information processing device 110 Communications Department 120 Input section 130 Control section 131 Reception 132 Generation part 133 Output section 134 Registration Department 135 Deleted section 136 Authentication Department 140 Storage section 150 Output section

Claims

1. An information processing device that responds to a question in a sentence, a first storage unit that stores a learning model that has learned a correspondence relationship between a question and an answer to the question; A second storage unit that stores raw information that is information on which the answer is based; a registration unit for registering additional information that may be a basis for the answer by the user in a third storage unit; A reception unit that receives questions from users; a generation unit that generates an answer to the question by using the learning model and by referring to the original information and the additional information; an output unit that outputs the answer generated by the generation unit; An information processing device comprising:

2. a fourth storage unit configured to store user account information including at least a user identifier for identifying the user; an authentication unit that accepts an input of user information including at least a user identifier and authenticates access from the user; the reception unit receives a question from the user when the authentication unit has authenticated the user, The registration unit registers the additional information in association with a user identifier indicating the user; The generation unit generates an answer to the question by using the learning model and by referring to the original information and additional information stored in association with a user identifier of the user who asked the question.

2. The information processing apparatus according to claim 1,

3. The information processing device further includes a deletion unit that deletes additional information designated by a user from the additional information registered by the user from the third storage unit.

3. The information processing apparatus according to claim 2.

4. the registration unit receives a permission input from a user permitting other users to refer to additional information added by the user, and stores the additional information in the third storage unit in association with permission information permitting other users to refer to the additional information; The generation unit generates an answer to the question by using the learning model and by referring to the original information, additional information registered by the user who asked the question, and additional information associated with permission information.

3. The information processing apparatus according to claim 2.

5. rank information indicating a rank of each of the users is associated with the account information, The generation unit generates an answer to the question by using the learning model and by referring to the original information, additional information registered by the user who asked the question, and additional information registered by users who are at a rank equal to or lower than the rank of the user corresponding to the question received by the reception unit.

3. The information processing apparatus according to claim 2.

6. A computer of an information processing device that responds to questions in text by providing text responses, a first storage step of storing a learning model that has learned a correspondence relationship between a question and an answer to the question; A second storage step of storing raw information that is information on which the answer is based; a registration step in which the user registers additional information that may be a basis for the answer in a third storage unit; a reception step for receiving a question from a user; a generating step of generating an answer to the question by using the learning model and by referring to the original information and the additional information; an output step of outputting the answer generated in the generating step; An information processing method for performing the above.

7. A computer that is an information processing device that responds to questions in text by writing a response in text, a first storage function for storing a learning model that has learned a correspondence relationship between a question and an answer to the question; A second storage function for storing original information that is information on which the answer is based; A registration function for allowing a user to register additional information that may be the basis of the answer in a third storage unit; A reception function for receiving questions from users; a generation function of generating an answer to the question by using the learning model and by referring to the original information and the additional information; an output function for outputting the answer generated by the generation function; An information processing program that realizes this.

Images