Information processing apparatus, information processing method, and information processing program

Abstract

To provide only data with assured reliability, as learning data of a generative model.SOLUTION: An information processing server 10 includes: a collection unit 131 which collects data; a determination unit 133 which determines reliability of the data collected by the collection unit 131, by using a determination model 134; and a transmission control unit 135 which transmits data which has been determined by the determination unit 133 to have predetermined reliability, as learning data of a generative model which generates data of a predetermined field.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to an information processing device, an information processing method, and an information processing program. [Background technology]

[0002] Generative AI (Artificial Intelligence), which uses trained machine learning models to generate text, images, audio, and video, is attracting attention (for example, Patent Documents 1 and 2).

[0003] For example, there is a large-scale language model (LLM), a natural language processing model trained using large amounts of text data, as a text generation AI. In particular, OpenAI's Generative Pre-trained Transformer (GPT) (chatGPT) has attracted attention and is gaining a large number of users. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 7133689 [Patent Document 2] Patent No. 7208314 Summary of the Invention [Problem to be solved by the invention]

[0005] Here, various data are used as training data for generative AI. For example, chatGPT uses a large amount of text data publicly available on the Internet as training data. Specifically, chatGPT uses various types of text data such as web pages, books, news articles, magazines, papers, Wikipedia, etc.

[0006] However, it is unclear whether the content of the vast amount of text data available on the internet is truly accurate. In other words, it is unclear whether the content of the data used as training data for generative AI is accurate.

[0007] Furthermore, data generated by generative AI (generative models) that have been trained using data whose authenticity is unknown may not be accurate.

[0008] The present invention has been made in consideration of the above, and aims to provide an information processing device, an information processing method, and an information processing program that can provide only data with guaranteed reliability as training data for a generative model. [Means for solving the problem]

[0009] In order to solve the above-mentioned problems and achieve the objectives, the information processing device of the present invention is characterized by having a collection unit that collects data, a judgment unit that uses a judgment model to judge the reliability of the data collected by the collection unit, and a transmission control unit that transmits data that is judged by the judgment unit to have a predetermined reliability as learning data for a generation model that generates data in a predetermined field. [Effects of the Invention]

[0010] According to the present invention, only data whose reliability is guaranteed can be provided as training data for a generative model. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a block diagram illustrating an example of a configuration of an information processing system according to an embodiment. [Figure 2] FIG. 2 is a block diagram illustrating an example of the configuration of the information processing server illustrated in FIG. [Figure 3] FIG. 3 is a diagram illustrating an example of the data configuration of the generative model field data illustrated in FIG. [Figure 4]FIG. 4 is a diagram illustrating the learning of the determination model shown in FIG. [Figure 5] FIG. 5 is a diagram illustrating an outline of the processing of the information processing system shown in FIG. [Figure 6] FIG. 6 is a sequence diagram showing the processing procedure of the information processing method according to the embodiment. [Figure 7] FIG. 7 is a block diagram showing an example of a configuration of an information processing system according to a modification of the embodiment. [Figure 8] FIG. 8 is a block diagram illustrating an example of the configuration of the information processing server illustrated in FIG. [Figure 9] FIG. 9 is a diagram illustrating an example of the data configuration of the generation model field data illustrated in FIG. [Figure 10] FIG. 10 is a sequence diagram showing a processing procedure of an information processing method according to a modified example of the embodiment. [Figure 11] FIG. 11 is a diagram illustrating an example of a computer that implements an information processing server by executing a program. DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. Note that the present invention is not limited to this embodiment. In addition, in the description of the drawings, the same parts are designated by the same reference numerals.

[0013] [Embodiment Mode] In the embodiment, a determination model is used to determine the reliability of various collected data, and only data determined to have a predetermined reliability, i.e., data with the highest reliability and guaranteed reliability, is transmitted as training data for a generative model that generates information in a predetermined field. As a result, in the embodiment, only data with guaranteed reliability is provided as training data for the generative model (generative AI).

[0014] [Information Processing Systems] FIG. 1 is a block diagram illustrating an example of a configuration of an information processing system according to an embodiment.

[0015] As shown in FIG. 1, an information processing system 100 according to the embodiment includes an information processing server 10 and a generating device 20 having a generative model 21 that generates data.

[0016] The information processing server 10 communicates with external servers 30-1 and 30-2 via a network 40. When the external servers are collectively referred to, they will be referred to as external servers 30.

[0017] The information processing server 10 collects data from the external server 30 and evaluates the reliability of the collected data. The information processing server 10 transmits only data that is determined to have a predetermined reliability, or data that is the most reliable and whose reliability is guaranteed, to the generation device 20 as training data for a generation model 21 (described later) of the generation device 20.

[0018] The information processing server 10 classifies the fields of the collected data and determines the reliability of only the data in a predetermined field corresponding to the generative model 21. The information processing server 10 transmits to the generating device 20 only the information in the predetermined field corresponding to the generative model 21, whose reliability is guaranteed.

[0019] The generating device 20 generates data using a generative model 21. The generative model 21 is a machine learning model that generates data in a predetermined field. In other words, the generative model 21 is a generative model specialized in a predetermined field. The predetermined field is medicine, law, finance, smart cities, factory management, agriculture, or chemistry.

[0020] The generative model 21 is, for example, a machine learning model that generates text data and is a natural language model that has been trained in advance using a large amount of text data. Subsequently, the generative model 21 is fine-tuned so as to perform a predetermined task using only reliable data in a predetermined field transmitted from the information processing server 10 as training data. Alternatively, the generative model 21 may be a machine learning model that generates image data.

[0021] [Information processing server] Next, a description will be given of the information processing server 10 shown in Fig. 2. Fig. 2 is a block diagram showing an example of the configuration of the information processing server 10 shown in Fig. 1.

[0022] 2, the information processing server 10 includes, for example, a communication unit 11, a storage unit 12, and a control unit 13. Note that the information processing server 10 may be connected to input devices such as a mouse and a keyboard, and output devices such as a display and a speaker.

[0023] The communication unit 11 controls communication related to various types of information. For example, the communication unit 11 controls communication with the external server 30 and communication with the generation device 20. The communication unit 11 receives various types of information from the external server 30 and outputs the information to the control unit 13. The communication unit 11 communicates with the generation device 20 and transmits reliable data to the generation device 20.

[0024] The storage unit 12 stores data and programs necessary for various processes by the control unit 13. For example, the storage unit 12 may be a semiconductor memory element such as a random access memory (RAM) or a flash memory, or a storage device such as a hard disk or an optical disk. For example, the storage unit 12 has generative model field data 121, collected data 122, and reliability-guaranteed data 123.

[0025] The generative model field data 121 is information indicating the field corresponding to the generative model 21 of the generating device 20 to which the information processing server 10 provides data. Fig. 3 is a diagram showing an example of the data configuration of the generative model field data 121 shown in Fig. 2.

[0026] 3, the generative model field data 121 includes, for example, items such as identification information of the generative model of the generating device 20, the field that the generating device corresponds to, and the reliability of the data provided to the generative model. Information on these items is registered, for example, during communication with the generating device 20 that provides the data. For example, it is registered that the generative model 21 corresponds to the medical field and transmits the most reliable data.

[0027] The collected data 122 is various data collected by the information processing server 10 from the external server 30.

[0028] The reliability-guaranteed data 123 is data collected by the information processing server 10, the reliability of which is guaranteed and which is to be transmitted to the generative model 21.

[0029] The control unit 13 has an internal memory for storing programs that define various processing procedures and necessary data, and executes various processes using these. Here, the control unit 13 may be, for example, an electronic circuit such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit), or an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0030] The control unit 13 includes a collection unit 131 , a category classification unit 132 , a determination unit 133 , and a transmission control unit 135 .

[0031] The collection unit 131 collects various data from the external server 30 via the communication unit 11. The collection unit 131 collects information published on the network. The data collected by the collection unit 131 is stored in the storage unit 12 as collected data 122.

[0032] The field classification unit 132 is provided before the determination unit 133. The field classification unit 132 classifies the fields of the data collected by the collection unit 131 and outputs only data in a predetermined field from the classified data to the determination unit 133. When the generative model 21 corresponds to the medical field, the field classification unit 132 outputs only data in the medical field from the classified data to the determination unit 133. The field classification unit 132 classifies the field to which the data belongs based on the title, frequently used terms, drawings, the industry of the publisher, etc. The field classification unit 132 may classify the data using a trained classifier. Furthermore, if the determination model 134 described below can determine whether the data is in the medical field when determining reliability, the field classification unit 132 may be omitted.

[0033] The determination unit 133 uses the determination model 134 to determine the reliability of the data collected by the collection unit 131. The determination unit 133 associates only the data that is most reliable (highest reliability) and can be transmitted to the generation model 21 with the determination result that the data is most reliable, and stores the data as reliability-guaranteed data 123 in the storage unit 12.

[0034] Here, the determination unit 133 may provide a pre-processing summary model before the determination model 134. For example, the summary model is a natural language processing model (e.g., a large-scale language model (LLM)) trained using a large amount of text data so as to output a summary of an input sentence. The determination unit 133 may generate a summary of each piece of data collected by the collection unit 131 using the summary model, and input the summary generated by the summary model to the determination model 134.

[0035] The determination model 134 is a machine learning model that, when data is input, outputs the reliability of the input data. In this case, a prompt is set in the determination model 134 to instruct the task of determining whether the reliability of the input health level is highest, high, standard, or low, and outputting the result in a predetermined format.

[0036] The determination model 134 is a model obtained by fine-tuning a natural language processing model (for example, a large-scale language model (LLM)) trained using a large amount of text data, using training data T1 described below. The determination model 134 is a machine learning model in which machine learning has been performed using peer-reviewed papers in a predetermined field, academic books in a predetermined field, the contents of national examinations in a predetermined field, and publications from administrative agencies that have jurisdiction over a predetermined field.

[0037] Fig. 4 is a diagram illustrating the learning of the determination model 134 shown in Fig. 2. When the field of the generation model 21 is medicine (see Fig. 3), as shown in Fig. 4, machine learning is performed on the determination model 134 using peer-reviewed papers T11 in the medical field, academic books T12 in the medical field, publications T13 from administrative agencies with jurisdiction over the medical field, and the contents of national medical examinations as training data T1.

[0038] When a peer-reviewed paper in the medical field T11, an academic book in the medical field T12, a publication from an administrative agency that has jurisdiction over the medical field T13, or the contents of a national medical examination are input into the judgment model 134, the parameter update unit 50 updates the parameters of the judgment model 134 so as to output the highest reliability.

[0039] The determination model 134 compares the words and contexts contained in, for example, peer-reviewed medical papers T11, academic books T12, publications T13 from administrative agencies with jurisdiction over the medical field, and the contents of national medical examinations with the words and contexts contained in the input data, thereby determining the reliability of the input data as, for example, highest, high, standard, or low. Specifically, the determination model 134 calculates the similarity between the words and contexts contained in the training data T1 and the words and contexts contained in the input data based on the feature vectors of the training data T1 and the feature vectors of the input data, and determines the reliability of the input data based on the calculated similarity. For example, a range of similarity is set corresponding to four levels of data reliability, for example, highest, high, standard, or low.

[0040] The parameter update unit 50 may be provided in the information processing server 10, or may be provided in a device different from the information processing server 10. The determination model 134 whose parameters have been optimized by the parameter update unit 50 is applied to the determination unit 133 of the information processing server 10.

[0041] The transmission control unit 135 transmits data determined by the determination unit 133 to have a predetermined reliability as training data for the generative model 21 that generates data in a predetermined field. When the field of the generative model 21 is medical (see FIG. 3), only data determined by the determination unit 133 to have the highest reliability is transmitted as training data for the generative model 21 to, for example, the generation device 20.

[0042] [Processing Overview] Fig. 5 is a diagram showing an outline of the processing of the information processing system 100 shown in Fig. 1. In the example of Fig. 5, a case will be described in which the field of the generative model 21 is medicine.

[0043] In the information processing server 10, the field classification unit 132 classifies the fields of the collected data, and outputs only data in the medical field, which is the field of the generative model 21, to the determination unit 133.

[0044] The determination model 134 of the determination unit 133 determines the reliability of the input medical field data and outputs the most reliable medical field data to the transmission control unit 135. The transmission control unit 135 transmits the most reliable medical field data as training data for the generation model 21.

[0045] The parameter update unit 60 uses the most reliable data in the medical field as training data to train the generative model 21 and update the parameters of the generative model 21. The parameter update unit 60 may be provided within the generating device 20, or may be provided in a device different from the generating device 20. The generative model 21, whose parameters have been optimized by the parameter update unit 60, is applied to the generating device 20.

[0046] [Processing Procedure] FIG. 6 is a sequence diagram showing the processing procedure of the information processing method according to the embodiment.

[0047] The information processing server 10 communicates with the external server 30 and collects data (step S1). The information processing server 10 classifies the collected data into categories and outputs only data in a predetermined category to the determination unit 133 (step S2). In the information processing server 10, the determination unit 133 uses the determination model 134 to determine the reliability of each piece of data collected by the collection unit 131 (step S3).

[0048] In the information processing server 10, the transmission control unit 135 transmits the data that the judgment unit 133 has determined to have a predetermined reliability, to, for example, the generation device 20, as learning data for a generation model that generates data in a predetermined field (step S4).

[0049] In the generating device 20, the generating model 21 is trained so as to perform a predetermined task using only data in a predetermined field that is transmitted from the information processing server 10 and whose reliability is guaranteed as training data (step S5).

[0050] [Effects of the embodiment] The information processing server 10 according to the embodiment collects data and determines the reliability of the collected data using the determination model 134. Then, the information processing server 10 transmits the data determined to have a predetermined reliability as training data for the generation model 21 that generates data in a predetermined field.

[0051] In this way, the information processing server 10 can provide only data whose reliability is guaranteed as training data for the generative model 21. This allows the generative model 21 to perform training using only data whose reliability is guaranteed as training data. As a result, the reliability of data in a predetermined field generated by the generative model 21 can also be guaranteed.

[0052] The determination model 134 is a machine learning model that has been trained using peer-reviewed papers in a predetermined field, academic books in a predetermined field, the contents of national examinations in a predetermined field, and publications from administrative agencies that have jurisdiction over a predetermined field. By using this determination model 134, the information processing server 10 can appropriately evaluate the reliability of data.

[0053] In addition, the information processing server 10 further provides a classification unit in front of the judgment unit 133 that classifies the fields of the collected data so that only data in a specified field can be output to the judgment unit 133, thereby improving the accuracy of the reliability evaluation by the judgment unit 133.

[0054] [Variations] Next, a modified example of the embodiment will be described below: Fig. 7 is a block diagram showing an example of the configuration of an information processing system according to the modified example of the embodiment.

[0055] As shown in Fig. 7, an information processing system 200 includes an information processing server 210 instead of the information processing server 10 of Fig. 1. The information processing system 200 also includes generation devices 20A, 20B, and 20C, each having first to third generation models 21A, 21B, and 21C that generate data. The first to third generation models 21A, 21B, and 21C generate data specialized for, for example, the fields of medicine, law, finance, smart cities, factory management, agriculture, or chemistry. The number of generation devices 20A, 20B, and 20C is just an example, and may be one or two, or may be four or more.

[0056] The information processing server 210 provides a determination model for each field of data generated by the first to third generation models 21A, 21B, and 21C, i.e., medical, legal, financial, smart city, factory management, agriculture, or chemistry. The information processing server 210 then determines the reliability of the data sorted for each field using the corresponding determination model.

[0057] The information processing server 210 transmits data determined to have a predetermined reliability (first to third reliability-guaranteed data) to generation devices 20A, 20B, and 20C having first to third generation models 21A, 21B, and 21C corresponding to the field to which the data belongs.

[0058] Fig. 8 is a block diagram showing an example of the configuration of the information processing server 210 shown in Fig. 7. As shown in Fig. 8, the information processing server 210 has a storage unit 212 and a control unit 213 instead of the storage unit 12 and the control unit 13 shown in Fig. 2.

[0059] The storage unit 212 has the same functions as the storage unit 12. The storage unit 212 has generative model field data 2121, collected data 122, field-classified data 2123, and reliability-guaranteed data 2124.

[0060] The generative model field data 2121 is information indicating the fields to which the first to third generative models 21A, 21B, and 21C of the generating devices 20A, 20B, and 20C, to which the information processing server 210 provides data, correspond.

[0061] Fig. 9 is a diagram showing an example of the data configuration of the generative model field data 2121 shown in Fig. 8. As shown in Fig. 9, the generative model field data 2121 includes items such as identification information of the generative model, the field supported by the generating device, and the reliability of the data provided to the generative model. Information on these items is registered, for example, during communication with the generating devices 20A, 20B, and 20C that provide the data.

[0062] For example, the first generative model 21A corresponds to the medical field and is registered to transmit the most reliable data. The second generative model 21B corresponds to the legal field and is registered to transmit the most reliable data. The third generative model 21C corresponds to the smart city field and is registered to transmit the most reliable data. Furthermore, the generative model 21D (not shown) corresponds to the factory management field and is registered to transmit the most reliable data. For the generative model 12E (not shown) and subsequent models, the corresponding field and the reliability of the data to be transmitted are also registered.

[0063] The field-classified data 2123 is data collected by the information processing server 210, and is data associated with information indicating the field classified by a field classification unit 2132 (described later).

[0064] The reliability-guaranteed data 2124 is data collected by the information processing server 210, whose reliability is guaranteed and which is sent to the first to third generation models 21A, 21B, and 21C, and is data associated with information indicating the classified field or identification information of the destination generation devices 20A, 20B, and 20C.

[0065] The control unit 213 has the same functions as the control unit 13. The control unit 213 has a collection unit 131, a category classification unit 2132, a determination unit 2133, and a transmission control unit 2135.

[0066] The category classification unit 2132 classifies the category of the data collected by the collection unit 131. The category classification unit 2132 associates the data collected by the collection unit 131 with information indicating the classified category and stores the data in the storage unit 212.

[0067] The determination unit 2133 determines the reliability of each data item using a determination model provided for each field, such as medicine, law, finance, smart city, factory management, agriculture, or chemistry. The determination unit 2133 includes a sorting unit 2136, a first determination model 2134-1, a second determination model 2134-2, and a third determination model 2134-3. The first determination model 2134-1 corresponds to the medicine field, the second determination model 2134-2 corresponds to the law field, and the third determination model 2134-3 corresponds to the smart city field. The determination unit 2133 includes determination models (not shown) specialized for the finance, factory management, agriculture, or chemistry fields.

[0068] The allocating unit 2136 allocates the data collected by the collecting unit 131 to each determination model (first determination model 2134-1, second determination model 2134-2, third determination model 2134-3) according to the field classified by the field classifying unit 2132.

[0069] The allocating unit 2136 inputs only data classified into the medical field by the field classification unit 2132 into a first determination model 2134-1 corresponding to the medical field. The allocating unit 2136 inputs only data classified into the legal field by the field classification unit 2132 into a second determination model 2134-2 corresponding to the legal field. The allocating unit 2136 inputs only data classified into the smart city field by the field classification unit 2132 into a third determination model 2134-3 corresponding to the smart city field.

[0070] The first determination model 2134-1, the second determination model 2134-2, and the third determination model 2134-3 determine the reliability of the data allocated by the allocation unit 2136 as, for example, the highest, high, standard, or low. The number of determination models is not limited to three, and may be one, two, or four or more, and is set according to the number of fields to which the generation model of the data destination corresponds.

[0071] The first judgment model 2134-1, the second judgment model 2134-2, and the third judgment model 2134-3 are machine learning models that, when data is input, output the reliability of the input data. The first judgment model 2134-1, the second judgment model 2134-2, and the third judgment model 2134-3 are machine learning models in which machine learning has been performed using the corresponding peer-reviewed papers in a predetermined field, academic books in a predetermined field, the contents of national examinations in a predetermined field, and publications from administrative agencies that have jurisdiction over a predetermined field.

[0072] The first determination model 2134-1 corresponds to, for example, the medical field of the first generation model 21A, and machine learning is performed using peer-reviewed papers in the medical field, academic books in the medical field, publications from administrative agencies with jurisdiction over the medical field, and the contents of national medical examinations as training data. When peer-reviewed papers in the medical field, academic books in the medical field, publications from administrative agencies with jurisdiction over the medical field, and the contents of national medical examinations are input to the first determination model 2134-1, the parameters of the first determination model 2134-1 are updated so as to output the highest reliability.

[0073] Furthermore, second determination model 2134-2 corresponds to, for example, the legal field of second generation model 21B, and machine learning is performed using peer-reviewed papers in the legal field, academic books in the legal field, publications from administrative agencies with jurisdiction over the legal field, and the contents of national legal examinations as training data. When peer-reviewed papers in the legal field, academic books in the legal field, publications from administrative agencies with jurisdiction over the legal field, and the contents of national legal examinations are input to second determination model 2134-2, parameters are updated so as to output the highest reliability.

[0074] The third determination model 2134-3 corresponds to, for example, the smart city field of the third generation model 21C, and performs machine learning using peer-reviewed papers in the smart city field, academic books in the smart city field, publications from administrative agencies with jurisdiction over the smart city field, and the contents of qualification exams required for engineers building smart cities as training data. When peer-reviewed papers in the smart city field, academic books in the smart city field, publications from administrative agencies with jurisdiction over the smart city field, and qualification exams required for engineers building smart cities are input to the second determination model 2134-2, the parameters of the third determination model 2134-3 are updated so as to output the highest reliability.

[0075] The determination unit 2133 associates the determination result that the data is the most reliable (highest) and can be transmitted to the first to third generation models 21A, 21B, 21C with information indicating the classified field or identification information of the destination generation devices 20A, 20B, 20C, and stores the data as reliability-guaranteed data 2124 in the storage unit 12 only for the data that is the most reliable among the data. The information indicating the field corresponds to the field corresponding to the determination model that is the output source of the reliability determination result. The identification information of the destination generation device is the identification information of the generation device that is the field corresponding to the determination model that is the output source of the reliability determination result.

[0076] The transmission control unit 2135 transmits the data that has been determined by the determination unit 2133 to have a predetermined reliability as training data for the first to third generation models 21A, 21B, and 21C.

[0077] The transmission control unit 2135 transmits, from the reliability guaranteed data 2124, information classified into the medical field or data whose destination generation device is the generation device 20A, as first reliability guaranteed data to the generation device 20A.

[0078] Furthermore, among the reliability-guaranteed data 2124, information classified into the field of law or data whose destination generating device is the generating device 20B is transmitted to the generating device 20B as second reliability-guaranteed data.

[0079] Furthermore, among the reliability-guaranteed data 2124, information whose classified field is smart city or data whose destination generation device is the generation device 20C is transmitted to the generation device 20C as second reliability-guaranteed data.

[0080] [Processing Procedure] FIG. 10 is a sequence diagram showing a processing procedure of an information processing method according to a modified example of the embodiment.

[0081] The information processing server 210 communicates with the external server 30 and collects data (step S11). In the information processing server 210, the field classification unit 2132 classifies the fields of the collected data (step S12). In the information processing server 210, the allocation unit 2136 allocates the collected data to each determination model according to the field classified by the field classification unit 2132 (step S13).

[0082] The information processing server 210 determines the reliability of the assigned data using each determination model (step S14).

[0083] Then, the information processing server 210 transmits the medical field data that has been determined to have a predetermined reliability (highest) to the generating device as training data for the generation model corresponding to the field in which this data has been classified.

[0084] The transmission control unit 2135 transmits the data that has been determined by the determination unit 2133 to have a predetermined reliability as training data for the first to third generation models 21A, 21B, and 21C.

[0085] The transmission control unit 2135 transmits the first reliability-guaranteed data of the reliability-guaranteed data 2124 to the generating device 20A (step S15). The transmission control unit 2135 transmits the second reliability-guaranteed data of the reliability-guaranteed data 2124 to the generating device 20B (step S17). The transmission control unit 2135 transmits the third reliability-guaranteed data of the reliability-guaranteed data 2124 to the generating device 20C (step S19).

[0086] In the generation devices 20A, 20B, and 20C, the first to third generation models 21A, 21B, and 21C are trained so as to perform a predetermined task using only the first to third reliability-guaranteed data transmitted from the information processing server 10, respectively, as training data (steps S16, S18, and S20).

[0087] In this way, the information processing server 210 classifies the fields of data and uses each determination model to determine the reliability of each field, thereby being able to provide the most reliable data as training data for each generation model corresponding to multiple fields.

[0088] [System configuration of the embodiment] The information processing servers 10 and 210 are conceptual functional entities and do not necessarily have to be physically configured as shown in the drawings. In other words, the specific form of distribution and integration of the functions of the information processing servers 10 and 210 is not limited to that shown in the drawings, and all or part of them can be functionally or physically distributed or integrated in any unit depending on various loads, usage conditions, etc.

[0089] Furthermore, all or any part of the processes performed by the information processing servers 10 and 210 may be realized by a CPU, a GPU (Graphics Processing Unit), and a program analyzed and executed by the CPU and the GPU. Furthermore, each process performed by the information processing servers 10 and 210 may be realized as hardware using wired logic.

[0090] Furthermore, among the processes described in the embodiments, all or part of the processes described as being performed automatically can be performed manually. Alternatively, all or part of the processes described as being performed manually can be performed automatically using a known method. In addition, the processing procedures, control procedures, specific names, and information including various data and parameters described above and illustrated can be changed as appropriate unless otherwise specified.

[0091] [program] 11 is a diagram showing an example of a computer in which the information processing server 10, 210 is realized by executing a program. The computer 1000 has, for example, a memory 1010 and a CPU 1020. The computer 1000 also has a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. These components are connected by a bus 1080.

[0092] The memory 1010 includes a ROM 1011 and a RAM 1012. The ROM 1011 stores a boot program such as a BIOS (Basic Input Output System). The hard disk drive interface 1030 is connected to a hard disk drive 1090. The disk drive interface 1040 is connected to a disk drive 1100. A removable storage medium such as a magnetic disk or optical disk is inserted into the disk drive 1100. The serial port interface 1050 is connected to a mouse 1110 and a keyboard 1120, for example. The video adapter 1060 is connected to a display 1130, for example.

[0093] The hard disk drive 1090 stores, for example, an OS (Operating System) 1091, an application program 1092, a program module 1093, and program data 1094. That is, a program that defines each process of the information processing server 10, 210 is implemented as a program module 1093 in which code executable by the computer 1000 is written. The program module 1093 is stored, for example, in the hard disk drive 1090. For example, a program module 1093 for executing the same process as the functional configuration of the information processing server 10, 210 is stored in the hard disk drive 1090. Note that the hard disk drive 1090 may be replaced with an SSD (Solid State Drive).

[0094] Furthermore, setting data used in the processing of the above-described embodiment is stored as program data 1094, for example, in memory 1010 or hard disk drive 1090. Then, CPU 1020 reads program module 1093 and program data 1094 stored in memory 1010 or hard disk drive 1090 into RAM 1012 as necessary and executes them.

[0095] The program module 1093 and program data 1094 are not limited to being stored in the hard disk drive 1090, but may also be stored in, for example, a removable storage medium and read by the CPU 1020 via the disk drive 1100 or the like. Alternatively, the program module 1093 and program data 1094 may be stored in another computer connected via a network (such as a local area network (LAN) or a wide area network (WAN)). The program module 1093 and program data 1094 may then be read by the CPU 1020 from the other computer via the network interface 1070.

[0096] Although the present invention has been described above as an embodiment, the present invention is not limited to the descriptions and drawings that form part of the disclosure of the present invention. In other words, other embodiments, examples, and operational techniques that can be made by those skilled in the art based on the present invention are all included in the scope of the present invention. [Explanation of symbols]

[0097] 10,210 information processing servers 11 Communications Department 12,212 storage unit 13,213 Control Unit 20,20A~20C generator 21 Generative Model 30, 30-1, 30-2 External Server 100,200 Information Processing Systems 121,2121 Generative Model Field Data 122 Collected Data 123,2124 Reliable data 131 Collection Department 132,2132 Field Classification Department 133,2133 Judgment section 134 Decision Model 135,2135 Transmission control section 2123 Category-Classified Data 2136 Distribution Department

Claims

1. a collection unit that collects data; a determination unit that determines the reliability of the data collected by the collection unit using a determination model; a transmission control unit that transmits the data that has been determined to have a predetermined reliability by the determination unit as training data for a generative model that generates data in a predetermined field; An information processing device comprising:

2. The information processing device according to claim 1, characterized in that the judgment model is a machine learning model that, when data is input, outputs the reliability of the input data, and is a machine learning model in which machine learning is performed using peer-reviewed papers in the specified field, academic books in the specified field, the contents of national examinations in the specified field, and publications of administrative agencies that have jurisdiction over the specified field.

3. The information processing device according to claim 1 , wherein the predetermined field is medicine, law, finance, smart cities, factory management, agriculture, or chemistry.

4. 2. The information processing apparatus according to claim 1, further comprising a classification unit, provided before the determination unit, for classifying the fields of the data collected by the collection unit and outputting only the data in the predetermined field to the determination unit.

5. The judgment models are provided according to the fields of medicine, law, finance, smart cities, factory management, agriculture, or chemistry, a classification unit disposed before the determination unit, which classifies the category of the data collected by the collection unit; 2. The information processing apparatus according to claim 1, wherein the determination unit includes a distribution unit that distributes the data collected by the collection unit to each determination model according to the field classified by the classification unit.

6. An information processing method executed by an information processing device, collecting data; determining the reliability of the data collected in the collecting step using a decision model; a step of transmitting the data determined to have a predetermined reliability in the determining step as training data for a generative model that generates data in a predetermined field; An information processing method comprising:

7. collecting data; determining the reliability of the data collected in the collecting step using a decision model; a step of transmitting the data determined to have a predetermined reliability in the determining step as training data for a generative model that generates data in a predetermined field; An information processing program that causes a computer to execute the above.

Images