Proposal system, proposal method, and computer program

Abstract

To promote utilization of personal data while ensuring protection of individual information, this proposition system comprises a graph generation unit, a reception unit, and a proposition unit. The graph generation unit uses element identification information for identifying personal data elements, which are data elements constituting personal data, and data location information relating to the personal data elements and representing information holders holding the personal data elements to generate a knowledge graph representing a relationship between a plurality of nodes including a plurality of the information holders. The reception unit receives an information request for requesting information of an information holder represented by data location information relating to personal data elements satisfying user needs. The proposition unit uses a model generated by machine learning to output a recommended information holding means that is an information holder satisfying requirements corresponding to the received information request. The model is generated by machine learning using the generated knowledge graph and information holder requirements corresponding to information requests.

Description

Proposal system, proposal method, and program storage medium

[0001] The present invention relates to a technology for utilizing personal data.

[0002] With the development of information and communication technology, a wide variety of data groups containing large amounts of data are being generated every day. Such data groups are called big data. Big data is increasingly being used in business, such as by collecting, storing, and analyzing big data to gain insights.

[0003] Big data can be broadly categorized into open data, personal data, and industrial data, focusing on data that can be generated by three entities: individuals, companies, and countries. Open data is data provided by national and local governments. Personal data is data related to individuals, including not only personal data that can identify individuals, but also personal attribute information, history of movements, actions, purchases, and other data, as well as information collected from wearable devices. Industrial data is data other than personal data, including data related to human "knowledge" such as various know-how, and data generated during communication between industrial machines known as M2M (Machine to Machine).

[0004] Patent Document 1 (JP 2022-113621 A) discloses a technology for supporting the coordination of transferring elderly patients or other patients from the facility where they are hospitalized to another facility such as a nursing home. In the configuration of Patent Document 1, information regarding whether or not the patient can be accepted is registered in the system as facility information, and facilities that meet the patient's requirements are extracted and presented using this facility information.

[0005] Japanese Patent Application Laid-Open No. 2022-113621

[0006] Data utilization businesses that want to utilize data often search for sources of data they want to use as reference when launching new projects, for example. Furthermore, the data they want to use as reference may not be personal data that can identify individuals, but may be associated with personal data. In other words, the data they want to use as reference may be personal data. Businesses that hold personal data may not disclose the status of their holdings of personal data from the perspective of protecting personal information. For this reason, data utilization businesses may find it difficult to find a business that holds the data they want to use as reference (personal data) in an amount that is effective for analysis (i.e., an appropriate business to access). Due to such cases and concerns about personal information protection, the utilization of personal data is slow to progress.

[0007] The present invention has been devised to solve the above-mentioned problems. That is, a main object of the present invention is to provide a technology that promotes the utilization of personal data while protecting personal information.

[0008] In order to achieve the above-mentioned object, one aspect of the proposal system of the present invention comprises: a graph generation unit that generates a knowledge graph representing the relationships between multiple nodes including multiple information holders using element identification information that identifies personal data elements, which are data elements that make up personal data, and data location information of the personal data elements that represent information holders that hold the personal data elements; a reception unit that receives an information request requesting information on the information holder represented by the data location information of the personal data elements that meets the user's needs; and a proposal unit that uses a model generated by machine learning using the generated knowledge graph and the requirements of the information holder that meets the information request to output an information holder that meets the requirements according to the received information request as a recommended information holder.

[0009] In one aspect, the proposed method of the present invention generates, by a computer, a knowledge graph representing the relationships between multiple nodes including multiple information holders, using element identification information that identifies personal data elements, which are data elements that make up personal data, and data location information of the personal data elements that represent information holders that hold the personal data elements; further, the computer receives an information request requesting information on the information holder represented by the data location information of the personal data elements that meets the user's needs; and outputs, as a recommended information holder, an information holder that meets the requirements according to the received information request, using a model generated by machine learning using the generated knowledge graph and the requirements of the information holder that meets the information request.

[0010] Furthermore, in one aspect, the program storage medium of the present invention stores a computer program for causing a computer to execute the following processes: receiving an information request requesting information on an information holder represented by data location information of a personal data element that meets a user's needs; and outputting an information holder that meets the requirements according to the received information request as a recommended information holder, using a model generated by machine learning using a knowledge graph representing the relationship between multiple nodes including multiple information holders, generated using element identification information that identifies the personal data element and the data location information of the personal data element that represents the information holder that holds the personal data element, and the requirements of the information holder that meets the information request.

[0011] According to the present invention, it is possible to promote the use of personal data while protecting personal information.

[0012] FIG. 1 is a diagram illustrating the configuration of a proposal system according to a first embodiment of the present invention. FIG. 2 is a diagram illustrating an example of personal data. FIG. 3 is a diagram illustrating data location information of personal data elements. FIG. 4 is a diagram illustrating an example of a knowledge graph. FIG. 5 is a flowchart illustrating an example of the operation of a processing device constituting the proposal system. FIG. 6 is a diagram illustrating another embodiment. FIG. 7 is a flowchart illustrating an example of the operation of a computer in another embodiment.

[0013] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0014] First Embodiment Fig. 1 is a diagram illustrating the configuration of a personal data distribution system including a proposal system according to the present invention. The personal data distribution system 100 is a system that supports the utilization of personal data. Personal data includes not only personal information (personal data) such as name, date of birth, and address that can identify an individual, but also various data related to an individual that can be found to have a relationship with that individual. Specific examples of data that can be found to have a relationship with an individual include personal attribute information (age, occupation, residential area, etc.), movement, behavior, and purchase history, and data from wearable devices.

[0015] In the example of FIG. 1 , the personal data distribution system 100 is connected to a plurality of information holders 50 and system users 80. Here, the information holders 50 are businesses that hold personal data. Specific examples of the information holders 50 include hospitals that hold patient data, which is personal data, and sports gyms and amusement facilities that hold membership information, which is personal data. Further specific examples of the information holders 50 include service providers such as logistics, retail, and information distribution companies that hold customer information, which is personal data. The system users 80 are businesses and local governments that use the personal data distribution system 100.

[0016] In the first embodiment, the personal data distribution system 100 has a function of presenting to the system user 80, as a recommended information holder, an information holder 50 that can be accessed to provide the personal data-related information desired by the system user 80. Note that a plurality of information holders 50 and a plurality of system users 80 can each be connected to the personal data distribution system 100. Here, the same reference numerals will be used to refer to the different information holders connected to the personal data distribution system 100, and the description herein will not distinguish between the multiple information holders. Similarly, the same reference numerals will be used to refer to the different system user businesses connected to the personal data distribution system 100, and the description herein will not distinguish between the multiple system user businesses.

[0017] Furthermore, in personal data distribution system 100, security measures are taken regarding connections with information holder 50 and system users 80, but a description of these will be omitted here.

[0018] The personal data distribution system 100 includes a proposal system 1, and further includes a holder attribute information DB (Database) 2, a personal attribute information DB (Database) 3, and a data location information DB (Database) 4, which are connected to the proposal system 1. The holder attribute information DB 2 is a database (storage device) that stores attribute information of information holders 50 (hereinafter also referred to as holder attribute information). Examples of holder attribute information include information on the business type, location, and service content of a business. Such holder attribute information is stored in the holder attribute information DB 2 in association with holder identification information that identifies the information holder 50.

[0019] The personal attribute information DB3 is a database (storage device) that stores attribute information (hereinafter also referred to as personal attribute information) about individuals whose personal data is stored in the information storage device 50. Examples of personal attribute information include information about various personal attributes such as age, city / town / village of residence, family structure, occupation, hobbies, and preferences. Such personal attribute information is associated with system personal identification information that identifies individuals and is stored in the personal attribute information DB3. Note that the system personal identification information is personal identification information unique to this system and is composed of information that differs from personal identification information of other systems, such as My Number, driver's license number, and passport number.

[0020] Personal attribute information is information related to an individual and is often associated with personal information. Here, it is considered one of the data elements (personal data elements) that make up personal data. Because the personal data distribution system 100 handles such personal attribute information (in other words, personal data), it is equipped with a consent management device (computer device) 7, as represented by the dotted line in FIG. 1 . The consent management device 7 is a device that has the function of obtaining consent from individuals corresponding to the personal data regarding the handling of the personal data when the system stores the personal data in the personal attribute information DB 3 or the like. Here, the configuration of the consent management device 7 is not limited as long as it has the function of obtaining consent regarding the handling of the personal data, and therefore a description thereof will be omitted.

[0021] The data location information DB4 is a database (storage device) that stores data location information. Data location information is information that associates element identification information, which identifies the type (content) of data elements (hereinafter also referred to as personal data elements) that constitute personal data, with location information that indicates the information storage unit 50 in which the personal data elements are stored. For example, assume that Hospital A and Clinic B, which are information storage units 50, each store personal data of Patient X. In this case, specific examples of Patient X's personal data stored by Hospital A and Clinic B are shown in FIG. 2. In the example of FIG. 2, Patient X's personal data stored in Hospital A and Clinic B includes personal information, personal attribute information, medical history information, and prescription history information. In such a case, for example, personal data elements such as personal information, personal attribute information, medical history, and prescription history may be set as personal data elements of the personal data. However, here, a specific example is shown in which personal data elements are set by further subdividing the personal data as follows. For example, as shown in FIG. 3, information indicating the medical visit details included in the medical history information is set as a personal data element. In this case, the element identification information for identifying the personal data element is set to be information on the name of the visit, which does not include the name of the disease but indicates the details of the visit, such as examination, test, and test results. On the other hand, the location information is information that indicates the information holder 50 in which such personal data element is held, and is composed of information on the name of the medical institution, such as Hospital A or Clinic B, and holder identification information. In such a case, one piece of data location information for the personal data element related to Patient X is, for example, information in which location information represented by information such as the name of the medical institution, "Clinic B," is associated with element identification information called "examination," which is associated with the date of visit, August 8, 2022.

[0022] As another example of a personal data element, for example, attributes such as "age 40s," "resides in XX city," and "company employee" included in personal attribute information may each be set as a personal data element. In such a case, an example of data location information for a personal data element is information in which element identification information, which is information on the attribute name representing "age 40s," is associated with location information represented by information such as the name of the information holder 50 that holds the personal attribute information. Note that personal data elements may be set appropriately taking into account the expected content of utilization of personal data, and are not limited to the above-mentioned example. Furthermore, the element identification information of a personal data element is associated with the personal data element and stored in the personal attribute information DB 3.

[0023] The holder attribute information DB 2, the personal attribute information DB 3, and the data location information DB 4 as described above are here separate and independent as a measure against information leaks due to cyber attacks or the like.

[0024] As shown in FIG. 1, the proposed system 1 constituting the personal data distribution system 100 includes an arithmetic device 10 and a processing device 20. The arithmetic device 10 and the processing device 20 are each computer devices and include processors 11, 21 and storage devices 30, 40. The storage devices 30, 40 include storage media for storing data and computer programs (hereinafter also referred to as programs) 33, 43. There are many types of storage devices, and the types of storage devices 30, 40 included in the arithmetic device 10 and the processing device 20, respectively, are not limited to one. Computer devices often include multiple types of storage devices. The types and numbers of storage devices 30, 40 included in the arithmetic device 10 and the processing device 20 are not limited here, and a description thereof will be omitted. Furthermore, when the arithmetic device 10 and the processing device 20 include multiple types of storage devices 30, 40, they will be collectively referred to as storage devices 30, 40.

[0025] The processors 11, 21 provided in the arithmetic device 10 and the processing device 20, respectively, are configured by processors such as a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit). The processors 11, 21 can have various functions based on the programs 33, 43 by reading and executing the programs 33, 43 stored in the storage devices 30, 40. Here, the processor 11 of the arithmetic device 10 includes a graph generation unit 15 and a learning unit 16 as functional units related to the utilization of personal data. The processor 21 of the processing device 20 includes a receiving unit 25 and a proposing unit 26.

[0026] The graph generation unit 15 in the computing device 10 generates a knowledge graph using AI (artificial intelligence) technology, using data stored in the holder attribute information DB 2, the personal attribute information DB 3, and the data location information DB 4. The knowledge graph generated here includes, as nodes, multiple information holders 50, individuals whose personal data is stored in the information holders 50, and element identification information related to the personal attribute information, and the relationships between these nodes are represented by edges. The graph generation unit 15 generates the knowledge graph using element identification information of personal data elements, data location information of personal data elements, attribute information of the information holders 50, etc. Figure 4 is a schematic diagram showing an example of a knowledge graph. In Figure 4, circles represent nodes in the knowledge graph, and lines connecting the nodes represent edges in the knowledge graph. The nodes constituting the knowledge graph are set appropriately by a designer or the like, taking into account the intended use of the knowledge graph to be generated. If necessary, open data obtained from information sources 90 outside the system, as shown by the dotted lines in Figure 1 (e.g., data on the location area of ​​the information holder 50 or data related to the patient's residential area) may also be used to generate the knowledge graph.

[0027] The learning unit 16 generates a proposal model using AI technology. Here, the proposal model is a model that receives an information request as input and outputs an information repository 50 that meets the requirements for meeting the information request as a recommended information repository. The information request is a request for information on an information repository represented by data location information for personal data elements that meet the user's needs. In the first embodiment, the information request is a request for a proposal (introduction) of an information repository 50 that holds personal data-related information desired by the system user (user) 80 (in other words, data for personal data elements that meet the user's needs) and is suitable for access. To give a specific example, assume that the system user 80 is planning a new service themed around "health promotion" in "XX City" and is searching for an information repository 50 that provides information (data) related to "health promotion" in "XX City" as an access destination. An example of an information request that meets such a user's needs is a request for a recommendation of a recommended information repository 50 that provides information (data) related to "health promotion" in "XX City." As a first requirement for an information holder 50 to respond to such an information request, for example, the information holder 50 must be an information holder 50 in which a personal data element corresponding to "XX City", which is the element identification information of the personal data element, is located (stored), and the number of pieces of data must be equal to or greater than a predetermined threshold. Furthermore, as a second requirement for an information holder 50 to respond to such an information request, the information holder 50 must be an information holder 50 that is directly connected to the node "health improvement" in the knowledge graph, or whose node-to-node distance to the node "health improvement" is equal to or less than a predetermined value.

[0028] The proposal model is generated by machine learning using the knowledge graph generated by the graph generation unit 15 and relational data representing the relationship between the information request and the requirements of the information holder 50 that meets the information request. The method for generating the proposal model is not limited to AI technology that generates a model using a knowledge graph. For example, the proposal model is generated by a technique called link prediction AI (artificial intelligence). Link prediction AI is an explainable AI technology. The proposal model generated using such explainable AI technology not only outputs information holders 50 that meet the requirements for meeting the information request as recommended information holders, but also outputs information explaining the basis for the recommendation. For example, suppose an information request is input to the proposal model requesting that information holders 50 that provide information (data) related to "health promotion" in "XX City" be recommended. In this case, the proposal model outputs name information representing the information holders 50, "Sports Gym D" and "Clinic B," as recommended information holders. Furthermore, the proposed model outputs information that, for example, "Sports Gym D" and "Clinic B" hold more than the number of pieces of personal data of individuals that are effective for analysis, including personal data elements related to "health promotion," as the basis for recommending these information holders 50.

[0029] As described above, the proposed model generated by the learning unit 16 is stored in the storage device 40 in the processing device 20.

[0030] The reception unit 25 in the processing device 20 receives an information request output from a request source such as a system user 80. Regarding input of an information request to the processing device 20, there are two possible cases: the information request is input from the request source to the processing device 20 via an information communication network, and the information request is input to the processing device 20 by, for example, manual input by an operator of the processing device 20 upon receiving the information request from the request source. The processing device 20 is configured to allow both of these information request input methods or one of them selected in advance.

[0031] When an information request is received by the receiving unit 25, the suggestion unit 26 inputs the received information request into the proposal model generated by the learning unit 16. The suggestion unit 26 then returns the information in the information holder 50 output from the proposal model to the request source as suggested information for the recommended information holder. If the proposal model also outputs information explaining the basis, the suggestion unit 26 also returns the explanation of the basis to the request source. Examples of replying methods include a method in which the processing device 20 returns a response to the system user 80's computer device via an information and communication network, or a method in which the operator of the processing device 20 returns a response using a communication method such as e-mail. Here, the replying method for the response generated by the suggestion unit 26 in response to the information request is not limited. The information output from the proposal model may not only be returned to the request source, but may also be notified to the operator of the processing device 20, for example, using a display device (not shown) connected to the processing device 20. Furthermore, the personal data distribution system 100 may have a function of acting as an intermediary between the recommended information holder 50 proposed by the proposal system 1 and the system user 80 who has sent the request.

[0032] The proposal system 1 has the configuration described above. An example of the operation of the processing device 20 in the proposal system 1 will be described below with reference to FIG. 5. FIG. 5 is a flowchart illustrating an example of the operation of the processing device 20. In this description of the operation, it is assumed that the proposal model is generated in advance by the computing device 10 and stored in, for example, the storage device 40. As described above, the proposal model is a model that receives an information request as input and outputs an information holder 50 that meets the requirements for responding to the information request as a recommended information holder 50. In this case, the information holder 50 recommended by the proposal model is an information holder 50 that is estimated to provide information related to personal data requested by a system user 80 when accessed by the system user 80.

[0033] When such a proposal model has been stored in the storage device 40, and the reception unit 25 of the processing device 20 receives an information request (step 101 in FIG. 5), the proposal unit 26 inputs the received information request into the proposal model (step 102). After that, when the proposal model outputs information on a recommended information holder 50 in response to the information request, the proposal unit 26 outputs the output information of the proposal model as proposal information to the request source that issued the information request (step 103).

[0034] The proposal system 1 in the first embodiment has the configuration described above. That is, when processing information requests related to personal data, the proposal system 1 performs processing using element identification information that identifies personal data elements and data location information for the personal data elements, rather than the personal data elements themselves, such as personal information (personal data) included in the personal data. Therefore, the proposal system 1 can provide the system user 80 with information that leads to the utilization of personal data while protecting personal information. In other words, the proposal system 1 can contribute to promoting the utilization of personal data while protecting personal information.

[0035] Furthermore, the proposal system 1 proposes information on recommended information holders 50 in response to an information request using a proposal model generated using a knowledge graph. By using a knowledge graph, the processing device 20 can predict information holders 50 that are highly relevant to matters related to the information request, even if the number of information holders 50 is enormous. This also contributes to promoting the use of personal data.

[0036] Second Embodiment A second embodiment of the present invention will be described below. In the description of the second embodiment, the same reference numerals will be used to designate components that are the same as those in the first embodiment, and duplicate descriptions thereof will be omitted.

[0037] In the second embodiment, the proposal system 1 constituting the personal data distribution system 100 outputs information from the information holder 50 with which it cooperates or shares functions. That is, in the proposal system 1 in the second embodiment, the information requests it receives (in other words, the anticipated information requests) differ from those in the first embodiment, and a knowledge graph is generated to respond to the information requests, and a proposal model is generated using the knowledge graph. In the proposal system 1 of the second embodiment, the configurations (functions) other than those related to the information requests are the same as in the first embodiment.

[0038] That is, in the second embodiment, the information request is a request to propose a combination of information holders 50 that cooperate or share functions in accordance with the user's needs. Here, a specific example of an information request is given. Assume that the information holder 50 is a medical institution. The information request is a request for information that meets the user's needs, namely, a proposal for a combination of information holders (medical institutions) 50 that are preferable for sharing functions, using information about medical visits, which is a personal data element of the patient. A requirement for satisfying such an information request is, for example, a requirement that a combination of multiple medical institutions visited by a single patient must have a threshold or greater number of patients who visited the combined medical institutions. For example, suppose that patient "001," who is in his or her 80s and suffers from diabetes, visits the cardiology department of medical institution AA. Furthermore, suppose that patient "001" also visits the nephrology department of medical institution BB. In this case, medical institution AA and medical institution BB hold personal data elements related to the medical visit history that constitute the personal data of patient "001." When referring to the data location information of personal data elements related to other patients, it is assumed that, as in the above, the number of patients visiting the cardiology department of medical institution AA and the nephrology department of medical institution BB is above a threshold. In such a case, medical institution AA and medical institution BB are recommended as a combination of linked information holders 50, since it is preferable for them to share patient medical record information when conducting examinations and treatments. In other words, the proposed system 1 outputs information on the recommended information holder 50, which is the combination of medical institution AA and medical institution BB, which are information holders 50 that meet the requirements for responding to the information request. In this case, the request source (system user 80) that issues the information request may be a local government or business operator proposing collaboration between medical institutions.

[0039] Another example of an information request is a request from a system user 80 to propose an information holder 50 suitable for sharing functions with the system user. Specifically, for example, when a patient is discharged from a medical institution, the medical institution may search for another medical institution to refer the patient to that will examine the patient and prescribe medication after discharge. In such a case, the system user 80 is both the medical institution where the patient is hospitalized and the information holder 50. An example of a requirement for responding to the information request described above is that the information holder (medical institution) 50 must have at least a predetermined number of patients with the same personal attributes as the patient's residential area, including multiple personal attributes. Specifically, a first requirement for a medical institution P to refer to a discharged patient "005" is that the medical institution (information holder 50) must have holder attribute information corresponding to the residential area of ​​the discharged patient "005." Furthermore, the second requirement is that the medical institution P holds personal data elements of medical examination history and that the information holder 50 (medical institution) is a place where many other patients associated with personal data elements representing the age attribute of the same generation as patient "005" have visited. The requirement that meets the information request is a requirement that satisfies both the first and second requirements.

[0040] In order to output information on recommended information holders 50 in response to the information request described above, in the proposal system 1 of the second embodiment, the learning unit 16 generates the following proposal model. That is, in the first embodiment, the information request requests information on information holders 50 recommended as information recipients, whereas in the second embodiment, the information request requests information on information holders 50 recommended as information holders for collaboration or functional sharing. In the second embodiment, a knowledge graph including multiple nodes set to respond to such information requests related to collaboration or functional sharing is generated by the graph generation unit 15 of the computing device 10. The learning unit 16 of the computing device 10 generates the proposal model by machine learning (e.g., link prediction AI) using the generated knowledge graph and requirements of information holders that respond to information requests related to collaboration or functional sharing.

[0041] In the processing device 20, as in the first embodiment, the reception unit 25 receives an information request, the proposal unit 26 inputs the information request into a proposed model, and the output of the proposed model is output as information for the recommended information holder 50 (an information holder 50 that cooperates or shares functions).

[0042] Similar to the first embodiment, the proposal system 1 of the second embodiment performs processing to respond to information requests involving personal data using element identification information that identifies personal data elements and data location information of the personal data elements, rather than the personal data elements themselves. As a result, similar to the first embodiment, the proposal system 1 of the second embodiment can contribute to promoting the utilization of personal data while protecting personal information.

[0043] Also in the second embodiment, the proposal system 1 proposes information on recommended information holders 50 in response to an information request using a proposal model generated using a knowledge graph. Therefore, even if the number of information holders 50 is enormous, the processing device 20 can predict an appropriate information holder 50 in response to an information request related to collaboration or functional sharing.

[0044] Other Embodiments The present invention is not limited to the first and second embodiments and may employ various other embodiments. For example, in the first and second embodiments, the proposal model outputs information about the recommended information holder 50 and an explanation of the reasons for recommending the information holder 50. The proposal model may be generated to output the following information as well. That is, the proposal model may be generated to output information representing a person profile related to the information request. As a specific example, suppose there is an information request for a recommendation of an information holder 50 that provides information (data) related to "health promotion" in "XX City." With regard to such an information request, the proposal model may be generated so that it outputs information such as "a man in his 80s who is undergoing rehabilitation" or "a woman in her 60s who swims for weight loss," as person profiles that are highly associated with the nodes "XX City" and "health promotion" in the knowledge graph.

[0045] FIG. 6 is a diagram illustrating another embodiment of a proposal system according to the present invention. The proposal system 70 includes a receiving unit 71, a proposal unit 72, and a graph generation unit 73. The receiving unit 71, the proposal unit 72, and the graph generation unit 73 are functional units implemented, for example, by a computer device executing a computer program. The graph generation unit 73 generates a knowledge graph representing the relationships between multiple nodes including multiple information holders. This knowledge graph is generated using element identification information identifying personal data elements, which are data elements constituting personal data, and data location information of personal data elements representing information holders that hold the personal data elements. The receiving unit 71 receives an information request. The information request is a request for information on an information holder represented by the data location information of personal data elements that meets the user's needs. The proposal unit 72 uses a model generated by machine learning. This model is generated by machine learning using the knowledge graph generated by the graph generation unit 73 and the requirements of the information holder that meets the information request. The proposal unit 72 uses the model to output information holders that meet the requirements of the received information request as recommended information holders.

[0046] 7 is a flowchart showing an example of the operation of a computer included in the proposal system 70. For example, assume that the above-described model has been generated by machine learning using a knowledge graph. First, the reception unit 71 receives an information request requesting information on an information holder represented by data location information of personal data elements that meet the user's needs (step 201 in FIG. 7). Thereafter, the proposal unit 72 outputs information on information holders that meet the requirements of the information request as information on recommended information holders (step 202).

[0047] The proposed system 70 in another embodiment has the above-mentioned configuration, thereby promoting the utilization of personal data while protecting personal information. This is because the proposed system 70 does not use the personal data itself, but rather uses element identification information that identifies the personal data elements that make up the personal data, and data location information of the personal data elements that indicates the information holder that holds the personal data elements. This has the effect of making it easier to provide information related to personal data.

[0048] The present invention has been described above using the above-described embodiments as exemplary examples. However, the present invention is not limited to the above-described embodiments. In other words, the present invention can be applied in various aspects that can be understood by a person skilled in the art within the scope of the present invention.

[0049] 1, 70 Proposal system 15, 73 Graph generation unit 25, 71 Reception unit 26, 72 Proposal unit

Claims

1. a graph generation means for generating a knowledge graph representing the relationship between a plurality of nodes including a plurality of information holders, using element identification information for identifying personal data elements that are data elements constituting personal data and data location information of the personal data elements that represents an information holder that holds the personal data elements; a receiving means for receiving an information request for information on an information holder represented by data location information of a personal data element that meets a user's needs; a suggestion means for outputting, as a recommended information holder, an information holder that satisfies the requirements according to the received information request, using a model generated by machine learning using the generated knowledge graph and the requirements of the information holder that meets the information request; The proposed system comprises:

2. The information request is a request to output information on a recommended information holder for collaboration or function sharing. The recommendation system of claim 1 .

3. The information request is a request to output information about a recommended information holder that requests the provision of information related to personal data. The recommendation system of claim 1 .

4. a learning means for generating a model to be used by the proposing means; The recommendation system of claim 1 .

5. The learning means generates a model using a link prediction technique that uses a knowledge graph. The proposal system of claim 4 .

6. By computer, generating a knowledge graph representing the relationship between a plurality of nodes including a plurality of information holders using element identification information for identifying personal data elements that are data elements constituting personal data and data location information of the personal data elements that represent information holders that hold the personal data elements; Furthermore, by computer, receiving an information request for information on an information holder represented by data location information of a personal data element that meets a user's needs; Using a model generated by machine learning using the generated knowledge graph and the requirements of information holders that meet the information request, an information holder that meets the requirements of the received information request is output as a recommended information holder. Proposal method.

7. receiving an information request for information on an information holder represented by data location information of a personal data element that meets a user's needs; a process of outputting an information holder that satisfies the requirements according to the received information request as a recommended information holder, using a knowledge graph that represents the relationship between a plurality of nodes including a plurality of information holders, the knowledge graph being generated using element identification information that identifies the personal data element and data location information of the personal data element that represents the information holder that holds the personal data element, and a model generated by machine learning using requirements for the information holder that satisfies the information request; A computer program for causing a computer to execute the above.