Information processing device and information processing method
Patent Information
- Application Number
- JP2024033800
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2024-03-06
- Publication Date
- 2024-11-13
- Estimated Expiration
- 2044-03-06
AI Technical Summary
【0013】 本発明によれば、匿名化されている複数のユーザ情報を結合することができるという効果を奏する。
Smart Images

Figure 0007583209000001_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to an information processing device and an information processing method. [Background technology]
[0002] Conventionally, user information, which is information about users, is collected from multiple businesses, and the collected information is combined to perform data analysis. For example, Patent Document 1 discloses a system that combines data on personal information of users corresponding to multiple businesses based on a combination key for combining multiple pieces of user information. [Prior art documents] [Patent documents]
[0003] [Patent Document 1] JP 2021-117679 A Summary of the Invention [Problem to be solved by the invention]
[0004] In order to protect the personal information of users, user information is anonymized. If multiple businesses each anonymize their user information, this user information cannot be combined, which causes a problem in that data analysis cannot be performed.
[0005] The present invention has been made in consideration of these points, and has an object to make it possible to combine a plurality of pieces of anonymous user information. [Means for solving the problem]
[0006] An information processing device according to a first aspect of the present invention includes a first acquisition unit that acquires a first information group including first attribute information that is attribute information indicating an attribute of a user collected by a first operator, and a second information group including second attribute information that is attribute information indicating an attribute of a user collected by a second operator different from the first operator, a generation unit that performs clustering of attribute information based on the first attribute information included in the first information group to generate a plurality of first clusters that are clusters corresponding to the first information group, and performs clustering of attribute information based on the second attribute information included in the second information group to generate a plurality of second clusters that are clusters corresponding to the second information group, and a parameter for combining a plurality of clusters in response to input of attribute information. a second acquisition unit that inputs the attribute information corresponding to each of the plurality of first clusters into a machine learning model that outputs a first parameter that is the parameter of each of the plurality of first clusters from the machine learning model, and inputs the attribute information corresponding to each of the plurality of second clusters into the machine learning model and acquires second parameters that are the parameters corresponding to each of the plurality of second clusters from the machine learning model; and a combination unit that combines one or more of the second clusters to at least any of the plurality of first clusters based on the first parameter of each of the plurality of first clusters and the second parameter of each of the plurality of second clusters.
[0007] The second acquisition unit may acquire one or more first parameters corresponding to each of the plurality of first clusters, and acquire one or more second parameters corresponding to each of the plurality of second clusters, using the machine learning model that outputs, in response to input of the attribute information, one or more parameters indicating characteristics of a person having the attribute information.
[0008] The generation unit may generate the first cluster and the second cluster associated with range information indicating a range of values of attributes summarized by clustering, and the second acquisition unit may acquire the first parameters from the machine learning model by inputting the attribute corresponding to each of the multiple first clusters and the range information corresponding to the attribute to the machine learning model, which outputs parameters for combining multiple clusters in response to an input of the attribute and the range information corresponding to the attribute, and may acquire the second parameters from the machine learning model by inputting the attribute corresponding to each of the multiple second clusters and the range information corresponding to the attribute to the machine learning model.
[0009] The information processing device may further include a distribution unit that distributes information corresponding to attribute information indicated by the second cluster combined with the first cluster to a terminal of a user having attribute information indicated by the first cluster.
[0010] The generation unit may calculate a first cluster membership degree indicating the degree to which a user corresponding to each of a plurality of records belongs to the first cluster based on a degree of match between attribute information indicated by the generated first cluster and attribute information of each of a plurality of records included in the first information group, and the distribution unit may preferentially distribute information corresponding to the attribute information indicated by the second cluster combined with the first cluster to a terminal of a user having a relatively high first cluster membership degree to the first cluster.
[0011] The first acquisition unit may acquire the first information group from a first device used by the first operator, and acquire a second information group from a second device used by the second operator, and the information processing device may further have a transmission unit that transmits information indicating the combination of the first cluster and the second cluster combined by the combination unit to at least one of the first device and the second device.
[0012] An information processing method according to a second aspect of the present invention includes the steps of: acquiring a first information group including first attribute information that is attribute information indicating an attribute of a user and collected by a first operator; and acquiring a second information group including second attribute information that is attribute information indicating an attribute of a user and collected by a second operator different from the first operator; clustering attribute information based on the first attribute information included in the first information group to generate a plurality of first clusters that are clusters corresponding to the first information group, and clustering attribute information based on the second attribute information included in the second information group to generate a plurality of second clusters that are clusters corresponding to the second information group; and calculating a number of clusters for combining the plurality of clusters in response to input of attribute information. the step of inputting the attribute information corresponding to each of the plurality of first clusters into a machine learning model that outputs parameters of the plurality of first clusters, and acquiring first parameters that are the parameters of each of the plurality of first clusters from the machine learning model, and inputting the attribute information corresponding to each of the plurality of second clusters into the machine learning model, and acquiring second parameters that are the parameters corresponding to each of the plurality of second clusters from the machine learning model; and a step of combining one or more of the second clusters to at least any of the plurality of first clusters based on the first parameters of each of the plurality of first clusters and the second parameters of each of the plurality of second clusters. Effect of the Invention
[0013] According to the present invention, it is possible to advantageously combine a plurality of pieces of anonymous user information. [Brief description of the drawings]
[0014] [Figure 1] FIG. 1 is a diagram illustrating an overview of an information processing system. [Diagram 2] FIG. 2 is a diagram illustrating a functional configuration of an information processing device. [Diagram 3] 4A to 4C are diagrams showing an example of a first information group and a second information group. [Figure 4]4 is a diagram showing an example of a first cluster generated from the first information group shown in FIG. 3. FIG. [Diagram 5] 4 is a diagram showing an example of a second cluster generated from the second information group shown in FIG. 3. FIG. [Figure 6] FIG. 11 is a diagram showing an example of a template of instruction information. [Figure 7] 13 is a diagram showing an example of instruction information including attribute information corresponding to a first cluster. FIG. [Figure 8] 8 is a diagram showing an example of text information acquired from a large-scale language model for instruction information corresponding to FIG. 7; FIG. [Figure 9] FIG. 4 is a diagram illustrating an example of a first parameter and a second parameter. [Figure 10] FIG. 13 is a diagram showing an example of coupled cluster information. [Figure 11] 4 is a flowchart showing a process flow in an information processing device. [Figure 12] FIG. 1 is a diagram illustrating an overview of an information processing system in which a first device and a second device form a cluster. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0015] [Outline of Information Processing System S] 1 is a diagram illustrating an overview of an information processing system S. The information processing system S includes an information processing device 1, a first device 2, and a second device 3, and is a system that combines user information, which is information about users of each of a plurality of businesses.
[0016] The information processing device 1 is, for example, a computer used by an aggregator that aggregates data and provides a service of providing the aggregated data. The information processing device 1 is communicably connected to a first device 2 used by the first operator and a second device 3 used by a second operator different from the first operator via a communication network (not shown) such as the Internet or a mobile phone line.
[0017] The information processing device 1 acquires from the first device 2 a first information group including first attribute information which is attribute information indicating user attributes collected by a first business operator, and acquires from the second device 3 a second information group including second attribute information which is attribute information indicating user attributes collected by a second business operator ((1) and (2) in FIG. 1).
[0018] The information processing device 1 performs clustering of the attribute information based on the first attribute information included in the first information group, and generates a plurality of first clusters that are clusters corresponding to the first information group. Similarly, the information processing device 1 performs clustering of the attribute information based on the second attribute information included in the second information group, and generates a plurality of second clusters that are clusters corresponding to the second information group ((3) in FIG. 1).
[0019] The information processing device 1 inputs attribute information corresponding to each cluster to a machine learning model that outputs parameters for combining multiple clusters in response to input attribute information ((4) in FIG. 1), and acquires parameters corresponding to each cluster from the machine learning model ((5) in FIG. 1). Specifically, the information processing device 1 inputs attribute information corresponding to each of multiple first clusters to the machine learning model, and acquires first parameters that are parameters for each of the multiple first clusters from the machine learning model. Similarly, the information processing device 1 inputs attribute information corresponding to each of multiple second clusters to the machine learning model, and acquires second parameters that are parameters corresponding to each of the multiple second clusters from the machine learning model.
[0020] The information processing device 1 combines one or more second clusters with at least one of the first clusters based on the first parameter of each of the first clusters and the second parameter of each of the second clusters ((6) in FIG. 1). In this way, the information processing device 1 can combine multiple pieces of anonymized user information.
[0021] [Functional configuration of information processing device 1] Next, a description will be given of the functional configuration of the information processing device 1. FIG.
[0022] As shown in FIG. 2, the information processing device 1 includes a communication unit 11, a storage unit 12, and a control unit 13. The communication unit 11 is a communication interface for transmitting and receiving data to and from the first device 2, the second device 3, and the like via a communication network.
[0023] The storage unit 12 is a storage medium that stores various data, and includes a ROM (Read Only Memory), a RAM (Random Access Memory), a hard disk, an SSD (Solid State Drive), a flash memory, etc. The storage unit 12 stores a program executed by the control unit 13. The storage unit 12 stores a program that causes the control unit 13 to function as a first acquisition unit 131, a generation unit 132, a second acquisition unit 133, a combination unit 134, and a transmission unit 135.
[0024] The control unit 13 is, for example, a CPU (Central Processing Unit). The control unit 13 executes a program stored in the storage unit 12 to function as a first acquisition unit 131, a generation unit 132, a second acquisition unit 133, a combination unit 134, and a transmission unit 135.
[0025] The first acquisition unit 131 acquires a first information group including first attribute information that is attribute information indicating an attribute of a user collected by a first business operator, and a second information group including second attribute information that is attribute information indicating an attribute of a user collected by a second business operator different from the first business operator. For example, the first acquisition unit 131 acquires the first information group from a first device 2 used by the first business operator, and acquires the second information group from a second device 3 used by the second business operator.
[0026] FIG. 3 is a diagram showing an example of a first information group and a second information group. FIG. 3(a) shows the first information group, and FIG. 3(b) shows the second information group. As shown in FIG. 3(a), the first information group includes an anonymous user ID as user identification information for identifying a user, and the user ID is associated with a static attribute "user's age" and a dynamic attribute that may change over time, indicating the type of smartphone used by the user, as attributes included in the first attribute information. Also, as shown in FIG. 3(b), the second information group includes an anonymous user ID for identifying a user. Also, the second information group includes a dynamic attribute "purchased product category" that is the category of the product purchased by the user, and a dynamic attribute "point use" that indicates the amount of points used by the user in shopping, as attributes included in the second attribute information, as attributes included in the second attribute information.
[0027] Although the first information group and the second information group each include an anonymous user ID, this is not limited to the above. For example, the first information group and the second information group may not include a user ID or may include a user ID in a different format that cannot be associated with each other.
[0028] The generation unit 132 performs clustering of the attribute information based on the first attribute information included in the first information group to generate one or more first clusters that are clusters corresponding to the first information group, and performs clustering of the attribute information based on the second attribute information included in the second information group to generate one or more second clusters that are clusters corresponding to the second information group. The generation unit 132 generates a plurality of first clusters and second clusters associated with range information that indicates the range of values indicated by the attributes summarized by the clustering.
[0029] For example, the generation unit 132 uses a clustering method such as the k-means method or LDA (Latent Dirichlet Allocation) to generate multiple first clusters that correspond to a first information group, and generate multiple second clusters that correspond to a second information group.
[0030] FIG. 4 is a diagram showing an example of a first cluster generated from the first information group shown in FIG. 3. FIG. 4(a) shows an example in which a first cluster ID for identifying a first cluster is assigned to each record included in the first information group. FIG. 4(b) shows first cluster information showing the contents of the first cluster. The first cluster information is, for example, information in which the first cluster IDs of a plurality of first clusters are associated with range information showing the range of attribute values corresponding to each of a plurality of attributes corresponding to the first cluster. For example, as a result of clustering a record with a first cluster ID of "1", it can be confirmed that "young age" is specified as the range information corresponding to the attribute "generation" included in the first attribute information, and "latest model" is specified as the range information corresponding to the attribute "model used".
[0031] After generating the first cluster, the generating unit 132 may calculate a degree of match between attribute information indicated by the generated first cluster and attribute information of each of the multiple records included in the first information group. Then, the generating unit 132 may calculate a first cluster membership degree indicating a degree to which a user corresponding to each of the multiple records belongs to the first cluster based on the calculated degree of match. For this calculation, a clustering method such as the well-known LDA (Latent Dirichlet Allocation) can be used.
[0032] FIG. 5 is a diagram showing an example of a second cluster generated from the second information group shown in FIG. 3. FIG. 5(a) shows an example in which a second cluster ID for identifying a second cluster is assigned to each record included in the second information group. FIG. 4(b) shows second cluster information showing the contents of a second cluster. The second cluster information is, for example, information in which the second cluster IDs of a plurality of second clusters are associated with range information showing the range of values of attributes corresponding to each of a plurality of attributes. For example, as a result of clustering a record with a second cluster ID of "1", it can be confirmed that "health / outdoors" is specified as the range information corresponding to the attribute "purchased product category" included in the second attribute information, and "few" is specified as the range information of the attribute "point use".
[0033] Note that the generating unit 132 may generate clusters so as to satisfy k-anonymity, that is, the number of data constituting the cluster, i.e., the number of users constituting the cluster is k or more. In this way, the information processing device 1 can perform clustering while ensuring the anonymity of users.
[0034] The second acquisition unit 133 inputs attribute information corresponding to each of the multiple first clusters to a machine learning model that outputs a combining parameter that is a parameter for combining multiple clusters in response to input of attribute information, and acquires the first parameter that is a combining parameter for each of the multiple first clusters from the machine learning model. Also, the second acquisition unit 133 inputs attribute information corresponding to each of the multiple second clusters to the machine learning model, and acquires the second parameter that is a combining parameter corresponding to each of the multiple second clusters from the machine learning model.
[0035] Here, the machine learning model is, for example, a generative AI (Artificial Intelligence) such as a large-scale language model (LLM) that outputs text information including parameters for combining multiple clusters in response to input of attribute information. The large-scale language model can output one or more combining parameters indicating characteristics of a person having attribute information in response to input of an attribute as attribute information and range information corresponding to the attribute.
[0036] First, the second acquisition unit 133 generates instruction information including attribute information. For example, a template of the instruction information is stored in the storage unit 12. FIG. 6 is a diagram showing an example of the template of the instruction information. In the example shown in FIG. 6, as user information, "@propaerty1" and "@propaerty2" are shown as parameters for specifying attributes, and "@values1" and "@values2" are shown as parameters for specifying range information corresponding to the attributes, of which "@value1" and "@value2" are shown to be taken.
[0037] The second acquisition unit 133 generates instruction information to be input to the large-scale language model by reflecting attributes corresponding to the clusters and range information corresponding to the attributes in a template of instruction information stored in the storage unit 12. FIG. 7 is a diagram showing an example of instruction information including attribute information corresponding to the first cluster. From the example shown in FIG. 7, it can be confirmed that the attributes of the first cluster having a first cluster ID of "1", "age" and "model used", are reflected in the parameters shown in the template of instruction information shown in FIG. 6, and that "young age" out of "young age" and "middle-aged and elderly age" and "latest model" out of "latest model" and "old model", which are range information corresponding to the attributes, are reflected.
[0038] Next, the second acquisition unit 133 inputs, to the large-scale language model, instruction information including attributes corresponding to each of the first clusters and range information corresponding to the attributes. Then, the second acquisition unit 133 acquires text information including one or more first parameters corresponding to each of the first clusters from the large-scale language model.
[0039] Similarly, the second acquisition unit 133 inputs instruction information including attributes corresponding to each of the multiple second clusters and range information corresponding to the attributes to the large-scale language model, and acquires text information including one or more second parameters corresponding to each of the multiple second clusters from the large-scale language model.
[0040] Fig. 8 is a diagram showing an example of text information acquired from a large-scale language model for the instruction information corresponding to Fig. 7. It can be seen from Fig. 8 that the first cluster contains values "3", "2", and "4" corresponding to the items "attitude", "subjective norm", and "possibility of behavior control", which indicate the characteristics of a person. These values are used as the first parameters.
[0041] It is desirable to specify the range of values that the parameter can take for each item of the first parameter, such as "on a 7-point scale" as shown in Figure 7. It is also desirable to specify more specifically, "on a 7-point scale (1: not applicable at all, 2: not applicable to a certain extent, 3: not applicable to a certain extent, 4: neither applicable nor unapplicable, 5: applicable to a certain extent, 6: applicable to a certain extent, 7: applicable to a certain extent)."
[0042] Next, the second acquisition unit 133 analyzes text information acquired from the large-scale language model in response to the instruction information corresponding to the first cluster, and extracts, as the first parameter, numerical values corresponding to each of the items "attitude," "subjective norm," and "possibility of behavior control" that indicate the characteristics of the person corresponding to the first cluster, contained in the text information, to acquire the first parameter. Similarly, the second acquisition unit 133 analyzes text information acquired from the large-scale language model in response to the instruction information corresponding to the second cluster, and extracts, as the second parameter, numerical values corresponding to each of the items "attitude," "subjective norm," and "possibility of behavior control" that are characteristics of the person corresponding to the second cluster, contained in the text information, and have the same index as the first parameter, to acquire the second parameter.
[0043] FIG. 9 is a diagram showing an example of a first parameter and a second parameter. FIG. 9(a) shows a first parameter acquired for the first cluster shown in FIG. 4, and FIG. 9(b) shows a second parameter acquired for the second cluster shown in FIG. 5. As shown in FIG. 9(a) and FIG. 9(b), it can be confirmed that values corresponding to the items "attitude", "subjective norm", and "possibility of behavior control" indicating a person's characteristics are acquired as the first parameter and the second parameter. Note that the items indicating a person's characteristics are "attitude", "subjective norm", and "possibility of behavior control", but are not limited thereto and may be other items indicating a person's characteristics.
[0044] The second acquisition unit 133 may generate a plurality of first parameters for each of the plurality of first clusters. The second acquisition unit 133 may generate a plurality of first parameters by changing a temperature parameter of the large-scale language model to output a plurality of parameters for the same input. In addition, when a plurality of combinations of attribute information correspond to the first cluster, the second acquisition unit 133 may generate a plurality of first parameters by causing the large-scale language model to generate a plurality of first parameters based on each combination. For example, when the first cluster is a young cluster and includes both the latest model and the old model, the second acquisition unit 133 inputs both "young", "latest model", and "young", "old model" into the large-scale language model and generates a first parameter corresponding to each combination. Such a parameter generation method can also be applied to the second parameters of each of the second clusters.
[0045] The combining unit 134 combines one or more second clusters with at least one of the multiple first clusters based on the first parameters of each of the multiple first clusters and the second parameters of each of the multiple second clusters acquired by the second acquiring unit 133. For example, the combining unit 134 calculates parameter similarity, such as cosine similarity, between the multiple first parameters of each of the multiple first clusters and the multiple second parameters of each of the multiple second clusters.
[0046] Then, the combining unit 134 combines the first cluster whose calculated similarity exceeds a predetermined threshold with the second cluster. For example, the combining unit 134 generates combined cluster information by combining the first cluster whose calculated similarity exceeds a predetermined threshold with the second cluster. Here, the combining unit 134 may combine, to one first cluster, a plurality of second clusters having second parameters whose similarity to the first parameter of the first cluster exceeds a predetermined threshold.
[0047] Note that the combining unit 134 combines the first cluster and the second cluster based on the similarity of a parameter such as cosine similarity, but is not limited to this. For example, the distance from the first cluster to the second cluster may be regarded as a transportation cost, and a combination of the first cluster and the second cluster that minimizes the transportation cost from one or more first clusters to one or more second clusters using a sinkhorn algorithm may be determined as the first cluster and the second cluster to be combined.
[0048] Fig. 10 is a diagram showing an example of combined cluster information. It can be seen from Fig. 10 that a first cluster with a first cluster ID of "1" and a second cluster with a second cluster ID of "2" are combined, and that a first cluster with a first cluster ID of "2" and a second cluster with a second cluster ID of "1" are combined.
[0049] The transmission unit 135 transmits information indicating the combination of the first cluster and the second cluster combined by the combining unit 134 to at least one of the first device 2 and the second device 3. For example, the transmission unit 135 distributes the combined cluster information shown in FIG. 10 as information indicating the combination of the first cluster and the second cluster combined by the combining unit 134 to at least one of the first device 2 and the second device 3.
[0050] The transmitting unit 135 may function as a distributing unit and distribute information corresponding to the attribute information indicated by the second cluster combined with the first cluster to a terminal of a user having the attribute information indicated by the first cluster. In this case, for example, the transmitting unit 135 selects information corresponding to the value of the second attribute information associated with the first cluster ID for a user of the first cluster ID. For example, in the example shown in FIG. 10, the transmitting unit 135 selects, as advertising information to be distributed to a user with a first cluster ID of "2", advertising information corresponding to "health and outdoors", which is attribute information corresponding to the second cluster ID "1" associated with the first cluster ID.
[0051] Then, the transmission unit 135 associates the user ID associated with the first cluster ID with the selected advertising information and transmits them to the first device 2, thereby causing the first device 2 to deliver the advertising information to a terminal used by the user of the user ID. For example, when the transmission unit 135 selects advertising information corresponding to "health and outdoor" as advertising information to be delivered to a user with a first cluster ID of "2", the transmission unit 135 transmits instruction information to the first device 2 instructing the first device 2 to deliver the advertising information to a terminal of the user with the user ID associated with the first cluster ID "2" shown in Fig. 4(a). In this way, the information processing device 1 can deliver effective information to the user.
[0052] Moreover, the transmitting unit 135 delivers information corresponding to the attribute information indicated by the second cluster combined with the first cluster to the terminal of the user having the attribute information indicated by the first cluster, but is not limited to this. For example, the transmitting unit 135 may deliver information corresponding to the attribute information indicated by the second cluster combined with the first cluster to the terminal of the user not having the attribute information indicated by the first cluster.
[0053] For example, when the generating unit 132 calculates the first cluster belonging degree for each user corresponding to the first information group, the transmitting unit 135 identifies users who have a relatively high first cluster belonging degree for the first cluster among users who do not have attribute information indicated by the first cluster. Then, the transmitting unit 135 may preferentially deliver information corresponding to the attribute information indicated by the second cluster combined with the first cluster to the terminal of the identified user who has a relatively high first cluster belonging degree.
[0054] For example, the transmission unit 135 may associate the user ID of a user having a relatively high degree of first cluster membership included in the first information group with the selected advertisement information and transmit the associated information to the first device 2, thereby causing the first device 2 to deliver the advertisement information to a terminal used by the user of the user ID. When a budgetary limit for information delivery is set, the transmission unit 135 may preferentially identify users having a high degree of first cluster membership within the budgetary limit.
[0055] [Operation flow] Next, a description will be given of the flow of processing related to the information processing device 1. Fig. 11 is a flowchart showing the flow of processing in the information processing device 1. First, the first acquisition unit 131 acquires a first information group from the first device 2, and acquires a second information group from the second device 3 (S1).
[0056] Next, the generation unit 132 clusters the attribute information based on the first attribute information included in the acquired first information group to generate multiple first clusters, and also clusters the attribute information based on the second attribute information included in the second information group to generate multiple second clusters (S2).
[0057] Next, the second acquisition unit 133 inputs attribute information corresponding to each of the first clusters to the machine learning model, and acquires text information including a first parameter that is a combining parameter for each of the first clusters. Then, the second acquisition unit 133 acquires the first parameter included in the acquired text information (S3).
[0058] Next, the second acquisition unit 133 inputs attribute information corresponding to each of the second clusters to the machine learning model, and acquires text information including second parameters that are combination parameters for each of the second clusters. Then, the second acquisition unit 133 acquires the second parameters included in the acquired text information (S4).
[0059] Next, the combining unit 134 generates combined cluster information by combining one or more second clusters with at least one of the first clusters based on the first parameters of the first clusters and the second parameters of the second clusters acquired by the second acquiring unit 133 (S5). Next, the transmitting unit 135 transmits the combined cluster information generated in S7 to at least one of the first device 2 and the second device 3 (S6).
[0060] [Variation 1] In the above embodiment, the information processing device 1 generates the first cluster and the second cluster, but this is not limited to the above. The devices that generate the first cluster and the second cluster may be the first device 2 and the second device 3, respectively. Fig. 12 is a diagram for explaining an overview of the information processing system S in the case where the first device 2 and the second device 3 generate clusters.
[0061] In the example shown in Fig. 12, the first device 2 generates a first cluster based on the first information group ((1) in Fig. 12) and transmits attribute information of the first cluster to the information processing device 1 ((2) in Fig. 12). The second device 3 generates a second cluster based on the second information group ((3) in Fig. 12) and transmits attribute information of the second cluster to the information processing device 1 ((4) in Fig. 12).
[0062] The information processing device 1 inputs attribute information corresponding to each cluster obtained from the first device 2 and the second device 3 to the machine learning model ((5) in Figure 12), and obtains parameters corresponding to each cluster from the machine learning model ((6) in Figure 12).
[0063] The information processing device 1 combines one or more second clusters with at least one of the multiple first clusters based on a first parameter of each of the multiple first clusters and a second parameter of each of the multiple second clusters ((7) in FIG. 12).
[0064] In this case, the first device 2 may hold association information between the first cluster ID and the user ID, and the second device 3 may hold association information between the second cluster ID and the user ID. Then, the transmission unit 135 of the information processing device 1 may transmit the advertisement information selected corresponding to the first cluster combined with the second cluster to the first device 2 in association with the first cluster ID of the first cluster, and the first device 2 may deliver the advertisement information to a terminal used by a user of the user ID associated with the first cluster ID. Similarly, the transmission unit 135 of the information processing device 1 may transmit the advertisement information selected corresponding to the second cluster combined with the first cluster to the second device 3 in association with the second cluster ID of the second cluster, and the second device 3 may deliver the advertisement information to a terminal used by a user of the user ID associated with the second cluster ID.
[0065] Furthermore, when the information processing device 1 generates the first cluster and the second cluster, a dedicated section for generating the first cluster and the second cluster may be provided so that the input data and the processing process cannot be confirmed by each other. In this case, the transmission unit 135 may associate the cluster ID of the first cluster with a user ID in the section corresponding to the first cluster, and distribute the advertising information selected corresponding to the first cluster to the terminal used by the user of the user ID. In this way, security and anonymity can be guaranteed when handling the first information group and the second information group.
[0066] [Variation 2] In the above embodiment, the information processing device 1 acquires each parameter by inputting text information into a large-scale language model, but the information input to the machine learning model is not limited to text information, and can be other than a large-scale language model that performs image and video processing, text analysis, and voice recognition. For example, a user's hobbies and preferences may be read from a wallpaper set by the user and reflected in the parameters. In addition, if a media image is analyzed, which is now widely used, such as predicting behavior using the user's voice information, various data can be used to acquire parameters, although the user's consent is required.
[0067] [Effects of information processing device 1] As described above, the information processing device 1 according to the present embodiment acquires a first information group including first attribute information indicating the attributes of a user collected by a first business operator and a second information group including second attribute information indicating the attributes of a user collected by a second business operator, generates a plurality of first clusters corresponding to the first information group, and generates a plurality of second clusters corresponding to the second information group. Then, the information processing device 1 inputs attribute information corresponding to each of the plurality of first clusters into a machine learning model, acquires a first parameter for each of the plurality of first clusters from the machine learning model, inputs attribute information corresponding to each of the plurality of second clusters into the machine learning model, acquires a second parameter for each of the plurality of second clusters from the machine learning model, and combines one or more second clusters with at least one of the plurality of first clusters based on the first parameter for each of the plurality of first clusters and the second parameter for each of the plurality of second clusters. In this way, the information processing device 1 can combine a plurality of pieces of anonymous user information.
[0068] Furthermore, this invention will make it possible to contribute to Goal 9 of the United Nations' Sustainable Development Goals (SDGs), which is to "build resilient infrastructure, promote industry, innovation and infrastructure."
[0069] Although the present invention has been described above using the embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes are possible within the scope of the gist of the present invention. For example, all or part of the device can be configured by distributing or integrating functionally or physically in any unit. In addition, new embodiments resulting from any combination of multiple embodiments are also included in the embodiments of the present invention. The effect of the new embodiment resulting from the combination combines the effect of the original embodiment. [Explanation of symbols]
[0070] 1. Information processing device 2 1st device 3 Second device 11 Communications Department 12 Storage section 13 Control section 131 First acquisition part 132 Generation part 133 Second Acquisition Department 134 Joint 135 Transmitter
Claims
1. a first acquisition unit that acquires a first information group including first attribute information that is attribute information indicating an attribute of a user collected by a first business operator, and a second information group including second attribute information that is attribute information indicating an attribute of a user collected by a second business operator different from the first business operator; a generation unit that performs clustering of attribute information based on first attribute information included in the first information group to generate a plurality of first clusters that are clusters corresponding to the first information group, and performs clustering of attribute information based on the second attribute information included in the second information group to generate a plurality of second clusters that are clusters corresponding to the second information group; a second acquisition unit that inputs attribute information corresponding to each of the first clusters into a machine learning model that outputs one or more parameters for combining a plurality of clusters in response to input of attribute information, the one or more parameters indicating characteristics of a person having the attribute information, and acquires first parameters that are the parameters of each of the first clusters from the machine learning model, and also inputs attribute information corresponding to each of the second clusters into the machine learning model, and acquires second parameters that are the parameters corresponding to each of the second clusters from the machine learning model; a combining unit that combines one or more of the second clusters to at least any one of the first clusters based on the first parameter of each of the first clusters and the second parameter of each of the second clusters; An information processing device having the above configuration.
2. the generation unit generates the first cluster and the second cluster associated with range information indicating a range of values of attributes grouped by the clustering; the second acquisition unit acquires the first parameters from the machine learning model by inputting the attributes corresponding to each of the first clusters and the range information corresponding to the attributes to the machine learning model, which outputs parameters for combining a plurality of clusters in response to an input of the attributes and the range information corresponding to the attributes, and acquires the second parameters from the machine learning model by inputting the attributes corresponding to each of the second clusters and the range information corresponding to the attributes to the machine learning model; The information processing device according to claim 1 .
3. A distribution unit that distributes information corresponding to the attribute information indicated by the second cluster combined with the first cluster to a terminal of a user having the attribute information indicated by the first cluster. The information processing device according to claim 1 .
4. the generation unit calculates a first cluster membership degree indicating a degree to which a user corresponding to each of a plurality of records belongs to the first cluster based on a degree of match between attribute information indicated by the generated first cluster and attribute information of each of a plurality of records included in the first information group; the distribution unit preferentially distributes information corresponding to attribute information indicated by the second cluster combined with the first cluster to a terminal of a user having a relatively high degree of first cluster membership with respect to the first cluster. The information processing device according to claim 3 .
5. The first acquisition unit acquires the first information group from a first device used by the first business operator and acquires a second information group from a second device used by the second business operator; a transmitting unit configured to transmit information indicating a combination of the first cluster and the second cluster combined by the combining unit to at least one of the first device and the second device. The information processing device according to claim 1 .
6. The computer executes acquiring a first information group including first attribute information that is attribute information indicating an attribute of a user collected by a first business operator, and a second information group including second attribute information that is attribute information indicating an attribute of a user collected by a second business operator different from the first business operator; clustering attribute information based on first attribute information included in the first information group to generate a plurality of first clusters corresponding to the first information group, and clustering attribute information based on the second attribute information included in the second information group to generate a plurality of second clusters corresponding to the second information group; inputting attribute information corresponding to each of the plurality of first clusters into a machine learning model that outputs one or more parameters for combining a plurality of clusters in response to input of attribute information, the one or more parameters indicating characteristics of a person having the attribute information, and acquiring first parameters that are the parameters of each of the plurality of first clusters from the machine learning model, and inputting attribute information corresponding to each of the plurality of second clusters into the machine learning model, and acquiring second parameters that are the parameters corresponding to each of the plurality of second clusters from the machine learning model; combining one or more of the second clusters with at least one of the first clusters based on the first parameter of each of the first clusters and the second parameter of each of the second clusters; An information processing method comprising the steps of:
Citation Information
Patent Citations
Customer information integration system and method for preparing integrated customer information database
JP2006018340A
Coordination server program, business operator server program, and data coordinated system
JP2021117679A
