JP2019070973A - Information processing system, information processing method, and program - Google Patents

Abstract

To provide new technologies related to databases.SOLUTION: A system (50) generates a database based on a first database and a second database. The first database (151) has feature data for each component in a first group, and the second database (351) has feature data for each component in a second group. The system divides a plurality of pairs of components between the first group and the second group into a plurality of clusters and provides cluster information; acquires first integrated feature data for each cluster from a first generation unit (10) that integrates the feature data of the first database for each cluster based on the cluster information; acquires second integrated feature data for each cluster from a second generation unit (30) that integrates the feature data of the second database for each cluster based on the cluster information; and generates a combined database (551) which has, for each cluster, data combining the first and second integrated feature data of the same cluster.SELECTED DRAWING: Figure 1

Description

  The present disclosure relates to an information processing system and an information processing method.

  Heretofore, it has been practiced to analyze the purchase behavior of a customer based on sales data of a product. In order to use for commercial activities, analysis of contact behavior with mass media and network content by customers is also performed.

  It is also conducted to collect various information such as purchase behavior of customers, contact behavior to mass media / network contents, lifestyle, etc. by questionnaire format or face-to-face question format.

  In recent years, each company has a huge database that includes data on such customers. However, each company is reluctant to provide data on these customers to the outside, mainly for the protection of personal information. These data, if provided externally by the company that owns it, is provided encrypted, provided that information leading to the identification of the customer has been largely deleted and provided, or intentionally (error) It is provided in the state changed so that it may contain (refer patent document 1).

JP, 2014-109647, A

  As described above, the provision of data on customers from data-holding companies is limited in terms of personal information protection. Therefore, according to the prior art, it is difficult to effectively utilize various data existing in society.

  Therefore, in one aspect of the present disclosure, it is desirable to be able to provide a new technology related to a database that can effectively utilize various data existing in society.

  An information processing system according to an aspect of the present disclosure is an information processing system that generates a new database based on the first and second databases. The first database has, for each of the members of the first group, feature data representing the first feature of the members. The second database has, for each of the members of the second group, feature data representing a second feature of the members.

  According to one aspect of the present disclosure, an information processing system includes a clustering unit, a first acquisition unit, a second acquisition unit, and a coupling unit. The clustering unit divides a plurality of construct pairs, which are a pair of constructs between the first group and the second group and in which two constructs belonging to the pair at least correspond to each other, into a plurality of clusters; It is configured to provide cluster information representing the cluster to which each of the constituent pairs belong.

  As mentioned above, one constituent pair is a pair of a first group of constituents and a second group of constituents at least corresponding to each other. The term "at least corresponding" as used herein includes "matching" in the concept. Thus, the entities of the first group of constituents and the second group of constituents belonging to one constituent pair may be a single entity. For example, each of a plurality of construct pairs may be a pair of constructs whose entities are assumed to be identical. When a construct pair is a pair of constructs having the same entity, "a plurality of construct pairs" is read as "a plurality of constructs common to the first group and the second group" It is also good.

  The first acquisition unit is configured to acquire first integrated feature data for each cluster from the first generation unit. The first generation unit integrates feature data corresponding to a plurality of constituent pairs possessed by the first database by statistical processing for each cluster based on cluster information acquired from the clustering unit, and thereby is integrated for each cluster It is configured to generate first integrated feature data as feature data.

  The second acquisition unit is configured to acquire second integrated feature data for each cluster from the second generation unit. The second generation unit integrates feature data corresponding to a plurality of construction body pairs possessed by the second database by statistical processing for each cluster based on cluster information acquired from the clustering unit, and thereby is integrated for each cluster It is configured to generate second integrated feature data as feature data.

  The combining unit is configured as a new database based on the first integrated feature data for each cluster acquired by the first acquisition unit and the second integrated feature data for each cluster acquired by the second acquisition unit. It is configured to generate a combined database having combined data for each cluster combining one integrated feature data and a second integrated feature data.

  According to this information processing system, the first integrated feature data and the second integrated feature data are meaningfully combined even if it is not possible to obtain the feature data without processing that the first database and the second database have. A meaningful combined database can be generated that corresponds to the combined database of the first and second databases.

  The feature data of the combined database generated by the information processing system is first and second combined feature data in which feature data in the first and second databases are integrated by statistical processing for each cluster. Therefore, according to the information processing system of the present disclosure, personal information can be protected. Therefore, according to one aspect of the present disclosure, it is possible to provide a new technology related to a database that can effectively utilize various data existing in society.

  According to one aspect of the present disclosure, the first and second groups of components may be consumers. In this case, the first database has, for each consumer of the first group, the feature data representing the first feature of the consumer, and the second database, for each consumer of the second group, the consumer May have feature data representing a second feature of

  According to one aspect of the present disclosure, each of the first group of constructs may be assigned a respective first identification code. The first database may store feature data of each of the first group of constituents in association with the first identification code of the constituents. A separate second identification code may be assigned to each of the second group of constructs. The second database may store the feature data of the second group of constituents in association with the second identification code of the constituents.

  According to one aspect of the present disclosure, the clustering unit identifies the plurality of constituent pairs based on the information indicating the correspondence between the first identification code and the second identification code, and determines the plurality of constituent pairs. May be divided into a plurality of clusters. The clustering unit may provide the first generation unit with cluster information representing a cluster to which each of the plurality of construction body pairs belongs in association with the first identification code. The clustering unit may provide, to the second generation unit, cluster information representing a cluster to which each of the plurality of construction body pairs belongs in association with the second identification code.

  According to one aspect of the present disclosure, the first and second databases use the identification code common to the first database and the second database to identify the feature data of each construct and the corresponding construct. It may be stored in association with the code. The clustering unit divides a plurality of constituent pairs corresponding to a pair of feature data associated with the same identification code between the first database and the second database as a plurality of constituent pairs into a plurality of clusters. The present invention may be configured to provide, as cluster information, cluster information representing a cluster to which each of a plurality of construction pairs belongs in association with an identification code as the first and second generation units.

  According to one aspect of the present disclosure, the first generation unit conceals / hashes the personal information of the plurality of constituents belonging to the first group with a specific function, and a plurality of members belonging to the first group The configuration unit may be configured to provide the clustering unit with a list including a secrecy value / hash value for each configuration.

  The second generation unit is a list of a plurality of constituents belonging to the second group, regarding each of the plurality of constituents belonging to the second group, concealing / hashing the personal information of the constituents with the specific function, , And may be configured to provide the clustering unit with a list including the concealment value / hash value for each construct.

  The clustering unit identifies a plurality of construct pairs based on the concealment value / hash value included in the list acquired from the first generation unit and the concealment value / hash value included in the list acquired from the second generation unit, The first and second generation units may be provided with cluster information representing a cluster to which each of the plurality of construct pair belongs, in association with the secret value / hash value.

  According to an aspect of the present disclosure, the clustering unit may be configured to divide the plurality of construct pairs into a plurality of clusters based on the degree of similarity between the plurality of construct pairs. According to such clustering based on the degree of similarity, feature data of similar configurations can be integrated, so that loss of valuable information can be suppressed even by integration of feature data for each cluster. A more meaningful combined database can be generated.

  According to an aspect of the present disclosure, the clustering unit may be configured to obtain similarity information that can specify the similarity of at least one of the first and second features between a plurality of constituent pairs. . In this case, the clustering unit divides the plurality of construct pairs into a plurality of clusters based on the acquired similarity information so that at least one of the first and second features is similar. May be configured.

  According to one aspect of the present disclosure, the first generation unit is a list of a plurality of constructs belonging to the first group, the list representing the similarity between the plurality of constructs related to the first feature, the clustering unit May be configured to provide. The second generation unit may be configured to provide the clustering unit with a list representing a plurality of constructs belonging to the second group and representing the similarity between the plurality of constructs relating to the second feature. Good. In this case, based on the list acquired from the first generation unit and the second generation unit, the clustering unit combines the plurality of structure pairs into a plurality of structures in which the first and second features are similar. It may be configured to be divided into clusters.

  According to one aspect of the present disclosure, the first generation unit is configured to provide the clustering unit with a list including a plurality of constituents belonging to the first group and including the first attribute value of each constituent. May be configured. The second generation unit may be configured to provide the clustering unit with a list including a plurality of constituent bodies belonging to the second group and including a second attribute value for each construction body. In this case, the clustering unit determines the degree of similarity between the plurality of structure pairs based on at least one of the first and second attribute values, and based on the determined degree of similarity, the plurality of structure pairs It may be configured to split into clusters.

  According to an aspect of the present disclosure, there is provided a computer program for causing a computer to implement at least one of a clustering unit, a first acquisition unit, a second acquisition unit, and a coupling unit included in the information processing system. May be The computer program may be stored in a computer readable non-transitory recording medium.

  According to an aspect of the present disclosure, an information processing system capable of communicating with a first external system including a first database and a second external system including a second database may be provided. The information processing system acquires a list of a plurality of constituents belonging to the first group from the first external system, and further acquires a list of a plurality of constituents belonging to the second group from the second external system. And a plurality of construction pairs each of which is a pair of constructions between the first group and the second group and at least two constructions corresponding to each other correspond to each other, based on the acquired list, into a plurality of clusters. And a clustering unit that provides cluster information representing a cluster to which each of a plurality of construct pairs belongs to the first and second external systems.

  The first external system integrates feature data corresponding to a plurality of constituent pairs possessed by the first database by statistical processing for each cluster based on the cluster information acquired from the clustering unit, and generates the second feature for each cluster generated thereby One integrated feature data may be configured to be provided to the information processing system. The information processing system may include a first acquisition unit that acquires first integrated feature data for each cluster from the first external system.

  The second external system integrates feature data corresponding to a plurality of constituent pairs possessed by the second database by statistical processing for each cluster based on the cluster information acquired from the clustering unit, and generates the second feature for each cluster generated thereby The second integrated feature data may be configured to be provided to the information processing system. The information processing system may include a second acquisition unit that acquires second integrated feature data for each cluster from the second external system.

  The information processing system performs first integrated feature data of the same cluster based on the first integrated feature data of each cluster acquired by the first acquisition unit and the second integrated characteristic data of each cluster acquired by the second acquisition unit. And the second integrated feature data may be provided with a combining unit that generates a combined database having combined data for each cluster.

  According to one aspect of the present disclosure, an information processing system may be provided that includes a second database configured to be communicable with an external system that includes the first database. The information processing system can include a clustering unit, an acquisition unit, a generation unit, and a coupling unit. The clustering unit divides a plurality of construct pairs, which are a pair of constructs between the first group and the second group and in which two constructs belonging to the pair at least correspond to each other, into a plurality of clusters; Cluster information is provided to the external system that represents the cluster to which each of the constituent pairs belong.

  The external system integrates feature data corresponding to a plurality of constituent pairs possessed by the first database by statistical processing on a cluster basis based on cluster information received from the clustering unit, thereby generating a first integrated feature for each cluster Provide data to the information processing system. The acquisition unit acquires first integrated feature data for each cluster from the external system.

  The generation unit generates second integrated feature data for each cluster by integrating feature data corresponding to a plurality of component pairs of the second database based on cluster information by statistical processing for each cluster. The combining unit is based on the first integrated feature data for each cluster acquired by the acquiring unit and the second integrated feature data for each cluster generated by the generating unit, the first integrated feature data and the second integrated feature of the same cluster A combined database is generated which has combined data combining data with each cluster.

  According to one aspect of the present disclosure, an information processing method may be provided for generating a new database based on the first and second databases. The method divides a plurality of pairs of members between the first group and the second group, in which two members belonging to the pair at least correspond to each other, into a plurality of clusters. Feature data corresponding to a plurality of construct pairs included in the first database, based on the clustering procedure providing cluster information representing the cluster to which each of the construct pairs belongs, and the cluster information provided by the clustering procedure, The first acquisition procedure for acquiring the first integrated feature data for each cluster from the device that generates the first integrated feature data as feature data integrated for each cluster by statistical processing, and the clustering procedure Based on the provided cluster information, feature data corresponding to a plurality of construction pairs possessed by the second database A second acquisition procedure for acquiring second integrated feature data for each cluster from the device for integrating the clusters by statistical processing for each cluster and thereby generating the second integrated feature data as feature data integrated for each cluster And first integrated feature data of the same cluster based on the first integrated feature data of each cluster acquired by the first acquisition procedure and the second integrated feature data of each cluster acquired by the second acquisition procedure And a combining procedure for creating a combined database having, per cluster, combined data combined with the second integrated feature data.

It is a block diagram showing composition of an information processing system of a first embodiment. It is a figure showing the composition of the 1st database and member list. It is a figure showing composition of the 2nd database and member list. It is a flowchart showing the coupling related process which a coupling device performs. It is a flowchart showing the 1st data provision process which a 1st data provision system performs. It is a flowchart showing the 2nd data provision process which a 2nd data provision system performs. It is a figure showing composition of the 1st cluster information and the 2nd cluster information. It is a flowchart showing the processing which a 1st and 2nd data provision system performs. FIG. 9A is an explanatory diagram regarding processing of the first database, and FIG. 9B is an explanatory diagram regarding processing of the second database. It is a figure showing composition of a binding database. It is a flowchart showing the member list production | generation process which the 1st data provision system of 2nd embodiment performs. It is a flowchart showing the member list production | generation process which the 2nd data provision system of 2nd embodiment performs. It is a flowchart showing the process which the coupling system of 2nd embodiment performs. It is a flowchart showing the 1st data provision process which the 1st data provision system of 3rd embodiment performs. It is a flowchart showing the 2nd data provision process which the 2nd data provision system of 3rd embodiment performs. It is a flowchart showing the process which the coupling system of 3rd embodiment performs. It is explanatory drawing regarding the related table which the coupling system of 4th embodiment has. It is a flowchart showing the coupling related processing which the coupling system of a 4th embodiment performs. It is a block diagram showing the composition of the information processing system of a fifth embodiment. It is a flowchart showing the coupling related process which the coupling system of 5th embodiment performs.

  Exemplary embodiments of the present disclosure are described below with reference to the drawings.

First Embodiment
As shown in FIG. 1, the information processing system 1 of the present embodiment includes a first data providing system 10, a second data providing system 30, and a coupling system 50. The coupling system 50 is configured to receive the first database 151 and the second database based on data on the first database 151 provided from the first data providing system 10 and data on the second database 351 provided from the second data providing system 30. It is comprised so that the joint database 551 which couple | bonded with the two database 351 may be produced | generated.

  The step of generating the combined database 551 includes the steps of processing the data in the first database 151 by the first data providing system 10 and the step of processing the data in the second database 351 by the second data providing system 30; Is included. The process of processing includes a process of integrating data on a plurality of individuals in the first database 151 and the second database 351 into data on a group.

  The coupling system 50 provides the first database 151 and the second database 351 with information (first cluster information 155 and second cluster information 355 described later) for controlling the processing. This provision enables the combination system 50 to generate a meaningful combination database 551 for consumers without receiving provision of data on individuals from the first data provision system 10 and the second data provision system 30. .

  Each of the first data providing system 10, the second data providing system 30, and the coupling system 50 is represented as a single device (machine) in FIG. 1 to simplify the description. However, the first data providing system 10, the second data providing system 30, and the coupling system 50 may each be configured with a plurality of machines.

  The first data providing system 10 includes a processor 11, a memory 13, and a storage device 15. The first data providing system 10 includes a communication interface (not shown), and is configured to be communicable with the coupling system 50 through the network NT.

  The processor (CPU) 11 executes processing in accordance with a program stored in the memory 13 or the storage device 15. The memory 13 includes a ROM, a RAM, and the like. The storage device 15 stores a first database 151.

  The first database 151 has, for each consumer belonging to the first group, feature data representing a feature of the consumer. The consumers belonging to the first group correspond to the constituents of the first group. Hereinafter, the consumers belonging to the first group will also be expressed as members of the first group.

  The configuration of the first database 151 is conceptually shown in the upper part of FIG. As can be understood from the upper part of FIG. 2, the first database 151 stores the feature data (x1, x2,...) Of each member in association with the member's customer number ID_A and connector ID_C.

  Each of the customer number ID_A and the connector ID_C is a member-specific identification code for identifying the corresponding member. However, the customer number ID_A is an identification code dedicated to the first group that is not used in the second database 351. At this point, the customer number ID_A is different from the connector ID_C commonly used in the first database 151 and the second database 351.

  The feature data represents the feature of the corresponding member by a plurality of elements x1, x2,. Examples of the elements x1, x2, ... include the member's age, gender, residence area, hobbies, and purchase experience and number of purchases for each product. When the first database 151 is managed by the specific company A, the information on the purchase experience and the number of purchases for each product included in the feature data of the first database 151 may be information on goods sold from the specific company A.

  The second data providing system 30 includes a processor 31, a memory 33, and a storage device 35. The second data providing system 30 includes a communication interface (not shown) and is configured to be communicable with the coupling system 50 through the network NT.

  The processor (CPU) 31 executes processing in accordance with a program stored in the memory 33 or the storage device 35. The memory 33 includes a ROM, a RAM, and the like. The storage device 35 stores a second database 351.

  The second database 351 has feature data representing the features of the consumer for each consumer belonging to the second group. The consumers belonging to the second group correspond to the constituents of the second group. Hereinafter, consumers belonging to the second group are also expressed as members of the second group.

  The configuration of the second database 351 is conceptually shown in the upper part of FIG. As can be understood from the upper part of FIG. 3, the second database 351 stores the feature data (y1, y2,...) Of each member in association with the member's customer number ID_B and connector ID_C.

  Each of the customer number ID_B and the connector ID_C is a member-specific identification code for identifying the corresponding member. However, the customer number ID_B is a second group dedicated identification code not used in the first database 151. The connector ID_C is an identification code commonly used with the first database 151 as described above. In the second database 351, a connector ID_C indicating the same value as that of the first database 151 is associated with the feature data of the same consumer as the first database 151.

  The feature data in the second database 351 represents the feature of the member by a plurality of elements y1, y2,. Examples of the elements y1, y2, ... include the member's age, gender, residence area, hobbies, and purchase experience and number of purchases for each product. However, at least a part of the elements y1, y2,... Is different from the elements x1, x2,. In that sense, the feature data in the second database 351 for the same consumer and the feature data in the first database 151 represent different types of features of the same consumer. The information on the purchase experience and the number of purchases for each product described as an example of the elements y1, y2, ... may be, for example, information on a product sold from a specific company B different from the specific company A.

  The connector ID_C is used to identify a consumer common to multiple databases. Specifically, it is used to associate feature data of the first database 151 and feature data of the second database 351 for the same consumer.

  According to the example shown in the upper part of FIG. 2 and the upper part of FIG. 3, the consumer of the feature data associated with A0003, A0004, A0005, A0006, A0007 as the customer number ID_A in the first database 151 is in the second database 351. It is the same as the consumer of the feature data associated with B0001, B0002, B0003, B0004, B0005 as customer number ID_B.

  The connector ID_C may be, for example, a customer number of a third group having more members than the first and second groups. The third group may be a group to which many consumers are members. The connector ID_C may be an identification code of a communication device such as a smartphone owned by a consumer, or may be an identification code of a consumer (for example, Cookie etc.) used for consumer tracking on the network. .

  The connector ID_C is advantageously a consumer identification code that can be commonly obtained without the cooperation of the administrator of the first database 151 and the administrator of the second database 351. The connector ID_C may be a value obtained by concealing these identification codes, specifically by hashing. Hashing can be performed using the same hash function in the first data providing system 10 and the second data providing system 30 so that the value of the connector ID_C of the same consumer is the same. In addition, the members of the first group and some of the members of the second group may not have the connector ID_C. In this case, information indicating that the connector is unknown is associated with the feature data of the corresponding member in the first database 151 and the second database 351.

  The coupling system 50 includes a processor 51, a memory 53, and a storage device 55. The coupling system 50 includes a communication interface (not shown), and is configured to be communicable with the first data providing system 10 and the second data providing system 30 through the network NT.

  The processor (CPU) 51 executes processing in accordance with a program stored in the memory 53 or the storage device 55. The memory 53 includes a ROM, a RAM, and the like. The storage device 55 stores a combined database 551 generated by the process executed by the processor 51.

  The processor 51 of the coupling system 50 receives the instruction to generate the coupling database 551 based on the first database 151 and the second database 351 from the user of the coupling system 50 through a user interface (not shown), as shown in FIG. Start processing

  When the connection related process is started, the processor 51 transmits a request signal for requesting a member list to the first data providing system 10 and the second data providing system 30 through the network NT (S110).

  When this request signal is received, the processor 11 of the first data providing system 10 starts the first data providing process shown in FIG. 5 and a member list listing members of the first group having feature data in the first database 151. 153 is generated (S310), and the generated member list 153 is transmitted to the coupling system 50 through the network NT (S320).

  Specifically, as illustrated in the lower part of FIG. 2, the processor 11 generates a member list 153 by representing the members of the first group by the connector ID_C. When generating the member list 153, the connector ID_C may be hashed. Here, the members of the first group whose connector ID_C is unknown are not listed in the member list 153. In the present embodiment, the feature data of the member whose connector ID_C is unknown is not used to generate the combined database 551. In other words, in the generation process of the combined database 551, the feature data of the member whose connector ID_C is unknown is regarded as nonexistent.

  Similarly, when the processor 31 of the second data providing system 30 receives the request signal, it starts the second data providing process shown in FIG. 6 and enumerates members of the second group having feature data in the second database 351. The generated member list 353 is generated (S410), and the generated member list 353 is transmitted to the coupling system 50 through the network NT (S420). Specifically, the processor 31 generates a member list 353 by representing the members of the second group with a connector ID_C as shown in the lower part of FIG. When generating the member list 353, the connector ID_C may be hashed using the same hash function as the first data providing system 10.

  The processor 51 of the coupling system 50 receives the member list 153 of the first group and the member list 353 of the second group thus transmitted from the first data providing system 10 and the second data providing system 30 (S120). The clustering process (S130) is executed based on the received member list 153, 353.

  In the clustering process (S130), the processor 51 specifies a member common to the first group and the second group (hereinafter, referred to as "common member") from the member list 153, 353 of the first group and the second group. (S131). The identification can be realized by collating the connector ID_C between the member lists 153 and 353.

  Thereafter, the processor 51 divides the common member into a plurality of clusters (S133). For example, the processor 51 can divide the common members into clusters of a predetermined number of members randomly or with a predetermined rule. For example, when the number of common members is M and the predetermined number of members per cluster is K, clusters can be generated by the number corresponding to the quotient α when dividing the value M by the value K. The value K is set to a value larger than 1 in terms of personal information protection. When there is one or more remainders β, members corresponding to the remainders β may be allocated randomly or according to a predetermined rule to any number of clusters corresponding to the quotient α.

  Furthermore, the processor 51 divides the members of the first group that are not common members specified from the member list 153 of the first group, that is, the non-common members of the first group into a plurality of clusters (S135). The clustering in S135 can be performed by the same method as the clustering in S133.

  Further, the processor 51 divides the members of the second group which are not common members specified from the member list 353 of the second group, that is, the non-common members of the second group into a plurality of clusters (S137). The clustering in S137 can be performed by the same method as the clustering in S135.

  The processor 51 divides the common member, the non-common member of the first group, and the non-common member of the second group into a plurality of clusters in this manner, and then the first cluster information 155 and the second cluster information 355. It generates (S140), transmits the first cluster information 155 to the first data providing system 10, and transmits the second cluster information 355 to the second data providing system 30 (S140).

  As shown in FIG. 7, the first cluster information 155 is generated by adding a cluster number, which is an identification code of a cluster to which each member belongs, to the member list 153 received from the first data providing system 10. The second cluster information 355 is generated by adding the cluster number of the cluster to which each member belongs to the member list 353 received from the second data providing system 30.

  The processor 11 of the first data providing system 10, upon receiving the first cluster information 155 (S330), executes processing on the first database 151 (S340). Specifically, the processor 11 executes the processing shown in FIG.

  That is, the processor 11 selects one of the clusters based on the received first cluster information 155 (S510), and integrates, by statistical processing, a plurality of feature data in the first database 151 of the members belonging to the selected cluster. Then, one feature data (hereinafter referred to as "integrated feature data") corresponding to the selected cluster is generated (S520).

  The processor 11 repeatedly generates the integrated feature data for each cluster by repeatedly executing the processes of S510 and S520 until determining that the integrated feature data is generated for all the clusters. Then, when integrated feature data of all clusters are generated (Yes in S530), the processing (S340) is ended, and the processed first database 157 including integrated feature data for each cluster is transmitted to the coupling system 50 (S350) ).

  The post-processing first database 157 is a database having integrated feature data for each cluster, which is data after statistical processing for the feature data, instead of the feature data for each member included in the first database 151. FIG. 9A partially shows the configuration of the first processed database 157.

  In S520, the processor 11 converts the values of the elements x1, x2, ... indicated by the feature data of the plurality of members belonging to the selected cluster into one statistical value for each element, and selects one corresponding to the cluster Generate integrated feature data.

  According to the example of the first cluster information 155 shown in FIG. 7, the feature data of the customer numbers A0003 and A0004 belong to the same cluster (cluster number 002). Therefore, regarding this cluster, in S520, the feature data of A0003 and the feature data of A0004 are converted into statistical values for each element x1, x2, ..., and integrated features including statistical values for each element x1, x2, ... Data is generated.

  As shown in FIG. 9A, with respect to the element x1, the value x1 [3] of A0003 and the value x1 [4] of A0004 are statistically processed, and the statistical value ST {x1 [3], x1 [4] is obtained. } Is described as the value of the element x1 in the integrated feature data of the cluster number 002. Here, ST {} indicates that it is a statistical value of values in parentheses {}. The statistical value may be an average value, a median value, a mode value, a maximum value and / or a minimum value, or a composition ratio. May be The type of statistical value is determined in advance for each type of elements x1, x2,.

  For example, when the element x1 represents an age, the statistical value ST {x1 [3], x1 [4]} may be an average value of the age x1 [3] and the age x1 [4]. When the element x1 represents a gender, the statistic ST {x1 [3], x1 [4]} may be a male and / or female ratio. When the element x1 represents the purchase experience of a product, ST {x1 [3], x1 [4]} may be a ratio of members who have the purchase experience of the product. When the element x1 represents the number of purchases of a product, ST {x1 [3], x1 [4]} may be one or a combination of an average, a median, and a maximum of the number of purchases of a product.

  Besides, according to the example of the first cluster information 155 shown in FIG. 7, the feature data of the customer numbers A0005, A0006 and A0007 belong to the same cluster (cluster number 003). Therefore, regarding this cluster, in S520 of processing, the feature data of A0005, the feature data of A0006, and the feature data of A0007 are converted into statistical values for each element x1, x2, ..., and integrated feature data is generated. Be done.

  As shown in FIG. 9A, the integrated feature data additionally has the value of the variable k. The variable k represents the number of members belonging to a cluster, and is described in integrated feature data when generating integrated feature data.

  The second data providing system 30 also executes the same process as the first data providing system 10 upon receiving the second cluster information 355 from the combined system 50. That is, when the processor 31 of the second data providing system 30 receives the second cluster information 355 (S430), the processor 31 executes processing on the second database 351 (S440). Here, the processing performed by the processor 31 is as shown in FIG.

  That is, the processor 31 integrates feature data included in the second database 351 for each cluster based on the second cluster information 355, and generates integrated feature data for each cluster. Specifically, the processor 31 converts the values of the elements y1, y2,... Indicated by the feature data of a plurality of members belonging to the corresponding cluster into one statistical value for each element, and To generate one integrated feature data corresponding to. Thereby, the processor 31 generates a second processed database 357 including integrated feature data for each cluster based on the second database 351. Thereafter, the processor 31 transmits the processed second database 357 to the coupling system 50 (S450).

  According to the example of the second cluster information 355 shown in FIG. 7, the feature data of the customer numbers B0001 and B0002 belong to the same cluster (cluster number 002). Therefore, regarding this cluster, in S520 of processing, the feature data of B0001 and the feature data of B0002 are converted into statistical values for each element y1, y2, ..., and the statistical values for each element y1, y2, ... Integrated feature data to be included is generated.

  An example of integrated feature data included in the second processed database 357 is shown in FIG. 9B. As shown in FIG. 9B, with respect to the element y1, the value y1 [1] of B0001 and the value y1 [2] of B0002 are statistically processed, and the statistical value ST {y1 [1], y1 [2] is obtained. } Is described as the value of the element y1 in the integrated feature data of the cluster number 002. The type of statistical value is determined in advance for each type of elements y1, y2,.

  The processor 51 of the coupling system 50 receives the processed first database 157 from the first data providing system 10 (S150), and further receives the processed second database 357 from the second data providing system 30 (S160). A combined database 551 is generated by combining the post-processing first database 157 and the post-processing second database 357, and the combined database 551 is stored in the storage device 55 (S170). Thereafter, the process shown in FIG. 4 is ended.

  The configuration of the combined database 551 is conceptually shown in FIG. In S170, the processor 51 of the coupling system 50 includes the first processed database 157 so as to couple integrated feature data of the same cluster between the first processed database 157 and the second processed database 357. A combined database 551 is generated by combining the integrated feature data for each cluster and the integrated feature data for each cluster included in the second processed database 357.

  The combined database 551 includes combined data in which the first integrated feature data and the second integrated feature data are combined for each cluster. Here, the first integrated feature data of one cluster is integrated feature data of the corresponding cluster of the processed first database 157, and the second integrated feature data is an integrated of the corresponding cluster of the second processed database 357. It is feature data.

  Although the information processing system 1 according to the present embodiment has been described above, according to the information processing system 1, the combination system 50 does not obtain individual feature data of the first database 151 and the second database 351. The integrated feature data for each cluster based on the first database 151 and the integrated feature data for each cluster based on the second database 351 are meaningfully combined to correspond to the combined database of the first database 151 and the second database 351. A meaningful combined database 551 can be generated.

  Converting individual feature data into cluster feature data (integrated feature data) helps protect personal information. Therefore, according to the present technology, it is possible to receive data provision relatively easily even from a database administrator who is reluctant to provide data from the viewpoint of personal information protection. Therefore, according to the present embodiment, it is possible to generate meaningful combined database 551 by meaningfully combining data on consumers scattered in society while protecting personal information. The combined database 551 can be retrieved from, for example, the storage device 55 and used for analysis of consumer behavior or the like.

  In addition, in the present embodiment, the coupling system 50 performs processing of the first database 151 in the first data providing system 10 and processing of the second database 351 in the second data providing system 30 as the first and second clusters. It controls by provision of the information 155 and 355. By this control, the first data provision system 10 and the second data provision system 30 can receive provision of integrated feature data for each common cluster with respect to the common members, and the combination system 50 can be provided for each same cluster. Integration feature data can be meaningfully combined.

  As a comparative example, consider an example in which the members of the first group and the members of the second group are separately clustered in the first data providing system and the second data providing system. In this case, the cluster group corresponding to the common member is different between the first data providing system and the second data providing system. Therefore, the combined system of the comparative example combines the combined feature data for each cluster of the first group with different members in the cluster and the combined feature data for each cluster of the second group to generate a combined database .

  As can be understood from this comparative example, according to the present embodiment, while processing the first database 151 and the second database 351, generating a more meaningful and valuable combined database 551 than the combined database of the comparative example. Can.

  In addition, in the comparative example, a state in which a member belonging to both of two clusters corresponding to two integrated feature data coupled to each other is unique may also occur. The possibility that such a situation occurs increases when connecting three or more databases. On the other hand, in the present embodiment, even if a large number of databases are connected, the above-mentioned state does not occur because clusters corresponding to a plurality of integrated feature data to be connected are configured by the same member. Therefore, according to the present embodiment, it is possible to provide a better combined database 551 also from the viewpoint of personal information protection.

Second Embodiment
Subsequently, an information processing system 1 of the second embodiment will be described. In the information processing system 1 of the second embodiment, the coupling system 50 uses the member list with distance information provided from the first data providing system 10 and the second data providing system 30 to generate a common member, a first group The information processing system 1 is different from the information processing system 1 of the first embodiment in that the non-common members of the second group and the non-common members of the second group are clustered. On the other hand, the information processing system 1 according to the second embodiment is configured in the same manner as the information processing system 1 according to the first embodiment in many other respects.

  Therefore, in the following, regarding the configuration of the information processing system 1 of the second embodiment, a configuration different from the information processing system 1 of the first embodiment will be selectively described, and the same configuration as the information processing system 1 of the first embodiment Description of the part is omitted. The configuration given the same reference numerals as the information processing system 1 of the first embodiment may be understood to be the same configuration as the information processing system 1 of the first embodiment unless there is an additional description.

  In the present embodiment, the processor 11 of the first data provision system 10 executes a member list generation process shown in FIG. 11 in S310 of the first data provision process (FIG. 5).

  In this member list generation process, the processor 11 calculates the distance D1 between members with respect to the members of the first group corresponding to the feature data attached with the connector ID_C in the first database 151 (S311).

  The distance D1 means the distance between members on the feature space corresponding to the elements x1, x2,. The distance D1 corresponds to the similarity between members for the features defined by the elements x1, x2,. It can be understood that the smaller the distance D1, the more similar the corresponding members are. The distance D1 may be, for example, a Euclidean distance. For example, the distance D1 [i, j] between the member i and the member j can be calculated according to the following equation. This equation is also shown in FIG.

D1 [i, j] = {Σ (xn [i] −xn [j]) ² } ^1/2
Here, the variable n takes a value from the value 1 to the value N corresponding to the number of elements of the feature data. xn in the case of n = 1 means the element x1, and xn in the case of n = 2 means the element x2. It may be understood that xn [i] and xn [j] are the values of element xn in the feature data of member i and member j, respectively. Here, the values of the element xn may be understood to be all expressed numerically. Σ (xn [i] −xn [j]) ² may be understood as the sum of (xn [i] −xn [j]) ² from n = 1 to n = N. In S311, the distance D1 [i, j] is calculated for all combinations of member i and member j.

  Thereafter, as in the first embodiment, the processor 11 is a member list 1531 in which the members of the first group are expressed by the connector ID_C, and the member list with distance information to which the information of the distance D1 calculated in S311 is added. 1531 is generated (S312). According to the example shown in FIG. 11, in the member list 1531, the distance D 1 [i, j] for each combination with an arbitrary member i is associated with the connector of the member j. In S320 (FIG. 5), the member list 1531 with distance information generated as described above is transmitted to the coupling system 50.

  Similarly, the processor 31 of the second data provision system 30 executes a member list generation process shown in FIG. 12 in S410 of the second data provision process (FIG. 6).

  In the member list generation process, the processor 31 calculates the distance D2 between members with respect to the members of the second group corresponding to the feature data attached with the connector ID_C in the second database 351 (S411).

  The distance D2 means the distance between members on the feature space corresponding to the elements y1, y2,. The distance D2 corresponds to the similarity between members for the features defined by the elements y1, y2,. It can be understood that the smaller the distance D2, the more similar the corresponding members are. The distance D2 may be, for example, a Euclidean distance. For example, the distance D2 [i, j] between member i and member j can be calculated according to the following equation. This equation is also shown in FIG.

D2 [i, j] = {Σ (yn [i] −yn [j]) ² } ^1/2
Here, the variable n takes a value from the value 1 to the value N corresponding to the number of elements of the feature data. yn [i] and yn [j] may be understood to be the value of element yn in the feature data of member i and member j, respectively. Here, the value of the element yn may be understood as being all expressed numerically. Σ (yn [i] −yn [j]) ² may be understood to be the sum of (yn [i] −yn [j]) ² from n = 1 to n = N. In S411, the distance D2 [i, j] is calculated for all combinations of member i and member j.

  Thereafter, as in the first embodiment, the processor 31 is a member list 3531 in which the members of the second group are expressed by the connector ID_C, and the member list with distance information to which the information of the distance D2 calculated in S411 is added. Generate 3531. According to the example shown in FIG. 12, in the member list 3531, the distance D2 [i, j] for each combination with an arbitrary member i is associated with the connector of the member j. In S420 (FIG. 6), the member list 3531 with distance information generated in this way is transmitted to the coupling system 50.

  The processor 51 of the coupling system 50 performs each of S133, S135, and S137 (see FIG. 4) based on the member list 1531 received from the first data providing system 10 and the member list 3531 received from the second data providing system 30. The process shown in FIG. 13 is performed.

That is, in S133, the processor 51 calculates the distance D between members for the common member as the combined distance D = (D1 ² + D2 ² ) ^1/2 of the distance D1 indicated by the member list 1531 and the distance D2 indicated by the member list 3531. (S610). This distance D corresponds to the similarity between members of the feature defined by the elements x1, x2, ..., y1, y2. The distance D = D [i, j] between member i and member j is calculated according to the equation D [i, j] = (D1 [i, j] ² + D 2 [i, j] ² ) ^1/2 be able to.

  Although the absolute position of each member on the feature space corresponding to the elements x1, x2,..., Y1, y2 can not be specified by the calculation of the distance D, the relative position between the members, ie, the distribution of the members It can be identified.

  Based on the distribution of common members on the feature space specified from the distance D, the processor 51 combines the members having similar distances by a predetermined number, ie, members having similar distances. Are divided into a plurality of clusters (S620). Each cluster is composed of the predetermined number of members. This clustering can be performed using the known k-means method (k-means method). Clustering may be performed using other known techniques.

Similarly, in S135, the processor 51 calculates the distance D = (D1 ² + D2 ² ) ^1/2 between the members with respect to the non-common member of the first group based on the distance D1 indicated by the member list 1531 (S610) . Here, since there is no information corresponding to the distance D2 regarding the non-common members of the first group, the distance D can be calculated by handling D2 = 0. In other words, in S610, the distance D = D [i, j] between the member i and the member j can be set to D = D1 [i, j].

  Based on the distribution of the non-common members of the first group on the feature space identified from the distance D, the processor 51 puts together the members having the similar distance D by a predetermined number, and the non-common members of the first group are separated. The common member is divided into a plurality of clusters (S620).

Similarly, in S137, the processor 51 calculates the distance D = (D1 ² + D2 ² ) ^1/2 between the members with respect to the non-common member of the second group based on the distance D2 indicated by the member list 3531 (S610) . Here, the distance D can be calculated by handling D1 = 0. In other words, in S610, the distance D = D [i, j] between the member i and the member j can be set to D = D2 [i, j].

  Based on the distribution of the non-common members of the second group on the feature space specified from the distance D, the processor 51 puts together the members having the similar distance D by a predetermined number, and The common member is divided into a plurality of clusters (S620).

  When the processor 51 executes the above processing in the clustering processing (S130), it transmits the first cluster information 155 and the second cluster information 355 based on the processing result to the first data providing system 10 and the second data providing system 30, respectively. (S140). Thereafter, processing similar to that of the first embodiment is performed.

  According to the present embodiment, based on the information of the distance D, the combination system 50 has the common member, the non-common member of the first group, and the second non-common member, the distance D is close (ie, the features are similar) ) Cluster as you group members. Therefore, compared with the first embodiment in which clustering is performed without considering the distance D, a more meaningful combined database 551 can be generated. That is, a plurality of feature data can be integrated so that meaningful consumer information is not lost by statistical processing, and the combined database 551 can be generated based on the integrated feature data.

  The information processing system 1 according to the second embodiment has been described above, but the distance D1 may not be calculated using all of the elements x1, x2,... Of the feature data included in the first database 151. , And may be calculated using some elements. Similarly, the distance D2 may be calculated using some of the elements y1, y2,... Of the feature data. Calculation of the distances D1 and D2 using only some elements is significant for the purpose of strengthening personal information protection.

Third Embodiment
Subsequently, an information processing system 1 of the third embodiment will be described. The information processing system 1 of the third embodiment is configured to perform clustering of members based on distance information, as in the second embodiment.

  In the following, regarding the configuration of the information processing system 1 of the third embodiment, a configuration different from the information processing system 1 of the first embodiment will be selectively described, and related to the same components as the information processing system 1 of the first embodiment. I omit explanation. The configuration given the same reference numerals as the information processing system 1 of the first embodiment may be understood to be the same configuration as the information processing system 1 of the first embodiment unless there is an additional description.

  In the present embodiment, when the processor 11 of the first data providing system 10 receives the request signal for the member list from the coupling system 50, it executes the first data providing process shown in FIG. 14 instead of the process shown in FIG. Do. In the first data provision processing, the processor 11 calculates the attribute value Z1 of each member for the members of the first group corresponding to the feature data attached with the connector ID_C in the first database 151 (S710). The right area in FIG. 14 and Z1 [i] shown below mean the attribute value Z1 for the member i.

  The attribute value Z1 [i] of the member i is, for example, a combination of basic attributes (for example, demographic attributes) such as the age, sex, residence area, and occupation of the member i into numerical values corresponding to the combinations. is there. For example, the attribute value Z1 [i] of member i may be a hash value calculated when a combination of basic attributes of member i is input to the hash function.

  However, the attribute value Z1 may be one obtained by hashing only one of the basic attributes. The attribute value Z1 may be, for example, hashing only the residence area of the member. Hashing is performed to make specific information on basic attributes of members secret from the outside of the first data providing system 10.

  The information on the basic attribute may be extracted from the feature data in the first database 151 or may be acquired from a database storing member information of a first group different from the first database 151 stored in the storage device 15. May be

  After calculating the attribute value Z1 of each member in S710, the processor 11 is the member list 1532 representing the members of the first group by the connector ID_C, and the member list 1532 to which the attribute value Z1 calculated in S710 is added Generate (S720). According to the example shown in the right area of FIG. 14, the member list 1532 is generated by associating the connector of member i with the attribute value Z1 [i] regarding member i.

  Furthermore, the processor 11 creates a conversion table 1533 for converting the combination (Z1_p, Z1_q) of the attribute values Z1 into the distance D1 (S730). The conversion table 1533 is a table in which, for each combination (Z1_p, Z1_q) of the attribute values Z1, a distance D1 corresponding to the combination (Z1_p, Z1_q) is described as shown in the right area of FIG.

  According to the conversion table 1533, the distance D1 between the member i and the member j when the attribute value Z [i] of the member i is the value Z1_p and the attribute value Z [j] of the member j is the value Z1_q Can be specified as the distance D1 associated with the corresponding combination (Z1_p, Z1_q) in the conversion table 1533. The distance D1 between the member i and the member j may be understood as the distance between the members on the feature space when the member i and the member j are arranged on the feature space defined by the basic attribute. This distance may be Euclidean distance as in the second embodiment.

  The processor 11 corresponds, for each combination (Z1_p, Z1_q) of the attribute value Z1, a point on the feature space corresponding to the combination of basic attributes corresponding to the attribute value Z1_p and a combination of basic attributes corresponding to the attribute value Z1_q. The conversion table 1533 can be generated by calculating the distance to a point on the feature space as the distance D1.

  The processor 11 transmits the generated member list 1532 and the conversion table 1533 to the coupling system 50 (S740). Thereafter, as in the first embodiment, the processor 11 receives the first cluster information 155 (S330), processes the first database 151 based on the first cluster information 155, and processes the first database 157 is generated (S340), the processed first database 157 is transmitted to the coupling system 50 (S350), and the first data provision processing is ended.

  Similarly, when the processor 31 of the second data providing system 30 receives a request signal for the member list from the coupling system 50, the processor 31 executes the second data providing process shown in FIG. 15 instead of the process shown in FIG. In the second data provision process, the processor 31 calculates the attribute value Z2 of each member for the members of the second group, as in S710 (S810). The right area in FIG. 15 and Z2 [i] shown below mean the attribute value Z2 for the member i.

  The attribute value Z2 [i] of the member i is obtained by encoding the combination of the basic attributes of the member i into a numerical value (for example, a hash value) corresponding to the combination, similarly to the above-described attribute value Z1. The information on the basic attribute may be extracted from feature data in the second database 351, or may be acquired from a database storing member information of a second group different from the second database 351 stored in the storage device 35. May be

  After calculating the attribute value Z2 of each member in S810, the processor 31 is a member list 3532 in which members of the second group are expressed by the connector ID_C, and the member list 3532 to which the attribute value Z2 calculated in S810 is added is used. Generate (S820). An example of the member list 3532 is shown in the right area of FIG.

  Furthermore, the processor 31 creates a conversion table 3533 for converting the combination (Z2_p, Z2_q) of the attribute values Z2 into the distance D2 (S830). The conversion table 3533 is a table in which, for each combination (Z2_p, Z2_q) of the attribute value Z2, a distance D2 corresponding to the combination (Z2_p, Z2_q) is described as shown in the right area of FIG.

  The processor 31 corresponds, for each combination (Z2_p, Z2_q) of the attribute value Z2, to a point on the feature space corresponding to the combination of basic attributes corresponding to the attribute value Z2_p and a combination of basic attributes corresponding to the attribute value Z2_q. The conversion table 3533 can be generated by calculating the distance between points on the feature space as the distance D2.

  The processor 31 transmits the generated member list 3532 and the conversion table 3533 to the coupling system 50 (S 840). Thereafter, as in the first embodiment, the processor 31 receives the second cluster information 355 (S430), processes the second database 351 based on the second cluster information 355, and processes the second database 352 after processing. 357 is generated (S440), the processed second database 357 is transmitted to the coupling system 50, and the second data provision processing is ended.

  Based on the member list 1532 and the conversion table 1533 received from the first data providing system 10, and the member list 3532 and the conversion table 3533 received from the second data providing system 30, the processor 51 of the coupling system 50 performs S133, S135, The process shown in FIG. 16 can be executed in each of S137 (see FIG. 4).

That is, in S133, the processor 51 calculates the distance D1 between members with respect to the common member based on the member list 1532 and the conversion table 1533 (S910). Further, the distance D2 between members is calculated based on the member list 3532 and the conversion table 3533 (S920). Then, based on the calculated distances D1 and D2, the distance D between members = (D1 ² + D2 ² ) ^1/2 is calculated (S930).

  After that, the processor 51 determines a predetermined number of members having similar distances D, in other words, similar characteristics, based on the distribution of common members on the feature space specified from the distance D, similar to the processing in S620. The common members are divided into a plurality of clusters (S940).

  Similarly in S135, the processor 51 divides the non-common members of the first group into a plurality of clusters such that the members having the similar distance D are grouped by a predetermined number in the procedure according to FIG. Similarly, in S137, the non-common members of the second group are divided into a plurality of clusters such that the members having the similar distance D are grouped by a predetermined number. The handling of the distances D1 and D2 that can not be specified in S135 and S137 is the same as that in the second embodiment.

  The information processing system 1 according to the third embodiment has been described above, but in the third embodiment as well, clustering is performed based on the distance between members in the feature space as in the second embodiment. 551 can be generated.

Fourth Embodiment
Subsequently, an information processing system 1 of the fourth embodiment will be described. In the following, regarding the configuration of the information processing system 1 of the fourth embodiment, a configuration different from the information processing system 1 of the first embodiment will be selectively described, and related to the same components as the information processing system 1 of the first embodiment. I omit explanation. The configuration given the same reference numerals as the information processing system 1 of the first embodiment may be understood to be the same configuration as the information processing system 1 of the first embodiment unless there is an additional description.

  In the information processing system 1 of the present embodiment, the coupling system 50 indicates the relationship between the customer number ID_A used in the first database 151 and the customer number ID_B used in the second database 351 in the storage apparatus 55. The table 553 is stored. The upper part of FIG. 17 conceptually shows a relation table 553.

  That is, the relation table 553 has information relating the customer number ID_A of the common member and the customer number ID_B for each of the common members belonging to both the first group and the second group. Although the relationship table 553 shown in the upper part of FIG. 17 further includes information of the connector ID_C of each member, this information is optional and may not be necessary.

  The relation table 553 may be generated based on information provided in advance from the administrator of the first database 151 and the second database 351. The relationship table 553 may be generated based on information provided by another person. For example, a company tracking user behavior on a network such as access to a web page may obtain customer numbers ID_A, ID_B of multiple companies related to the same user by the tracking. The relationship table 553 can be generated based on information available from such a company. The relationship table 553 may be generated based on information obtained by the coupling system 50 tracking user behavior on the network.

  In the present embodiment, the processor 51 of the coupling system 50 executes coupling-related processing shown in FIG. 18 in place of the processing shown in FIG. 4 based on an instruction from the user.

  In the connection related processing, the processor 51 does not request and receive a member list, and performs clustering processing corresponding to S130 in S1010. In this clustering process (S1010), a common member is specified with reference to the relationship table 553 (S131), the common member is divided into a plurality of clusters (S133), and a plurality of non-common members of the first group are further divided. (S135), and the non-common member of the second group is divided into a plurality of clusters (S137).

  After that, the processor 51 generates the first cluster information 1554 and the second cluster information 3554 in S1020, and transmits the first cluster information 1554 to the first data providing system 10, and the second data providing system 30 to the second data providing system 30. Two cluster information 3554 is transmitted.

  The first cluster information 1554, as shown in the lower left area of FIG. 17, represents the members of the first group in the member list of the first group, which is represented by the customer number ID_A of the first group used by the first database 151. Is generated with the cluster number of the cluster to which. The second cluster information 3554 is, as shown in the lower right area of FIG. 17, each member of the second group, represented by the second group customer number ID_B used by the second database 351, in each member list of the second group. It is generated with the cluster number of the cluster to which the member belongs.

  In the subsequent S1030, the processor 51 receives the processed first database 157 from the first data providing system 10, and further receives the processed second database 357 from the second data providing system 30 (S1040). A combined database 551 is generated by combining the back first database 157 and the processed second database 357, and the combined database 551 is stored in the storage device 55 (S1050). Thereafter, the process shown in FIG. 17 is ended.

  The processor 11 of the first data providing system 10 does not execute the processing of S310 and S320 shown in FIG. 5 and receives the first cluster information 1554 from the coupling system 50 (S330), the customer possessed by the first cluster information 1554 The processed first database 157 can be generated and transmitted based on the association information between the number ID_A and the cluster number (S340, S350).

  When the processor 31 of the second data providing system 30 receives the second cluster information 3554 from the coupling system 50 without executing the processes of S410 and S420 shown in FIG. 6 (S430), the customer included in the second cluster information 3554 The processed second database 357 can be generated and transmitted based on the association information between the number ID_B and the cluster number (S440, S450).

  As mentioned above, although the information processing system 1 of 4th embodiment was demonstrated, the information processing system 1 of this embodiment also has an effect similar to 1st embodiment.

Fifth Embodiment
Subsequently, an information processing system 5 of the fifth embodiment will be described. In the following, regarding the configuration of the information processing system 5 of the fifth embodiment, a configuration different from the information processing system 1 of the first embodiment will be selectively described, and related to the same components as the information processing system 1 of the first embodiment. I omit explanation. The configuration given the same reference numerals as the information processing system 1 of the first embodiment may be understood to be the same configuration as the information processing system 1 of the first embodiment unless there is an additional description.

  As shown in FIG. 19, the information processing system 5 of this embodiment has a configuration in which the function corresponding to the second data providing system 30 of the first embodiment is incorporated in the coupling system 80. Specifically, the information processing system 5 includes a data providing system 70 corresponding to the first data providing system 10 of the first embodiment, and a combination corresponding to the second data providing system 30 and the coupling system 50 of the first embodiment. And a system 80.

  The data providing system 70 includes a processor 71, a memory 73, and a storage device 75. The storage device 75 stores the first database 151. The data providing system 70 may be understood to have the same configuration as the first data providing system 10 of the first embodiment. It may be understood that the processing executed by the processor 71 is basically the same as the processing executed by the processor 11 of the first embodiment.

  The coupling system 80 is configured to be communicable with the data providing system 70 through the network NT. The coupling system 80 includes a processor 81, a memory 83, and a storage device 85. The storage device 85 stores a second database 351. The storage apparatus 85 stores the coupled database 551 when the coupled database 551 is generated by the process executed by the processor 81.

  When an instruction to generate the combined database 551 is input from the user, the processor 81 executes the connection-related process shown in FIG. 20 instead of the process shown in FIG.

  In this connection-related process, the processor 81 transmits a request signal for requesting a member list to the data providing system 70 through the network NT (S1110), and acquires the member list 153 of the first group from the data providing system 70 (S1110) S1120).

  Thereafter, clustering processing is executed based on the acquired member list 153 (S1130). In the clustering process (S1130), the processor 81 collates the member list 153 with the second database 351, and identifies a common member commonly existing in the first group and the second group. Furthermore, among members represented in the member list 153 of the first group, members other than common members are specified as non-common members of the first group. In addition, among members of the second group having feature data in the second database 351, members other than common members are specified as non-common members of the second group.

  The processor 81 determines the plurality of common members, the non-common members of the first group, and the non-common members of the second group in the same manner as the processing of S133, S135, and S137 in the first embodiment. Split into clusters.

  The processor 81 transmits the first cluster information 155 to the data providing system 70 when the clustering processing in S1130 is finished (S1140), and provides the processed first database 157 based on the first cluster information 155 as data. It acquires from the system 70 (S1150).

  Furthermore, the processor 81 integrates feature data possessed by the second database 351 in the storage apparatus 85 by statistical processing for each cluster based on the result of the clustering process in S1130 to generate integrated feature data for each cluster. . Thereby, the processor 81 generates a second processed database 357 including integrated feature data for each cluster based on the second database 351 (S1160).

  Thereafter, the processor 81 generates a combined database 551 by combining the first processed database 157 obtained from the data providing system 70 and the second processed database 357 generated by itself, thereby generating a combined database 551. Are stored in the storage device 85 (S1170). Thereafter, the process shown in FIG. 20 is ended.

  The combined database 551 generated by the combined system 80 may be used by a company on the side of the combined system 80 to analyze consumer behavior and to determine an advertisement delivery target, or may be provided to a company on the first database 151 side. Similarly to the first embodiment, the technology relating to the information processing system 5 of the present embodiment can facilitate the provision of data from a company reluctant to provide data from the viewpoint of protection of personal information, and further, the statistics Based on the converted data, it is possible to generate a joint database meaningful for consumer behavior analysis and the like.

  Although the first to fifth embodiments have been described above, the present disclosure is not limited to the above embodiments, and various aspects can be adopted. For example, the technical ideas of the second to fourth embodiments may be applied to the fifth embodiment. When the technical concept of the fourth embodiment is applied to the fifth embodiment, for example, the coupling system 80 can have the relation table 553 and the processing of S1110 and S1120 in FIG. 20 can be omitted.

  The connector ID_C may not be pre-assigned to a member. The connector ID_C may be generated by encoding detailed personal information of each member held by the administrator of the first database 151 and the second database 351. The data providing system 10, 30 may have a function of storing such personal information and encoding the personal information to generate a connector ID_C. Examples of the personal information mentioned here include information such as name, address, telephone number, and e-mail address. A hash function may be used for encoding.

  That is, the connector ID_C may be a hash value generated by inputting personal information into the hash function. If personal information is encoded with a common hash function in the first database 151 and the second database 351 to generate the connector ID_C, any of the feature data of the first database 151 and the feature data of the second database 351 regarding the same person. Can also associate the same connector ID_C. Basically, personal information is not leaked to the outside by using an irreversible hash value for the connector ID_C. The personal information to be converted into the hash value is not limited to the above specific example as long as the connector ID_C is information that is approximately unique to each individual. Instead of the connector ID_C, a part of the personal information without coding may be provided to the combined system for the identification of the common member together with the customer number ID_A, ID_B.

  Besides, in the above embodiment, although a plurality of members common to the first group and the second group are divided into a plurality of clusters, a plurality of pairs of members in the first group and members in the second group Pairs may be divided into a plurality of clusters, and a combined database 551 may be generated based on the clustering result. The pair may be a pair of members at least corresponding to each other between the first group and the second group. A pair may be a pair of members in which two members belonging to the pair at least correspond to each other. For example, a pair of members that at least correspond to each other may be a pair of members that are presumed to be identical to each other or likely to be identical to each other.

  For example, in the fourth embodiment, the relationship table 553 includes information that associates the common member customer number ID_A and the customer number ID_B, but the relationship table 553 includes at least members of the first group and the second group that correspond to each other. It may be a table showing a pair with a member of by the association of the customer number ID_A and the customer number ID_B. For example, the relationship table 553 may be a table indicating pairs of members of the first group and members of the second group estimated to be identical, by associating the customer number ID_A with the customer number ID_B.

  Such a relationship table 553 can be generated, for example, by collation of a cookie (Cookie) list. As is well known, cookies are used to identify the person who accesses the web page. By tracking consumer behavior on the network, it is possible to create a cookie list associated with customer number ID_A and a cookie list associated with customer number ID_B. Then, when the matching degree between the cookie list associated with the customer number ID_A and the cookie list associated with the customer number ID_B is high, the members of the first group corresponding to the customer number ID_A and the first corresponding to the customer number ID_B It can be said that members of the two groups are likely to be the same consumer.

  Therefore, if it is presumed that the members of the first group and the members of the second group whose degree of coincidence in the cookie list is equal to or higher than the reference are the same person, the relation table 553 is assumed to be the same as that of the first group. It is possible to create a table showing pairs of members and members of the second group by associating the customer number ID_A with the customer number ID_B.

  The techniques of the second embodiment and the third embodiment that perform clustering based on distance information are also applicable to the information processing system 1 using such a relationship table 553. In this case, from the first data providing system 10 to the coupling system 50, a member list representing each member by the customer number ID_A instead of the connector ID_C as a member list replacing the member lists 1531, 1532 shown in FIGS. You can send the list. Similarly, the second data providing system 30 transmits a member list representing each member by the customer number ID_B as a member list replacing the member lists 3531 and 3532 shown in FIGS. 12 and 15 from the coupling system 50. Can.

  In addition, although the example which processes and connects two databases was demonstrated in the said embodiment, when processing and connecting three or more databases, it can not be overemphasized that the technique of the said embodiment can be applied. Thus, the present disclosure may be exploited for the purpose of processing and combining three or more databases. In this case, a plurality of databases may be linked around one database, or a plurality of databases may be linked in series.

  The database may be a database having feature data of a group whose members (members) are objects and / or places associated with the consumer. In recent years, consumer behavior is closely related to portable terminals such as smart phones. Therefore, the first database 151 and the second database 351 may have feature data for each portable terminal corresponding to the consumer.

  Furthermore, the combined database 551 may be configured as a database having reference information to the first processed database 157 and the second processed database 357. That is, the combined database 551 may not have the substance of integrated feature data, and the combined data of each cluster is the link information to the integrated feature data of the first processed database 157 and the second processed database 357. Alternatively, it may be configured in a form having address information.

  In the second embodiment, clustering is performed using information of the distance D1 and the distance D2, but only one of the distance D1 and the distance D2 may be used for clustering. In this case, one of the unnecessary distances D1 and D2 may be regarded as the value zero to calculate the combined distance D. Information on one of the unnecessary distances D1 and D2 may not be provided from the data providing system 10 or 30 to the coupling system 50. In the third embodiment, as the attribute values Z1 and Z2 used in place of the distances D1 and D2, similarly, a modification is conceivable in which only one of them is used.

  The functions of one component in the above embodiment may be distributed to a plurality of components. The functions of multiple components may be integrated into one component. A part of the configuration of the above embodiment may be omitted. At least a part of the configuration of the above embodiment may be added to or replaced with the configuration of the other above embodiments. All aspects included in the technical concept specified from the wording described in the claims are an embodiment of the present disclosure.

  The correspondence between terms is as follows. The processes of S130, S1010 and S1130 executed by the processors 51 and 81 of the combined system 50 and 80 correspond to an example of the process realized by the clustering unit. The processes of S150, S1030, and S1150 executed by the processors 51 and 81 correspond to an example of the process implemented by the first acquisition unit. The processes of S160, S1040 and S1160 executed by the processors 51 and 81 and the process of the processor 81 reading out the processed database 357 generated in S1160 in S1170 correspond to an example of the process realized by the second acquisition unit. The processes of S170, S1050, and S1170 executed by the processors 51 and 81 correspond to an example of the process realized by the coupling unit. The process of S340 executed by the processor 11 of the first data providing system 10 corresponds to an example of the process implemented by the first generation unit. The process of S440 executed by the processor 31 of the second data providing system 30 and the process of S1160 executed by the processor 81 of the coupling system 80 correspond to an example of the process implemented by the second generation unit.

DESCRIPTION OF SYMBOLS 1 ... Information processing system, 5 ... Information processing system, 10 ... 1st data provision system, 11 ... Processor, 13 ... Memory, 15 ... Storage apparatus, 30 ... 2nd data provision system, 31 ... Processor, 33 ... Memory, 35 ... Storage device 50: Coupling system 51: Processor 53: Memory 55: Storage device 70: Data providing system 71: Processor 73: Memory 75: Storage device 80: Coupling system 81: Processor 81 83: memory, 85: storage device, 151: first database, 153: member list, 155: first cluster information, 157: first database after processing, 351: second database, 353: member list, 355: second Cluster information, 357 ... second database after processing, 551 ... combined data Database, 553 ... relationship table, 1531 ... member list, 1532 ... member list, 1533 ... conversion table, 1554 ... first cluster information, 3531 ... member list, 3532 ... member list, 3533 ... conversion table, 3554 ... second cluster information , NT ... Network.

Claims (17)

An information processing system for generating a new database based on first and second databases, wherein the first database has feature data representing a first feature of a first group of components, for each component of the first group. In the information processing system, the second database includes, for each of the members of the second group, feature data representing a second feature of the members,
A plurality of construction pairs, each of which is a pair of constructions between the first group and the second group and in which two constructions belonging to the pair at least correspond to each other, are divided into a plurality of clusters; A clustering unit that provides cluster information representing a cluster to which each of the constituent pairs belong;
Based on the cluster information acquired from the clustering unit, feature data corresponding to the plurality of structure pairs included in the first database are integrated by statistical processing for each cluster, whereby the integration is performed for each cluster. A first acquisition unit that generates a first integrated feature data as feature data, and a first acquisition unit that acquires a first integrated feature data for each cluster from the first generation unit;
Based on the cluster information acquired from the clustering unit, feature data corresponding to the plurality of constituent pairs included in the second database are integrated by statistical processing for each cluster, whereby the integration is performed for each cluster. A second acquisition unit that generates second integrated feature data as the second feature data, and a second acquisition unit that acquires second integrated feature data for each of the clusters from the second generation unit that generates second integrated feature data as the feature data;
Based on the first integrated feature data for each cluster acquired by the first acquisition unit and the second integrated feature data for each cluster acquired by the second acquisition unit, as the new database A combining unit that generates a combined database including combined data obtained by combining the first integrated feature data and the second integrated feature data for each of the clusters;
An information processing system comprising: The information processing system according to claim 1, wherein
The members of the first and second groups are consumers, and the first database has, for each consumer of the first group, feature data representing a first feature of the consumer, The second database is an information processing system including, for each consumer of the second group, feature data representing a second feature of the consumer. The information processing system according to claim 1 or 2, wherein
Each of the first group of constructs is assigned a separate first identification code, and the first database comprises: feature data for each of the first group of constructs; Store in association with the identification code,
Each of the second group of constructs is assigned a separate second identification code, and the second database is configured to include feature data for each of the second group of constructs in the second group of said constructs. Store in association with the identification code,
The clustering unit identifies the plurality of construction body pairs based on the information indicating the correspondence between the first identification code and the second identification code, divides the plurality of construction pairs into the plurality of clusters, and determines the plurality of clusters as the cluster information. Providing cluster information representing a cluster to which each of the plurality of component pairs belongs in association with the first identification code, and providing the second generator with the plurality of component pairs. An information processing system for providing cluster information representing a cluster to which each belongs in association with the second identification code. The information processing system according to any one of claims 1 to 3, wherein
An information processing system, wherein each of the plurality of component pairs is a pair of components whose entities are assumed to be identical. The information processing system according to claim 1 or 2, wherein
The first and second databases store the feature data of each structure in association with the identification code of the corresponding structure using an identification code common to the first database and the second database. ,
The clustering unit includes, as the plurality of constituent pairs, a plurality of constituent pairs corresponding to the pair of the feature data in which the same identification code is associated between the first database and the second database. An information processing system for dividing the data into clusters and representing, as the cluster information, the first and second generation units in association with the identification code and representing the clusters to which each of the plurality of component pairs belongs. The information processing system according to any one of claims 1 to 5, wherein
The information processing system, wherein the clustering unit divides the plurality of constituent pairs into the plurality of clusters based on the degree of similarity between the plurality of constituent pairs. The information processing system according to claim 6, wherein
The clustering unit acquires similarity information that can specify the similarity of at least one of the first and second features between the plurality of construction body pairs, and the plurality of the clustering units are based on the acquired similarity information. An information processing system, wherein a construction pair is divided into the plurality of clusters such that construction pairs in which at least one of the first and second features is similar are put together. The information processing system according to claim 6 or 7, wherein
The first generation unit provides the clustering unit with a list representing a similarity between a plurality of constructs related to the first feature, the list being a list of a plurality of constructs belonging to the first group.
The second generation unit provides the clustering unit with a list representing a similarity between a plurality of constructs related to the second feature, the list being a list of a plurality of constructs belonging to the second group.
The clustering unit is configured to combine the plurality of structure pairs into a structure pair having similar first and second features based on the list acquired from the first generation unit and the second generation unit. An information processing system for dividing into a plurality of clusters. The information processing system according to claim 6, wherein
The first generation unit provides the clustering unit with a list including a plurality of constructs belonging to the first group and including attribute values of the constructs.
The clustering unit determines similarity between the plurality of construct pairs based on the attribute value, and divides the plurality of construct pairs into the plurality of clusters based on the determined similarity. system. The information processing system according to claim 6, wherein
The first generation unit provides the clustering unit with a list including a plurality of constituents belonging to the first group and including a first attribute value of each of the constituents.
The second generation unit provides the clustering unit with a list including a plurality of constituents belonging to the second group, the list including a second attribute value for each of the constituents.
The clustering unit determines the similarity between the plurality of construction pairs based on the first and second attribute values, and the plurality of construction pairs are determined based on the determined similarity. An information processing system that divides into clusters of. The information processing system according to any one of claims 1 to 10, wherein
The first generation unit and the first database are provided in a first external system, and the second generation unit and the second database are provided in a second external system independent of the first external system. An information processing system configured to be communicable with the first and second external systems. An information processing system capable of communicating with first and second external systems, wherein
The first external system is provided with a first database having feature data representing a first feature of each of the first group of components,
In the information processing system, the second external system includes, for each of the members of the second group, a second database having feature data representing a second feature of the members.
Obtaining a list of a plurality of components belonging to the first group from the first external system, and further obtaining a list of a plurality of components belonging to the second group from the second external system; Based on the acquired list, a plurality of clusters of a pair of members between the first group and the second group, in which two members belonging to the pair at least correspond to each other, are referred to as a plurality of clusters A clustering unit for dividing the information into a plurality of clusters and providing cluster information representing a cluster to which each of the plurality of construct pairs belongs to the first and second external systems;
The first external system integrates feature data corresponding to the plurality of construct pairs included in the first database by statistical processing on the basis of the cluster information acquired from the clustering unit. A first acquisition unit for acquiring the generated first integrated feature data for each cluster from the first external system;
The second external system integrates feature data corresponding to the plurality of construct pairs included in the second database by statistical processing on the basis of the cluster information acquired from the clustering unit. A second acquisition unit for acquiring the generated second integrated feature data for each cluster from the second external system;
The first integrated feature data of the same cluster based on the first integrated feature data for each cluster acquired by the first acquisition unit and the second integrated feature data for each cluster acquired by the second acquisition unit A combination unit that generates a combination database including, for each of the clusters, combination data obtained by combining the second integrated feature data and the second integrated feature data;
An information processing system comprising: An information processing system configured to be communicable with an external system,
The external system comprises, for each component of the first group, a first database having feature data representing a first feature of the component;
The information processing system comprises, for each of the members of the second group, a second database having feature data representing a second feature of the members,
The information processing system further includes
A plurality of construction pairs, each of which is a pair of constructions between the first group and the second group and in which two constructions belonging to the pair at least correspond to each other, are divided into a plurality of clusters; A clustering unit that provides the external system with cluster information representing a cluster to which each of the constituent pairs belongs;
The external system integrates feature data corresponding to the plurality of construct pairs included in the first database by statistical processing on the basis of the cluster information received from the clustering unit, and generates the data by the statistical processing. An acquisition unit for acquiring first integrated feature data for each cluster from the external system;
Generation is performed to generate second integrated feature data for each cluster by integrating feature data corresponding to the plurality of component pairs of the second database based on the cluster information by statistical processing for each cluster Department,
The first integrated feature data of the same cluster and the second integrated feature data of the same cluster based on the first integrated feature data of each cluster acquired by the acquisition unit and the second integrated feature data of the cluster generated by the generation unit. A combining unit for creating a combined database having combined data for each cluster, combined data being combined with integrated feature data;
An information processing system comprising: An information processing system for generating a new database based on first and second databases, wherein the first database has feature data representing a first feature of a first group of components, for each component of the first group. The second database includes, for each of the second group of structures, feature data representing a second feature of the structures, and the first and second groups include at least a plurality of structures in common. In an information processing system that includes in part:
A clustering unit that divides the plurality of common constituents into a plurality of clusters and provides cluster information representing a cluster to which each of the plurality of common constituents belongs;
Based on the cluster information acquired from the clustering unit, feature data of the plurality of common constituents possessed by the first database are integrated by statistical processing for each cluster, whereby the integration is performed for each cluster A first acquisition unit that generates a first integrated feature data as feature data; a first acquisition unit that acquires a first integrated feature data for each cluster from the first generation unit;
Based on the cluster information acquired from the clustering unit, feature data of the plurality of common constituents included in the second database are integrated by statistical processing for each cluster, whereby the integration is performed for each cluster A second generation unit configured to generate second integrated feature data as feature data; and a second acquisition unit configured to obtain second integrated feature data for each cluster from the second generation unit;
Based on the first integrated feature data for each cluster acquired by the first acquisition unit and the second integrated feature data for each cluster acquired by the second acquisition unit, as the new database A combining unit that generates a combined database including combined data obtained by combining the first integrated feature data and the second integrated feature data for each of the clusters;
An information processing system comprising:   In order to cause a computer to realize functions as a clustering unit, a first acquisition unit, a second acquisition unit, and a coupling unit included in the information processing system according to any one of claims 1 to 12 and claim 14. Programs. An information processing method for generating a new database based on first and second databases,
The first database has feature data representing a first feature of the first group of components, and the second database includes the feature data of the second group of components. Having feature data representing the second feature,
The method is
A plurality of construction pairs between the first group and the second group, in which two constructions belonging to the pair at least correspond to each other, are divided into a plurality of clusters; A clustering procedure that provides cluster information representing the cluster to which each of the construct pairs belongs;
Based on the cluster information provided by the clustering procedure, feature data corresponding to the plurality of construct pairs included in the first database are integrated by statistical processing for each cluster, whereby the integration is performed for each cluster A first acquisition procedure for acquiring the first integrated feature data for each cluster from the device for generating the first integrated feature data as the selected feature data;
Based on the cluster information provided by the clustering procedure, feature data corresponding to the plurality of constituent pairs included in the second database are integrated by statistical processing for each cluster, whereby the integration is performed for each cluster A second acquisition procedure for acquiring second integrated feature data for each cluster from a device for generating second integrated feature data as the selected feature data;
The same as the new database based on the first integrated feature data for each cluster acquired by the first acquisition procedure and the second integrated feature data for each cluster acquired by the second acquisition procedure A combining procedure for creating a combined database having combined data for each cluster that combines the first integrated feature data of the cluster and the second integrated feature data;
Information processing method including: An information processing method for generating a new database based on first and second databases,
The first database has feature data representing a first feature of the first group of components, and the second database includes the feature data of the second group of components. Having feature data representing a second feature, the first and second groups at least partially including a plurality of common components,
The method is
A clustering procedure for dividing the common plurality of structures into a plurality of clusters and providing cluster information representing a cluster to which each of the common plurality of structures belongs;
Based on the cluster information provided by the clustering procedure, feature data of the plurality of common constituents possessed by the first database are integrated by statistical processing for each cluster, whereby the integration is performed for each cluster A first acquisition procedure for acquiring the first integrated feature data for each cluster from the device for generating the first integrated feature data as the selected feature data;
Based on the cluster information provided by the clustering procedure, feature data of the plurality of common constituents possessed by the second database are integrated by statistical processing for each cluster, whereby the integration is performed for each cluster A second acquisition procedure for acquiring second integrated feature data for each cluster from a device for generating second integrated feature data as the selected feature data;
The same as the new database based on the first integrated feature data for each cluster acquired by the first acquisition procedure and the second integrated feature data for each cluster acquired by the second acquisition procedure A combining procedure for creating a combined database having combined data for each cluster that combines the first integrated feature data of the cluster and the second integrated feature data;
Information processing method including:

Images