Collaborative data analysis system and collaborative data analysis method

Abstract

To analyze and utilize information that cannot be understood from information held by a single organization, by combining non-personally identifiable information collaborated among multiple organizations related to an objective service.SOLUTION: A collaborative data analysis system extracts usage performance data and statistical information data related to an objective service; calculates a usage rate of the objective service based on the usage performance data and the statistical information data extracted; calculates attribute items related to the use of the objective service based on the calculated usage rate and the extracted statistical information data; calculates, from each attribute value for each attribute item related to the use of the objective service, a distribution of influence that the attribute item has on the use of the objective service; calculates attribute values that have singular values in the distribution of influence on the use of the objective service as attribute values that influence the use of the objective service; and outputs the calculated attribute values and attribute items that influence the use of the objective service as attribute information that becomes candidates for recommending the service.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a federated data analysis system and a federated data analysis method for analyzing data that is federated between organizations. [Background technology]

[0002] In recent years, there has been a growing trend to link and utilize the diverse data held by multiple organizations, such as local governments and private businesses, in order to implement effective policies and create services in local areas.However, when it comes to data held by organizations such as local governments and private businesses, especially data related to individuals, such as personal attributes and behavioral history, it can be difficult to link and analyze the data between organizations due to privacy concerns.

[0003] Therefore, a technique is known that can facilitate analysis by statistically processing data collected from a plurality of users and visualizing the attributes and behavior of the users (see, for example, Patent Document 1). In the data analysis system disclosed in Patent Document 1, in response to a request from a service receiving provider device, the service platform provider device performs statistical processing of collected data in accordance with the request using a statistical processing unit, creates attribute analysis data or people flow analysis data based on the obtained statistical data, and transmits it to the service receiving provider device. This technology is expected to facilitate visualization and support analysis by statistically processing a huge amount of user attribute information and behavior information. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2019-23851 Summary of the Invention [Problem to be solved by the invention]

[0005] However, conventional technology does not disclose a method for multiple organizations, such as local governments and private businesses, to link, combine, and utilize information held by each organization that has been processed so that individuals cannot be identified.

[0006] The present invention has been made in consideration of these circumstances, and aims to combine, analyze, and utilize information that does not identify individuals and other information that is shared among multiple organizations. [Means for solving the problem]

[0007] In order to solve the above problems, one of the representative linked data analysis systems of the present invention is a system that includes a terminal used by a user, a server that performs data analysis, and a service provided by the user. It is data that is structured in a form that cannot identify individuals. Usage data and Public data about residents and a database that stores statistical information data related to attributes that constitute usage history data, are connected to each other via a network, the terminal transmits input information related to a service targeted by a user to the server, and outputs attribute information that is a candidate for recommending the target service received as an output from the server, and the server has a function of extracting usage history data and statistical information data related to the target service from the database based on the input information, a function of calculating a usage rate of the target service based on the extracted usage history data and statistical information data, a function of calculating attribute items related to usage of the target service based on the calculated usage rate and the extracted statistical information data, a function of calculating a distribution of the influence that the attribute items have on usage of the target service from each attribute value for each attribute item related to usage of the target service, and a function of calculating the distribution of the influence that the attribute items have on usage of the target service. Based on this, the estimated influence of each attribute value is Singular Values is The system comprises a processing unit having a function of calculating an attribute value as an attribute value that affects the use of the target service, and a function of outputting the calculated attribute value that affects the use of the target service and the attribute item having the attribute value as attribute information that is a candidate for recommending the target service, and a memory unit that stores each piece of information calculated by the processing unit using each of the functions. According to the above aspect, for example, by combining and analyzing usage history data for each service, which has been appropriately processed so that individuals who have used each service cannot be identified and is provided by businesses that provide each service (service providers), with statistical data on residents, such as that held by local governments, it becomes possible to realize an analysis that enables recommendations to be made that takes into account aspects of individual attributes that are not included in the usage history data. [Effects of the Invention]

[0008] According to the present invention, it is possible to perform analysis across multiple organizations by combining information held by each organization that has been processed so that individuals cannot be identified with other information, and to utilize this information. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a block diagram showing an example of the overall configuration of a linked data analysis system according to the present invention. [Figure 2] 2 is a block diagram showing an example of a hardware configuration of a user terminal according to the present invention; FIG. [Figure 3] FIG. 2 is a block diagram showing an example of a hardware configuration of an analysis server according to the first embodiment of the present invention. [Figure 4] FIG. 10 is a diagram showing an example of the contents of a usage record data table. [Figure 5] 10A and 10B are diagrams showing examples of the contents of a statistical information data table and a number of people information data table. [Figure 6] FIG. 10 is a diagram illustrating an example of the contents of a service usage rate data table. [Figure 7] FIG. 10 is a diagram showing an example of the contents of an attribute similarity information data table. [Figure 8] FIG. 10 is a diagram showing an example of the contents of a service usage characteristics information data table. [Figure 9] FIG. 10 is a diagram illustrating an example of the contents of a service usage impact distribution data table. [Figure 10]FIG. 10 is a diagram illustrating an example of a series of sequences in Example 1, from when a user terminal accepts input information, to when an analysis server calculates attribute information that is a candidate for recommending the input target service, to when the user terminal outputs and displays the information. [Figure 11] FIG. 10 is a diagram showing an example of the configuration of a registration screen that is presented to a user by a user terminal and that accepts input of necessary information and data. [Figure 12] FIG. 10 is a diagram showing a processing flow in a step in which the analysis server calculates the characteristics of the service usage attributes of the target service. [Figure 13] FIG. 10 is a diagram showing the processing flow in the step in which the analysis server determines whether or not the attribute item is an "attribute item related to the use of the service" in the target service. [Figure 14] FIG. 10 is a flowchart showing the processing flow in the step in which the analysis server calculates the "distribution of influence on service usage" and the "attribute value that influences service usage" for the target service. [Figure 15] FIG. 10 is a diagram showing an example of the configuration of a result screen that presents attribute information that is a candidate for recommending a target service in the first embodiment. [Figure 16] FIG. 10 is a block diagram showing an example of a hardware configuration of an analysis server according to a second embodiment of the present invention. [Figure 17] FIG. 10 is a diagram showing an example of the contents of a target attribute statistical information data table. [Figure 18] FIG. 10 is a diagram showing an example of the contents of a target attribute time-series information data table. [Figure 19] FIG. 10 is a diagram showing an example of the contents of a service suitability information data table. [Figure 20] FIG. 11 is a diagram illustrating an example of a series of sequences from receiving input information from a user terminal to calculating information on candidate services to be recommended for a target attribute by the analysis server, and outputting and displaying the information on the user terminal in the second embodiment. [Figure 21] FIG. 10 is a diagram showing an example of the configuration of a registration screen that is presented to a user by a user terminal and that accepts input of necessary information and data. [Figure 22]FIG. 10 is a diagram showing a processing flow of a step for calculating a degree of match of a selected candidate service in a target attribute. [Figure 23] FIG. 10 is a diagram showing a processing flow of a step for calculating an increase tendency degree of a selected candidate service in a target attribute. [Figure 24] FIG. 11 is a diagram showing an example of the configuration of a result screen that presents information about services that are candidates for recommendation for a target attribute in the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, with reference to the drawings, embodiments of the present invention will be described with reference to Examples 1 and 2. However, the present invention should not be construed as being limited to the following description of Examples 1 and 2. In addition, in the configurations of the embodiments of the present invention described below, the same parts or parts having similar functions are denoted by the same reference numerals in different drawings, and duplicated explanations may be omitted.

[0011] In order to facilitate understanding of the invention, the position, size, shape, range, etc. of each component in the drawings may not represent the actual position, size, shape, range, etc. Therefore, the present invention is not necessarily limited to the position, size, shape, range, etc. disclosed in the drawings.

[0012] In this specification, various programs executed by the processing unit or parts thereof may be referred to as "units," "means," "functions," etc. Furthermore, the configuration of each embodiment of the present invention may be configured as a single computer, or any part of the input device, output device, processing device, and storage device may be configured as another computer connected via a network. [Example]

[0013] Example 1 is a linked data analysis system and a linked data analysis method for calculating candidate attributes for recommendation to a user who wants to know candidate attributes for recommending a target service using usage history data that has been appropriately processed so that individuals who have used the target service cannot be identified and resident statistical information data (e.g., public statistical information data provided by a local government, etc.), and presenting the calculated attribute information to the user.

[0014] Here, the user who is the main user of the user terminal is, for example, a local government employee who intends to use recommendations to residents to implement some kind of local policy, such as promoting activities among residents or revitalizing the area.

[0015] In Example 1, it is assumed that the system is used for the purpose of knowing the attributes of residents who are candidates for recommending a service in order to encourage the use of the service in the area. Note that in Example 1 and Example 2 described later, the service may be a procedure at an administrative agency.

[0016] FIG. 1 is a block diagram showing an example of the overall configuration of a linked data analysis system 1 according to the present invention. The linked data analysis system 1 is a service system made up of a user terminal 101, an analysis server 102, and a database (hereinafter abbreviated as "DB") server 103. The DB server 103 stores a usage history DB 104 and a statistical information DB 105. The user terminal 101, analysis server 102, and DB server 103 are communicatively connected via a network 106 such as the Internet. Furthermore, the analysis server 102 and DB server 103 may be implemented as the same device, or may be implemented separately.

[0017] FIG. 2 is a block diagram showing an example of the hardware configuration of the user terminal 101 according to the present invention. Here, the user of the user terminal is a user who wants to know the attributes that are candidates for recommending a certain service.

[0018] The user terminal 101 can be realized by a general computer, and includes, for example, a control unit 201, a network interface unit 202, an input unit 203, and an output unit 204. These units are connected by a BUS 205 or the like.

[0019] The control unit 201 is configured with a CPU or the like and executes each program. The control unit 201 receives input information from the input unit 203 and sends output information to the output unit 204. Note that the functions of each program may be realized by hardware such as an LSI. The control unit 301 of the analysis server 102, which will be described later, has a similar configuration.

[0020] The network interface unit 202 transmits IP packets transferred by the control unit 201 to the analysis server 102 via a network 106 such as the Internet, and also transfers IP packets received by the user terminal 101 via a network 106 such as the Internet to the control unit 201. The network interface unit 202 is configured, for example, by a network card or the like.

[0021] The input unit 203 transmits input from the outside to the control unit 201, and is composed of, for example, a keyboard, a mouse, and the like. The output unit 204 outputs information from the control unit 201 to the outside, and is composed of, for example, a display, a printer, or the like.

[0022] FIG. 3 is a block diagram showing an example of the hardware configuration of the analysis server 102 according to the first embodiment of the present invention. The analysis server 102 can be realized by a general server device, and one example of its configuration includes a control unit 301, a network interface unit 302, a web service providing unit 303, a data access unit 304, a data registration unit 305, an input information registration unit 306, an input information storage unit 307, a service usage rate calculation unit 308, a service usage rate storage unit 309, a service feature calculation unit 310, an attribute similarity information storage unit 311, a service feature storage unit 312, and a recommendation attribute calculation unit 313. These units are connected by a BUS 314 or the like.

[0023] 3 do not need to be physically separated, and one element may serve the other. Note that input and output devices that are typically included in servers are not specifically shown in FIG. The control unit 301 is composed of a CPU and the like, and executes the following programs.

[0024] The network interface unit 302 transmits IP packets transferred by the control unit 301 to the user terminal 101 via a network 106 such as the Internet, and also transfers IP packets received by the analysis server 102 via a network 106 such as the Internet to the control unit 301. The network interface unit 302 is configured, for example, by a network card.

[0025] The web service providing unit 303 transmits data written in a format that can be displayed on a web browser to the user terminal 101 via a network 106 such as the Internet. Furthermore, the web service providing unit 303 receives information such as operation details on the web browser of the user terminal 101 via a network 106 such as the Internet, and based on the received information, the data access unit 304, data registration unit 305, input information registration unit 306, service usage rate calculation unit 308, service feature calculation unit 310, and recommendation attribute calculation unit 313 execute their respective processes. This web service providing unit 303 is, for example, a program having the above-mentioned web service functions.

[0026] The data access unit 304 is a program having the function of accessing data in the usage history DB 104 and statistical information DB 105 in the DB server 103 based on information received from the user terminal 101 and information stored in the input information storage unit 307 described later, and performing processing to extract the necessary information.

[0027] The data registration unit 305 is a program having a function of performing processing to register data in the usage record DB 104 and the statistical information DB 105 in the DB server 103 based on information received from the user terminal 101 .

[0028] The input information registration unit 306 is a program that has the function of receiving information, etc., entered on the web browser of the user terminal 101, related to the target service (target service) for which candidate attributes to be recommended are to be calculated, via a network 106 such as the Internet, and registering information related to the target service in the input information storage unit 307 based on the received information.

[0029] The input information storage unit 307 stores some or all of the information related to the target service registered by the input information registration unit 306 . The input information storage unit 307, the service usage rate storage unit 309 (described later), the attribute similarity information storage unit 311, and the service feature storage unit 312 may be configured as multiple storage units each having a different storage method depending on the size, format, etc. of the data to be stored. In addition, some or all of the data equivalent to the data stored in the input information storage unit 307, the service usage rate storage unit 309 described below, the attribute similarity information storage unit 311, and the service feature storage unit 312 may be provided by other multiple servers connected to the network 106 such as the Internet.

[0030] The service usage rate calculation unit 308 is a program that has the function of calculating the usage rate (service usage rate) of the target service based on the information accessed by the data access unit 304 and extracted from the usage history DB 104 and the statistical information DB 105, and registering information related to the calculated service usage rate in the service usage rate storage unit 309. The service utilization rate storage unit 309 stores part or all of the information about the service utilization rate of the target service calculated by the service utilization rate calculation unit 308 .

[0031] The service feature calculation unit 310 is a program having the functions of calculating information regarding the similarities between attributes (attribute similarity information) based on the information in the service usage rate storage unit 309 and the statistical information DB 105, and registering the calculated attribute similarity information in the attribute similarity information storage unit 311, and calculating information regarding the characteristics of attributes related to the use of the target service (service usage attribute characteristics) based on the information in the service usage rate storage unit 309 and the attribute similarity information storage unit 311, and registering the calculated information regarding the service usage attribute characteristics in the service feature storage unit 312.

[0032] The attribute similarity information storage unit 311 stores a part or all of the attribute similarity information calculated by the service feature calculation unit 310 . The service feature storage unit 312 stores part or all of the information about the service use attribute features of the target service calculated by the service feature calculation unit 310 .

[0033] The recommendation attribute calculation unit 313 is a program having a function of calculating attribute information that is a candidate for recommending a target service based on the information in the service feature storage unit 312 and transmitting the information to the user terminal 101 .

[0034] In addition, some or all of the programs and data executed by the control unit 201 of the user terminal 101 and the control unit 301 of the analysis server 102 may be stored in advance in a non-transitory storage device (not shown) within the computer that realizes these devices, or may be imported from a non-transitory storage medium, or may be imported via a network from an information processing device equipped with an external non-transitory storage device.

[0035] Next, various data tables used in the first embodiment will be described with reference to FIGS. <Usage history data table 401>

[0036] FIG. 4 is a diagram showing an example of the contents of the usage record data table 401. As shown in FIG. The usage history data table 401 is stored in the usage history DB 104 in the DB server 103, and is composed of a user column group 402, a service column 403, and a number of users column 404. Note that one data table may contain information about multiple services, or a different data table may be used for each service.

[0037] The User column group 402 is included in the usage record of the service and is composed of columns of attribute items of people who have used the service. For example, in the case of information regarding the usage record of hot spring facility A, in which the number of users is tallied using two attribute items, "age" and "gender," the User column group 402 is composed of an age column 405 and a gender column 406. Each attribute item column that makes up the User column group 402 stores information that uniquely identifies the attribute value of the attribute item corresponding to the column.

[0038] The service column 403 stores information that uniquely identifies a service. The number of users column 404 stores the number of people who have the attribute stored in each attribute item column of the user column group 402 and have used the service stored in the service column 403 . <Statistical Information Data Table 501 and Number of People Information Data Table 511>

[0039] 5 shows an example of the contents of the statistical information data table 501 and the number of people information data table 511. The statistical information data table 501 and the number of people information data table 511 are stored in the statistical information DB 105 in the DB server 103.

[0040] The statistical information data table 501 is made up of attribute column groups 502 and 503, an attribute item column 504, and attribute value column groups 505-508. The attribute column groups 502 and 503 are composed of columns of attribute items corresponding to attributes that serve as parameters when aggregating statistical information. For example, if you want to aggregate statistical information related to the attribute item column 504 for each combination of attribute values ​​of "age" and "gender," the attribute column groups are composed of an age column 502 and a gender column 503.

[0041] Each attribute column constituting the attribute column groups 502 and 503 stores information that uniquely identifies an attribute value in an attribute item corresponding to the attribute column. The attribute item column 504 stores information that uniquely identifies an attribute item.

[0042] The attribute value column group 505 to 508 is made up of columns of the attribute values ​​of the attribute items stored in the attribute item column 504 . For example, if the attribute item stored in attribute item column 504 is "area of ​​residence" and its attribute values ​​are four types, "area A," "area B," "area C," and "area D," it is composed of area A column 505, area B column 506, area C column 507, and area D column 508.

[0043] Each attribute value column constituting attribute value column groups 505 to 508 stores the percentage of people who have the attribute value corresponding to each attribute value column among people who have the attribute values ​​stored in attribute column groups 502 and 503. In this specification, the information composed of the values ​​of the attribute value columns 505 to 508 in a given row may be referred to as the "distribution" of the attribute items stored in the attribute item column 504 among the attributes stored in the attribute columns 502 and 503 of the corresponding row.

[0044] Next, the number of people information data table 511 is made up of attribute column groups 512 and 513 and a total number column 514 . The attribute column groups 512 and 513 are composed of columns of attribute items corresponding to attributes that serve as parameters when aggregating statistical information. For example, if you want to aggregate the total number of people who have each attribute value for each combination of "age" and "gender," the attribute column groups are composed of an age column 512 and a gender column 513. The total number column 514 stores the total number of people who have the attribute values ​​stored in the attribute column groups 512 and 513 . <Service Usage Rate Data Table 601>

[0045] FIG. 6 is a diagram showing an example of the contents of the service usage rate data table 601. As shown in FIG. The service usage rate data table 601 is stored in the service usage rate storage unit 309 of the analysis server 102 and is composed of a user column group 602 , a service column 603 , a usage rate column 604 , and a usage rate ranking column 605 .

[0046] The User column group 602 is included in the usage history of the service and is composed of columns of attribute items of people who have used the service. For example, if the usage history of hot spring facility A is information in which the number of users is tallied using two attribute items, "age" and "gender," the User column group 602 is composed of an age column 606 and a gender column 607. The column of each attribute item that makes up the User column group 602 stores information that uniquely identifies the attribute value of the corresponding attribute item. The service column 603 stores information that uniquely identifies a service.

[0047] The utilization rate column 604 stores the percentage of people who have the attribute value stored in each attribute item column of the user column group 602 and who have used the service stored in the service column 603 . The utilization rate ranking column 605 stores the ranking of each service stored in the service column 603, in descending order of the utilization rate stored in the utilization rate column 604. <Attribute Similarity Information Data Table 701>

[0048] FIG. 7 is a diagram showing an example of the contents of the attribute similarity information data table 701. As shown in FIG. The attribute similarity information data table 701 is stored in the attribute similarity information storage unit 311 of the analysis server 102 and is composed of a user column group 702 , an attribute item column 703 , a similarity column 704 with the top attribute, and a similarity ranking column 705 .

[0049] The User column group 702 is included in the usage record of the target service and is composed of columns of attribute items of people who have used the target service. For example, if the target service is hot spring facility A and the usage record of hot spring facility A is information in which the number of users is tallied using two attribute items, "age" and "gender," the User column group 702 is composed of an age column 706 and a gender column 707. The column of each attribute item that makes up the User column group 702 stores information that uniquely identifies the attribute value of the corresponding attribute item. The attribute item column 703 stores information that uniquely identifies an attribute item.

[0050] The similarity column 704 with the number one attribute stores the similarity between the attribute ranked first in usage rate for the target service and the attribute stored in each attribute item column of the user column group 702, for the distribution of the attribute item stored in the attribute item column 703. The similarity ranking column 705 stores, for each service stored in the attribute item column 703, a ranking in descending order of the similarity with the first-ranked attribute stored in the similarity column 704. <Service Usage Characteristics Information Data Table 801>

[0051] FIG. 8 is a diagram showing an example of the contents of the service usage characteristics information data table 801. As shown in FIG. The service usage characteristics information data table 801 is stored in the service characteristics storage unit 312 of the analysis server 102 and is composed of a service column 802 , an attribute item column 803 , and an attribute value column 804 . The service column 802 stores information that uniquely identifies a service.

[0052] The attribute item column 803 stores “attribute items related to the use of the service” for the service stored in the service column 802 . The attribute value column 804 stores, for the service stored in the service column 802, an “attribute value that affects the use of the service” that corresponds to the attribute item stored in the attribute item column 803. <Service Usage Impact Distribution Data Table 901>

[0053] FIG. 9 is a diagram showing an example of the contents of the service usage impact distribution data table 901. As shown in FIG. The service usage impact distribution data table 901 is stored in the service feature storage unit 312 of the analysis server 102 and is composed of a service column 902 , an attribute item column 903 , an attribute value column 904 , an estimated impact column 905 , and a standard deviation column 906 .

[0054] The service column 902 stores information that uniquely identifies a service. The attribute item column 903 stores “attribute items related to the use of the service” for the service stored in the service column 902 . The attribute value column 904 stores information that uniquely identifies the attribute value corresponding to the attribute item stored in the attribute item column 903 .

[0055] The estimated impact column 905 stores the estimated impact that the attribute stored in the attribute value column 904 will have on the use of the service stored in the service column 902 . The standard deviation column 906 stores the value of the standard deviation calculated when estimating the estimated influence stored in the estimated influence column 905 .

[0056] FIG. 10 is a diagram showing an example of a sequence 1001 up to presenting attribute information that is a candidate for recommending a target service to a user. Figure 10 shows the processing flow of a series of sequences 1001 in which the user terminal 101 accepts input from the user regarding a target service for which the user wishes to calculate candidate attributes to recommend, the analysis server 102 calculates candidate attribute information for recommending the target service based on the information in the usage history DB 104 and statistical information DB 105 in the DB server 103, and sends it to the user terminal, and the user terminal 101 outputs the received information and presents it to the user.

[0057] Note that communication between the analysis server 102 and the DB server 103, such as data access from the analysis server 102 to the DB server 103 and data registration from the analysis server 102 to the DB server 103, is omitted.

[0058] In step 1001 (s1001), the user terminal 101 receives input from the user regarding a target service for which candidate attributes to be recommended are to be calculated, and transmits this input information to the analysis server 102. Examples of target services include "hot spring facility A," "sports gym B," and "museum C."

[0059] In step 1002 (s1002), the analysis server receives input information related to the target service from the user terminal 101, and registers the received input information in the input information storage unit 307 of the analysis server .

[0060] In step 1003 (s1003), the analysis server 102 accesses the usage history DB 104 in the DB server 103 and checks whether usage history data for the input target service exists in the usage history DB 104. If it exists (Yes), the process proceeds to step 1007 (s1007); if it does not exist (No), the process sends a message to the user terminal 101 that it does not exist.

[0061] In step 1004 (s1004), when the user terminal 101 receives information indicating that no usage history data exists, it outputs a message prompting the user to input usage history data for the target service.

[0062] In step 1005 (s1005), the user terminal 101 receives input relating to the usage record data of the target service from the user, and transmits this input information to the analysis server 102.

[0063] In step 1006 (s1006), the analysis server 102 receives input information related to the usage history data of the target service from the user terminal 101, and registers the usage history data of the target service in the usage history DB 104 based on this received information.

[0064] In step 1007 (s1007), the analysis server 102 accesses the DB server 103 and extracts information on the attribute items (usage history attribute items) included in the relevant usage history from the usage history data table 401 (Figure 4) related to the usage history of the target service, and also extracts information on the attribute items (candidate attribute items) for which statistical information is compiled for each attribute (usage history attribute) of the usage history attribute item from the statistical information data table 501 (Figure 5).

[0065] In step 1008 (s1008), the analysis server 102 transmits to the user terminal 101 the information on the usage record attribute items and candidate attribute items extracted in step 1007 (s1007).

[0066] In step 1009 (s1009), the user terminal 101 outputs the information on the usage record attribute items and candidate attribute items received from the analysis server 102.

[0067] In step 1010 (s1010), the user terminal 101 receives input from the user about information relating to additional statistical information data that the user wishes to use, and transmits this input information to the analysis server 102.

[0068] In step 1011 (s1011), the analysis server 102 receives input information relating to the additional statistical information data that the user wishes to use from the user terminal 101, and registers the additional statistical information data that the user wishes to use in the statistical information DB 105 based on this received information.

[0069] Here, the registration screen of the user terminal 101 will be described. FIG. 11 is a diagram showing an example of the configuration of a registration screen 1101 that is presented to the user by the user terminal 101 and is used to accept input of necessary information and data. A registration screen 1101 is displayed in the window of a web browser, etc. This registration screen 1101 has a target service input area 1102, a send button 1103, a usage history data registration area 1104, and an attribute item information confirmation area 1107 arranged thereon.

[0070] The user inputs information that can uniquely identify the target service, such as the name of the target service for which the user wants to calculate candidate attributes to be recommended, into the target service input area 1102. When the send button 1103 is pressed, the user terminal 101 transmits the content input into the target service input area 1102 to the analysis server 102.

[0071] The usage history data registration area 1104 is output and displayed in step 1004 (s1004). In the usage history data registration area 1104, a usage history data input area 1105 and a registration button 1106 are arranged.

[0072] When the usage history data registration area 1104 is displayed, the user inputs information related to the usage history data of the target service in the usage history data input area 1105. This input can be done by uploading a data file or by inputting the location of the usage history data. When the registration button 1106 is pressed, the user terminal 101 transmits the information input in the usage history data input area 1105 to the analysis server 102.

[0073] The attribute item information confirmation area 1107 is output and displayed in step 1009 (s1009). In the attribute item information confirmation area 1107, a statistical information data input area 1108, a registration button 1109, and an attribute item display area 1110 are arranged. In the attribute item display area 1110, information on the usage record attribute items and candidate attribute items sent by the analysis server 102 is output.

[0074] When the attribute item information confirmation area 1107 is displayed, the user checks the information on the usage history attribute items and candidate attribute items displayed in the attribute item display area 1110, and if there is statistical information they would like to use other than the displayed candidate attribute items, they input information related to the statistical information they would like to use in the statistical information data input area 1108. This input method can be done by uploading a data file or by inputting the location of the statistical information data. When the registration button 1109 is pressed, the user terminal 101 sends the information input in the statistical information data input area 1108 to the analysis server 102.

[0075] Returning to FIG. 10, the processing flow from step 1011 (s1011) onwards will be described. In step 1012 (s1012), the analysis server 102 calculates the service usage rate of the target service based on the information in the usage history data table 401 (Figure 4) of the target service stored in the usage history DB 104 and the number of users information data table 511 (Figure 5) stored in the statistical information DB 105.

[0076] Specifically, the analysis server 102 calculates the service usage rate by dividing the value of the number of users in the number of users column 404 in the usage history data table 401 (FIG. 4) by the number of users value stored in the total number column 514 in a row in which the attribute values ​​stored in the attribute columns 512 and 513 in the number of people information data table 511 (FIG. 5) match the attribute values ​​stored in the User column group 402 of the corresponding row in the usage history data table 401 (FIG. 4). The calculated service usage rate information is stored in the service usage rate storage unit 309 of the analysis server 102 as a service usage rate data table 601 (FIG. 6).

[0077] If there are multiple usage history attribute items, the service usage rate may be calculated by focusing on all or some of the usage history attribute items. The calculation may also be performed for attributes that combine multiple usage history attribute items, or for each usage history attribute item. When focusing on some of the usage history attribute items or performing calculations for each usage history attribute item, the service usage rate is calculated after recounting the number of users for each attribute value of the usage history attribute items of interest or a combination thereof.

[0078] Next, in step 1013 (s1013), the analysis server 102 calculates information regarding the attribute characteristics (service usage attribute characteristics) related to the usage of the target service based on the information in the service usage rate data table 601 (Figure 6) stored in the service usage rate memory unit 309 and the statistical information data table 501 (Figure 5) stored in the statistical information DB 105. Furthermore, information about the similarities between attributes calculated during the process (attribute similarity information) is registered in the attribute similarity information storage unit 311, and information about the calculated service use attribute features is registered in the service feature storage unit 312.

[0079] Here, the detailed flow of step 1013 (s1013) will be explained. 12 is a diagram showing the processing flow in step 1013 (s1013) in which the analysis server 102 calculates the characteristics of the service usage attributes of the target service. In the following, since the analysis server 102 is the processing subject of each step, the subject will not be described.

[0080] First, in step 1201 (f1201), for rows in the service usage rate data table 601 (FIG. 6) where the service stored in the service column 603 matches the target service, the service usage rate values ​​stored in the usage rate column 604 are ranked in descending order (usage rate ranking) and stored in the usage rate ranking column 605.

[0081] In step 1202 (f1202), one candidate attribute item that has not been selected in step 1013 (s1013) is selected from the candidate attribute items extracted in step 1007 (s1007).

[0082] In step 1203 (f1203), it is determined whether the candidate attribute item selected in step 1202 (f1202) is an "attribute item related to service usage" of the target service, and if the determination result is "attribute item related to service usage" (Yes), it is stored in the service usage characteristic information data table 801 (Figure 8) in the service characteristic storage unit 312, and the process proceeds to step 1204 (f1204); if not (No), the process proceeds to step 1207 (f1207).

[0083] Furthermore, the detailed flow of step 1203 (f1203) will be explained. 13 shows the detailed flow of step 1203 (s1203), in which the analysis server 102 determines whether the attribute item is related to the use of the service in question. In the following, the analysis server 102 is the processing entity for each step, so the entity will not be described.

[0084] In step 1301 (f1301), the DB server 103 is accessed, and information aggregated for each usage record attribute for which the service usage rate has been calculated for the selected candidate attribute item in the statistical information data table 501 (FIG. 5) is extracted from the statistical information DB 105. For example, if the selected candidate attribute item is "area of ​​residence" and the attribute items of the usage record of the target service are "age" and "gender," the attribute column groups 502 and 503 in the statistical information data table 501 are composed of the "age" column and the "gender" column, and information on rows where the attribute item column 504 matches "area of ​​residence" is extracted.

[0085] In step 1302 (f1302), the attribute value calculated in step 1201 (f1201) and stored in the User column group 602 in the row with the highest usage rate ranking stored in the usage rate ranking column 605 of the service usage rate data table 601 (FIG. 6) is extracted as the highest usage rate attribute.

[0086] In step 1303 (f1303), the similarity between the attribute with the highest usage rate and each of the other usage history attributes is calculated for the distribution of the selected candidate attribute items, and the calculated value is stored in the similarity column 704 with the highest usage rate attribute in the attribute similarity information data table 701 (Figure 7) in the attribute similarity information storage unit 311.

[0087] Specifically, for the statistical information data table 501 (FIG. 5) extracted in step 1301 (f1301), the similarity is calculated between a vector (distribution vector) composed of values ​​in the attribute value column group 505 to 508 in rows in which the attribute values ​​stored in the attribute column group 502 and 503 match the most frequently used attribute extracted in step 1302 (f1302), and the distribution vector in each row in the statistical information data table 501 extracted in step 1301 (f1301).

[0088] For example, suppose the most popular attribute is "age" "30-39" and "gender" "male." Looking at statistical information data table 501 (FIG. 5), for the candidate attribute item "area of ​​residence," attribute value column group 505-508 is composed of region A column 505, region B column 506, region C column 507, and region D column 508. The distribution vector of the most popular attribute ("age" "30-39" and "gender" "male") is a vector composed of the values ​​of region A column 505, region B column 506, region C column 507, and region D column 508, and is (0.26, 0.2, 0.45, 0.09). When calculating the similarity between the most popular attribute for the distribution of "area of ​​residence" and another attribute with a history of usage, for example, "age" = "30-39" and "gender" = "female," the similarity is calculated between the distribution vector of the most popular attribute (0.26, 0.2, 0.45, 0.09) and the distribution vector of the attributes "age" = "30-39" and "gender" = "female" (0.23, 0.18, 0.42, 0.17). Here, the similarity may be, for example, cosine similarity.

[0089] In step 1304 (f1304), the similarity values ​​with the most frequently used attribute stored in the similarity column 704 with the most frequently used attribute in the attribute similarity information data table 701 (FIG. 7) are ranked in descending order (similarity ranking), and stored in the similarity ranking column 705. Note that the similarity ranking of the row corresponding to the most frequently used attribute (in FIG. 7, the row where "age" is "30-39" and "gender" is "male") is set to first.

[0090] In step 1305 (f1305), the degree of agreement between the usage history attributes ranked in the top N by usage rate and the usage history attributes ranked in the top N by similarity, and the degree of agreement between the usage history attributes ranked in the bottom N by usage rate and the usage history attributes ranked in the bottom N by similarity are calculated, and the average of the degrees of agreement between the usage history attributes ranked in the top N and the bottom N by similarity is calculated.

[0091] Specifically, first, the system extracts the usage history attributes stored in the User column group 602 from rows in which the values ​​in the usage rate ranking column 605 in the service usage rate data table 601 (FIG. 6) are in the top 1 to N positions. The system compares the extracted N types of usage history attributes with the usage history attributes stored in the User column group 702 corresponding to rows in the attribute similarity information data table 701 (FIG. 7) in which the values ​​in the similarity ranking column 705 are in the top 1 to N positions, and calculates the percentage of matches among the N types.

[0092] Similarly, for the bottom 1 to N usage history attributes, the percentage of matches among the N types is calculated by comparing the usage history attributes whose values ​​in the usage rate ranking column 605 of the service usage rate data table 601 (Figure 6) are bottom 1 to N with the usage history attributes whose values ​​in the similarity ranking column 705 of the attribute similarity information data table 701 (Figure 7) are bottom 1 to N. Here, N may be set by the user, or may be a value determined according to the number of types of usage record attributes.

[0093] In step 1306 (f1306), it is determined whether the degree of match calculated in step 1305 (f1305) is higher than a threshold, and if it is higher than the threshold, the result is "Yes," otherwise, the result is "No." Here, this threshold may be set by the user, or may be a value determined according to the number of types of usage history attributes.

[0094] Next, returning to FIG. 12, in step 1204 (f1204), if the judgment in step 1203 (f1203) is "Yes," the estimated impact on the use of the target service for each attribute value of the candidate attribute item is calculated as the "distribution of impact on service use," and this calculation result is stored in the estimated impact column 905 of the service usage impact distribution data table 901 (FIG. 9) in the service feature storage unit 312.

[0095] Specifically, for each attribute value of the candidate attribute item, in a row where the attribute values ​​stored in each column of the User column group 502 and 503 in the statistical information data table 501 (FIG. 5) match the usage history attribute stored in the User column group 602 of the row in which the value of the usage rate ranking column 605 in the service usage rate data table 601 (FIG. 6) is 1st to Nth, the values ​​stored in the corresponding attribute value columns 505 to 508 are weighted by the usage rate value stored in the usage rate column 604 of the row corresponding to each usage history attribute in the service usage rate data table 601 (FIG. 6) and averaged.

[0096] If there is a bias in the number of people who have each attribute value of the candidate attribute item among all residents, that bias may be taken into consideration. For example, the value calculated for each attribute value of the candidate attribute item using the above method may be further compared to the proportion of people who have the corresponding attribute value among all residents, or the difference between the distribution of the candidate attribute item for each attribute (the value stored in the corresponding attribute value columns 505 to 508) and the corresponding candidate attribute item among all residents may be weighted by the value of each usage rate and then averaged. At this time, the standard deviation when calculating this weighted average may be calculated and stored in the standard deviation column 906 of the service usage impact distribution data table 901 (FIG. 9).

[0097] In step 1205 (f1205), it is determined whether the estimated influence of each attribute value of the candidate attribute item is a singular value based on the "distribution of influence on service usage" calculated in step 1204 (f1204). If the determination result is a singular value (Yes), the process proceeds to step 1206 (f1206); if not (No), the process proceeds to step 1207 (f1207).

[0098] In step 1206 (f1206), the corresponding attribute value is calculated as an "attribute value that influences the use of the service" and stored in the attribute value column 804 of the service usage feature information data table 801 (FIG. 8) in the service feature storage unit 312. As a method for determining whether or not a value is a singular value, for example, it is possible to determine whether or not the value of the estimated influence of each attribute value, or the value obtained by subtracting the standard deviation from the estimated influence, is larger than the reciprocal of the number of types of attribute values ​​of the corresponding candidate attribute item.

[0099] In step 1207 (f1207), it is determined whether all the candidate attribute items extracted in step 1007 (s1007) have been selected, and if "Yes", the process proceeds to step 1208 (f1208), and if "No", the process returns to step 1202 (f1202).

[0100] In step 1208 (f1208), the "distribution of influence on service usage" and the "attribute value that influences service usage" are calculated for each usage record attribute item.

[0101] Here, the detailed flow of step 1208 (f1208) will be explained. 14 shows the detailed flow of step 1208 (s1208), in which the analysis server 102 calculates the "distribution of the impact on service usage" and the "attribute values ​​that impact service usage" for the target service. In the following, the analysis server 102 is the processing entity for each step, so the entity will not be described.

[0102] In step 1401 (f1401), if there are multiple usage history attribute items, one of the usage history attribute items that has not been selected in step 1208 (f1208) is selected.

[0103] In Step 1402 (f1402), the service usage characteristics information is stored in the service characteristics storage unit 312 in the service usage characteristics information data table 801 (FIG. 8).

[0104] In step 1403 (f1403), the service usage rate is calculated based on only the selected usage record attribute items. Specifically, similar to step 1012 (s1012), the service usage rate of the target service is calculated based on the information in the usage record data table 401 (FIG. 4) of the target service in the usage record DB 104 and the number of people information data table 511 (FIG. 5) in the statistical information DB 105.

[0105] At this time, the usage history data table 401 and the number of people information data table 511 are re-aggregated using only the selected usage history attribute items. For example, if the usage history attribute items are "age" and "sex" and "age" is selected in step 1401 (f1401), the number of people or the total number of people for all attribute values ​​related to "sex" are re-aggregated for each attribute value of "age".

[0106] In step 1404 (f1404), based on the value of the service usage rate calculated in step 1403 (f1403), the "distribution of the impact on service usage" is calculated as the estimated impact for each attribute value of the selected usage history attribute item, and this calculation result is stored in the service usage impact distribution data table 901 (Figure 9) of the service feature storage unit 312. Specifically, the estimated impact of a certain attribute value of the selected usage history attribute item is calculated by dividing the value of the service usage rate for the corresponding attribute value of the selected usage history attribute item by the sum of the values ​​of the service usage rates for all attribute values ​​of the selected usage history attribute item.

[0107] In step 1405 (f1405), similarly to step 1205 (f1205) and step 1206 (f1206), it is determined whether the estimated impact of each attribute value of the corresponding usage record attribute item is a singular value based on the "distribution of impact on service usage" calculated in step 1404 (f1404), and if the determination result is a singular value (YES), proceed to step 1406 (f1406); if not (No), return to step 1401 (f1401).

[0108] In step 1406 (f1406), the attribute value that takes a singular value is calculated as the "attribute value that affects the use of the service" and stored in the service use characteristic information data table 801 (FIG. 8) of the service characteristic storage unit 312.

[0109] Finally, in step 1407 (f1407), it is determined whether all the usage record attribute items have been selected, and if "Yes", the process ends, and if "No", the process returns to step 1401 (f1401).

[0110] Returning to the processing flow of FIG. 10, the processing flow from step 1014 (s1014) onwards will be explained. In step 1014 (s1014), the analysis server 102 extracts the values ​​of the attribute item column 803 (attribute items related to the use of the service) and the attribute value column 804 (attribute values ​​that affect the use of the service) of the row corresponding to the target service based on the contents of the service usage characteristic information data table 801 (Figure 8) stored in the service characteristic memory unit 312, as candidate attribute information for recommendation.

[0111] The processing of step 1012 (s1012) and step 1013 (s1013) is periodically performed for all services for which the usage record data table 401 (FIG. 4) exists in the usage record DB 104, and the results are stored and updated in the service usage characteristic information data table 801 (FIG. 8) in the service characteristic storage unit 312. As a result, if there is no additional statistical information that the user wants to use in step 1010 (s1010), it is possible to omit step 1012 (s1012) and step 1013 (s1013) and perform step 1014 (s1014) based on the contents of the service usage characteristic information data table 801, which is periodically updated.

[0112] In step 1015 (s1015), the analysis server 102 transmits to the user terminal 101 the attribute information extracted in step 1014 (s1014) that is a candidate for recommendation.

[0113] Finally, in step 1016 (s1016), the user terminal 101 outputs the attribute information of the candidates for recommendation received from the analysis server 102 and presents it to the user of the user terminal 101.

[0114] Next, the result screen of the user terminal 101 will be described. FIG. 15 is a diagram showing an example of the configuration of a result screen 1501 that presents attribute information that is a candidate for recommending a target service in the first embodiment. A result screen 1501 is displayed in a window of a web browser or the like, and includes a target service display area 1502 and a recommendation attribute confirmation area 1503 .

[0115] In the target service display area 1502, information on the target service input by the user in step 1001 (s1001) is displayed. The recommendation attribute confirmation area 1503 displays attribute information that is a candidate for recommendation and that the user terminal 101 has received from the analysis server 102. For each attribute that is a candidate for recommendation, the estimated influence value, the number of attribute values ​​in the corresponding attribute item, and other attribute values ​​are presented to the user.

[0116] In Example 1, by combining and analyzing statistical data on residents held by local governments and usage history data held by service providers in a form that does not identify individuals who have used their services, it is possible to indicate attribute information that is a candidate for recommending the target service, in addition to attribute information not included in usage history.

[0117] Furthermore, this analysis can be useful for developing effective local policies without having to understand the services used by each resident. Furthermore, by providing feedback to service providers, for example, the results of this analysis can enable the service providers to consider improvements and new ideas for the services they provide. [Example]

[0118] Example 2 is a collaborative data analysis system and a collaborative data analysis method for, when a user wants to know candidate services to be recommended for a target attribute, calculating candidate services to be recommended using usage history of the candidate services to be recommended (candidate services) and statistical information of residents, and presenting information on the calculated candidate services to be recommended to the user. In the following description of the second embodiment, the parts common to the first embodiment will be omitted and only the different parts will be described. In particular, the same effects and advantages due to the same configurations will not be described one by one.

[0119] As in Example 1, the user who is the main user of the user terminal is, for example, a local government official or other person who intends to implement some kind of local policy, such as promoting resident activity or revitalizing the area, by using recommendations to residents.

[0120] In Example 2, it is assumed that the system will be used to find out which services are candidates for recommendation for certain attributes in order to promote local activities for residents with certain attributes, in line with the purpose of the policy.

[0121] FIG. 16 is a block diagram showing an example of the hardware configuration of the analysis server 102 according to the second embodiment of the present invention. The analysis server 102 can be realized by a general server device, and includes a target attribute information storage unit 1601, a service aptitude information calculation unit 1602, a service aptitude information storage unit 1603, and a recommendation service calculation unit 1604 in addition to the components of the first embodiment shown in Fig. 3, namely, a control unit 301, a network interface unit 302, a web service providing unit 303, a data access unit 304, a data registration unit 305, an input information registration unit 306, an input information storage unit 307, a service usage rate calculation unit 308, a service usage rate storage unit 309, a service feature calculation unit 310, an attribute similarity information storage unit 311, and a service feature storage unit 312. The individual units are connected by a BUS 314 or the like.

[0122] The target attribute information storage unit 1601 stores some or all of the statistical information regarding the target attributes extracted by the data access unit 304 from the statistical information DB 105 of the DB server 103 based on the information received from the user terminal 101 and the information stored in the input information storage unit 307.

[0123] The target attribute information storage unit 1601 and the service aptitude information storage unit 1603 (described later) may be configured with a plurality of storage units each having a different storage method depending on the size and format of the data to be stored.

[0124] In addition, some or all of the data equivalent to the data stored in the target attribute information storage unit 1601 and the service suitability information storage unit 1603 described later may be provided by multiple other servers connected to the network 106 such as the Internet.

[0125] The service suitability information calculation unit 1602 is a program that has the function of calculating the suitability (service suitability information) of each service for the target attribute based on the information in the service feature memory unit 312 and the target attribute information memory unit 1601, and registering the service suitability information in the service suitability information memory unit 1603.

[0126] The service suitability information storage unit 1603 stores part or all of the information relating to the service suitability information calculated by the service suitability information calculation unit 1602 .

[0127] The recommended service calculation unit 1604 is a program having a function of calculating candidate services to be recommended for the target attribute based on the information in the service suitability information storage unit 1603 and transmitting the calculated services to the user terminal 101 .

[0128] Next, various data tables used in the twenty-first embodiment will be described with reference to FIGS. <Subject Attribute Statistical Information Data Table 1701>

[0129] FIG. 17 is a diagram showing an example of the contents of the target attribute statistical information data table 1701. As shown in FIG. The target attribute statistical information data table 1701 is stored in the target attribute information storage unit 1601 of the analysis server 102 , and is composed of a target attribute column group 1702 , an attribute item column 1703 , an attribute value column 1704 , and a ratio column 1705 . The target attribute column group 1702 is composed of columns for each attribute item in the target attribute received from the user terminal 101 .

[0130] For example, if the target attributes are "age" of "40-49" and "whether or not the user has children under 15 years old" of "yes," the target attribute column group 1702 is made up of two types of attribute item columns: an age column 1706 and a whether or not the user has children under 15 years old column 1707. Each attribute item column making up the target attribute column group 1702 stores information that uniquely identifies the attribute value of the attribute item corresponding to the attribute item column, based on the information of the target attribute received from the user terminal 101.

[0131] The attribute item column 1703 stores information that can uniquely identify an attribute item. The attribute value column 1704 stores information that can uniquely identify an attribute value. The percentage column 1705 stores information about the percentage of people who have the attribute value stored in the attribute value column 1704 for the attribute item stored in the attribute item column 1703 among the target attributes. <Target attribute time series information data table 1801>

[0132] FIG. 18 is a diagram showing an example of the contents of the target attribute time-series information data table 1801. As shown in FIG. The object attribute time-series information data table 1801 is stored in the object attribute information storage unit 1601 of the analysis server 102 and is composed of an object attribute column group 1802 , an attribute item column 1803 , an attribute value column 1804 , and a time-series data column group 1805 . Similar to the target attribute column group 1702 in the target attribute statistical information data table 1701 (FIG. 17), the target attribute column group 1802 is composed of columns for each attribute item in the target attribute received from the user terminal 101. The column for each attribute item stores information that uniquely identifies the attribute value in the attribute item corresponding to the column.

[0133] The attribute item column 1803 and attribute value column 1804 store information that uniquely identifies an attribute item and information that uniquely identifies an attribute value, respectively, similar to the attribute item column 1703 and attribute value column 1704 in the target attribute statistical information data table 1701.

[0134] The time-series data column group 1805 is composed of past time-series columns, and each time-series column stores, at a corresponding time, the number of people who have the attribute value stored in the attribute value column 1804 for the attribute item stored in the attribute item column 1803 among the target attributes. For example, column 1806, which indicates the time point of November 2021, stores the number of people who meet the corresponding conditions as of November 2021. <Service Aptitude Information Data Table 1901>

[0135] FIG. 19 is a diagram showing an example of the contents of the service suitability information data table 1901. As shown in FIG. The service aptitude information data table 1901 is stored in the service aptitude information storage unit 1603 of the analysis server 102 and is composed of a service column 1902 , a match column 1903 , an increase tendency column 1904 , an aptitude column 1905 , and a recommendation priority column 1906 .

[0136] The service column 1902 stores information that uniquely identifies a service. The match degree column 1903 stores the value of the match degree in the target attribute for the service stored in the service column 1902 .

[0137] The increase tendency column 1904 stores the value of the increase tendency in the target attribute for the service stored in the service column 1902 .

[0138] The suitability column 1905 stores the suitability value for the target attribute for the service stored in the service column 1902 .

[0139] The recommendation priority column 1906 stores the priority value of the service stored in the service column 1902 to be recommended for the target attribute.

[0140] 20 is a diagram showing an example of a sequence 2001 from receiving input information from the user terminal 101 to calculating information on candidate services to be recommended for the target attribute by the analysis server 102, and displaying the information on the user terminal in Example 2. Note that, as with the explanation of Example 1, the exchange between the analysis server 102 and the DB server 103 is omitted. In step 2001 (s2001), the user terminal 101 receives input of information relating to target attributes for which candidate services to be recommended are to be calculated from the user, and transmits the input information to the analysis server 102.

[0141] In step 2002 (s2002), the analysis server receives input information relating to the target attribute from the user terminal 101, and registers the received information in the input information storage unit 307 of the analysis server .

[0142] In step 2003 (s2003), the analysis server 102 accesses the usage history DB 104 of the DB server 103 via the data access unit 304, extracts information about services for which the usage history data table 401 (FIG. 4) exists, and transmits the extracted information about the services to the user terminal 101. In FIG. 20, information about services is abbreviated as service information.

[0143] In step 2004 (s2004), the user terminal 101 receives the information sent from the analysis server 102 and outputs the received information about the service (service information).

[0144] In step 2005 (s2005), the user terminal 101 receives an input from the user regarding the selection of candidate services to be recommended (hereinafter referred to as candidate services), and transmits this input information to the analysis server 102.

[0145] In step 2006 (s2006), the analysis server receives input information related to the candidate target services from the user terminal 101, and registers the received information in the input information storage unit 307 of the analysis server .

[0146] Here, the registration screen of the user terminal 101 will be described. FIG. 21 is a diagram showing an example of the configuration of a registration screen 2101 that is presented to the user by the user terminal 101 and that is used to accept input of necessary information and data related to target attributes. A registration screen 2101 is displayed in a window of a web browser or the like, and includes a target attribute input area 2102, a send button 2105, and a candidate target service selection area 2106.

[0147] The target attribute input area 2102 is arranged with a target attribute item input area 2103 and a target attribute value input area 2104 for that target attribute item. The user inputs the attribute item and attribute value of the target attribute in the target attribute item input area 2103 and the target attribute value input area 2104, respectively, and presses the send button 2105. At this time, it is also possible to input multiple attribute items in cases where the target attribute does not have a single attribute item, such as when "age" is "40-49" and "presence of children under 15" is "yes." Upon receiving the press of the send button 2105, the user terminal 101 transmits the information on the target attribute input in the target attribute input area 2102 to the analysis server 102.

[0148] The candidate service selection area 2106 is arranged with a candidate service display area 2107, check boxes 2108, and a send button 2109. When the candidate service selection area 2106 is displayed in step 2004 (s2004), the user checks the service information sent from the analysis server 102 in step 2003 (s2003) and displayed in the candidate service display area 2107, and selects the candidate service by, for example, checking the check box 2108 of the service to be selected as the candidate service. After the candidate service is selected, the user terminal 101, which has received a press of the send button 2109, sends the information entered in the candidate service selection area 2106 to the analysis server 102.

[0149] Returning to the sequence shown in FIG. 20, the processing flow from step 2007 (s2007) onwards will be described. In step 2007 (s2007), the analysis server 102 accesses the statistical information DB 105 of the DB server 103 using the data access unit 304, extracts information related to the target attribute from the statistical information data table 501 and the number of people information data table 511 (Figure 5), and stores it in the target attribute statistical information data table 1701 (Figure 17) and the target attribute time-series information data table 1801 (Figure 18) in the target attribute information storage unit 1601.

[0150] In step 2008 (s2008), the analysis server 102 selects one unselected candidate service from the candidate services stored in the input information storage unit 307 in step 2006 (s2006).

[0151] In step 2009 (s2009), the analysis server 102 accesses the usage history DB 104 of the DB server 103 via the data access unit 304, and extracts usage history information relating to the selected candidate service.

[0152] In step 2010 (s2010) and step 2011 (s2011), the analysis server 102 performs the calculation process of the service usage rate in step 1012 (s1012) and the calculation process of the service characteristics in step 1013 (s1013) for the selected candidate target services, as described in the previous Example 1.

[0153] As in the first embodiment, the processing of step 2010 (s2010) and step 2011 (s2011) is periodically performed for all services for which the usage record data table 401 (FIG. 4) of the usage record DB 104 exists, and the results are stored and updated in the service usage characteristic information data table 801 (FIG. 8) of the service characteristic storage unit 312. In this way, step 2010 (s2010) and step 2011 (s2011) may be omitted, and the subsequent steps may be performed based on the contents of the service usage characteristic information data table 801, which is periodically updated.

[0154] In step 2012 (s2012), the analysis server 102 calculates information (aptitude information) relating to the suitability of the selected candidate target service for the target attribute based on the information in the target attribute information storage unit 1601 and the service feature storage unit 312.

[0155] As a method for performing step 2012 (s2012), two analysis methods (Analysis 1 and Analysis 2) will be described. First, the method of analysis 1 calculates the degree of match of the selected candidate target service in the target attribute as suitability information from the perspective of the similarity between the distribution of candidate attribute items in the target attribute and the "distribution of impact on service usage" of the selected candidate target service.

[0156] 22 is a diagram showing the processing flow of the step of calculating the degree of match of the selected candidate service in the target attribute as part of analysis 1. In the following, the analysis server 102 is the processing entity for each step, so the entity will not be described. In step 2201 (f2201), the "attribute items related to service usage" (attribute item column 803) of the selected candidate target service are extracted from the service usage characteristic information data table 801 (Figure 8) stored in the service characteristic storage unit 312, and one that has not been selected in this processing flow is selected from among them.

[0157] In step 2202 (f2202), for the "attribute items related to service usage" selected in step 2201 (f2201), the similarity between the distribution of statistical information data for the target attribute and the "distribution of impact on service usage" is calculated based on the information in the target attribute statistical information data table 1701 (Figure 17) of the target attribute information storage unit 1601 and the information in the service usage impact distribution data table 901 (Figure 9) of the service feature storage unit 312, and the similarity is stored in the service suitability information storage unit 1603.

[0158] For example, when the "attribute item related to service usage" is "area of ​​residence" and the corresponding attribute values ​​are four types, namely, "Area A," "Area B," "Area C," and "Area D," the distribution of statistical information data for the target attribute (when the age column 1706 of the target attribute column group 1702 shown in FIG. 17 is "40-49" and the presence / absence column 1707 of children under 15 is "yes") is expressed as (0.36, 0.19, 0.38, 0.07) according to the target attribute statistical information data table 1701 shown in FIG. 17, while the "distribution of impact on service usage" is expressed as (0.21, 0.18, 0.42, 0.19) according to the service usage impact distribution data table 901 shown in FIG. 9. Then, the similarity between these two vectors is calculated in the same manner as in step 1303 (f1303) of the first embodiment.

[0159] In step 2203 (f2203), it is determined whether or not all "attribute items related to service usage" for the selected candidate service have been selected. If "Yes," proceed to step 2204 (f2204); if "No," return to step 2201 (f2201).

[0160] In step 2204 (f2204), the degree of match of the selected candidate service in the target attribute is calculated and stored in the service suitability information storage unit 1603. Specifically, the average value of the similarities in all the "attribute items related to service usage" of the selected candidate service calculated in step 2202 (f2202) is calculated as the degree of match of the selected candidate service. According to the above analysis 1, the suitability of each service for the target attribute can be calculated taking into account the perspective of attribute items other than the attribute item selected by the user as the target attribute.

[0161] Next, the method of Analysis 2 calculates the degree of increase in the number of people who have the "attribute values ​​that affect the use of the service" of the selected candidate target service in the target attribute as suitability information. 23 is a diagram showing the processing flow of the step of calculating the degree of increase tendency of the selected candidate target service in the target attribute as part of Analysis 2. In the following, the analysis server 102 is the processing subject of each step, so the subject will not be described.

[0162] In step 2301 (f2301), the "attribute values ​​that affect the use of the service" (attribute value column 804) of the selected candidate target service are extracted from the service usage characteristic information data table 801 (Figure 8) stored in the service characteristic storage unit 312, and one that has not been selected in this processing flow is selected from among them.

[0163] In step 2302 (f2302), time series data relating to the "attribute value that affects the use of the service" selected in step 2301 (f2301) for the target attribute is extracted from the target attribute time series information data table 1801 (Figure 18) stored in the target attribute information storage unit 1601.

[0164] In step 2303 (f2303), it is verified whether a period can be detected from the time series data extracted in step 2302 (f2302). One method for detecting a period is to use an autocorrelation function, for example. If a period can be detected (Yes), the process proceeds to step 2304 (f2304); if a period cannot be detected (No), the process proceeds to step 2305 (f2305).

[0165] In step 2304 (f2304), the degree of increase periodicity is calculated based on the change trend for the period going back the detected cycle from the present. Specifically, if the time series data is divided into months, the change trend for the period (month n) going back the detected cycle from the present can be calculated. For example, using difference, the difference between the number of people in the month one month after month n (month (n+1)) minus the number of people in month n, and the difference between the number of people in the month two months after month n (month (n+2)) minus the number of people in month (n+1) can be calculated. If both are positive, the period is considered to be a period of periodic increase, and the degree of increase periodicity is set to 1. If either is negative, the degree of increase periodicity can be set to 0. Alternatively, the peak period can be detected, and a determination can be made as to whether the current period is approaching the maximum peak, and the degree of increase periodicity can be calculated based on the determination result.

[0166] In step 2305 (f2305), the degree of increasing periodicity is set to 0, assuming that the period cannot be detected.

[0167] In step 2306 (f2306), the most recent increase rate is calculated based on the average value of the increase in the number of people for each unit of time in the most recent period. Here, if the average value of the increase in the number of people is positive, the most recent increase rate is the ratio of the average value to the most recent number of people with the corresponding attribute value, and if the average value of the increase in the number of people is negative, the rate is 0. For example, if the unit of time is "month," the increase in the number of people is calculated for each of the most recent m months, and the calculated values ​​are averaged. In this case, the most recent period (m months) for which the average value is calculated may be fixed in advance or may be set by the user.

[0168] In step 2307 (f2307), the average value of the increase periodicity calculated in step 2304 (f2304) or step 2305 (f2305) and the most recent increase degree calculated in step 2306 (f2306) is calculated and used as the increase tendency degree of the selected "attribute value affecting service usage."

[0169] In step 2308 (f2308), it is determined whether or not all "attribute values ​​that affect the use of the service" have been selected for the candidate service in question. If the answer is "Yes," proceed to step 2309 (f2309); if the answer is "No," return to step 2301 (f2301).

[0170] In step 2309 (f2309), the degree of increase tendency of the selected candidate target service in the target attribute is calculated and stored in the service suitability information storage unit 1603. Specifically, for the selected candidate target service, the average value of the degree of increase tendency of all the "attribute values ​​that affect the use of the service" calculated in step 2307 (f2307) is calculated as the degree of increase tendency of the candidate target service. According to the above analysis 2, the suitability of each service for the target attribute can be calculated based on the time series characteristics of the target attribute.

[0171] As described above, the method for implementing step 2012 (s2012) may be either the above-described analysis 1 or analysis 2, or a combination of both. Alternatively, a method may be used in which the service usage rate data table 601 (FIG. 6) is used to use the usage rate of attributes included in the User column group among the target attributes for each candidate target service as information related to the suitability of each service for the target attribute, or this method may be combined with analysis 1 or analysis 2. The suitability value is determined from the suitability value obtained based on the selected analysis method for implementing step 2012 (s2012), i.e., the match value if analysis 1 is selected, or the increase tendency value if analysis 2 is selected.

[0172] Returning to the sequence shown in FIG. 20, the processing flow from step 2013 (s2013) onwards will be described. In step 2013 (s2013), the analysis server 102 determines whether or not all candidate services have been selected, and if "Yes", proceeds to step 2014 (s2014), and if "No", returns to step 2008 (s2008).

[0173] In step 2014 (s2014), the analysis server 102 calculates the service suitability information for all candidate services, and then assigns recommendation priorities to each candidate service based on the contents of the service suitability information data table 1901 (FIG. 19).

[0174] In step 2015 (s2015), the analysis server 102 transmits to the user terminal 101 the suitability value and recommendation priority order information (abbreviated as service information in FIG. 20) calculated for each candidate service.

[0175] In step 2016 (s2016), the user terminal 101 receives information on each candidate target service from the analysis server 102, and from this information outputs the minimum and maximum suitability values ​​as a range of suitability values, and accepts input of the suitability threshold value from the user.

[0176] In step 2017 (s2017), the user terminal 101 outputs information on the suitability and recommendation priority of the services whose suitability values ​​are greater than the suitability threshold value received as input, as candidates for recommendation for the target attribute, and presents this information to the user.

[0177] Next, the result screen of the user terminal 101 will be described. FIG. 24 is a diagram showing an example of the configuration of a result screen 2401 that presents information about services that are candidates for recommendation for a target attribute in the second embodiment. A result screen 2401 is displayed in a window of a web browser or the like, and includes a target attribute display area 2402, a threshold confirmation area 2403, and a recommendation service confirmation area 2407.

[0178] In the target attribute display area 2402, information on the target attribute input by the user in step 2001 (s2001) is displayed. In the threshold confirmation area 2403, an appropriateness range display area 2404, a threshold input area 2405, and a setting button 2406 are arranged.

[0179] The user of the user terminal 101 checks the range of suitability values ​​of the candidate services from the values ​​displayed in the suitability range display area 2404, and inputs the suitability threshold in the threshold input area 2405. After that, when the setting button 2406 is pressed, the user terminal 101 outputs information about services, among the candidate services, whose suitability is greater than the value input in the threshold input area 2405, to the recommended service confirmation area 2407.

[0180] The recommendation priority value, service name, and service suitability for each service that is a candidate for recommendation are displayed in the recommended service confirmation area 2407. In other words, information about the candidate services for recommendation is effectively presented to the user.

[0181] In Example 2, by combining and analyzing the statistical data of residents held by a local government and the usage history data of multiple service providers in a form that does not identify individuals who have used each service, it becomes possible to analyze candidate services to be recommended for target attributes from the perspective of attributes other than those not included in the usage history. This has the effect of leading to effective recommendations to residents without knowing the services used by each resident.

[0182] It should be noted that the present invention is not limited to the above-described first and second embodiments. It is not necessarily limited to those having all of the configurations described above. It is also possible to replace part of the configuration of one embodiment with the configuration of another embodiment, or to add the configuration of another embodiment to the configuration of one embodiment. It is also possible to add, delete, or replace part of the configuration of each embodiment with another configuration. [Explanation of symbols]

[0183] 1: Collaborative data analysis system, 101: User terminal, 102: Analysis server, 103: DB server, 104: Usage record DB, 105: Statistical information DB, 201, 301: control unit, 202, 302: network interface unit, 203: Input section, 204: Output section, 205, 314: BUS (bus), 303: Web service providing unit, 304: Data access unit, 305: Data registration unit, 306: Input information registration unit, 307: Input information storage unit, 308: Service utilization rate calculation unit, 309: Service usage rate storage unit, 310: Service feature calculation unit, 311: attribute similarity information storage unit, 312: service feature storage unit, 313: Recommendation attribute calculation unit, 401: Usage history data table, 501: statistical information data table, 511: number of people information data table, 601: Service usage rate data table, 701: Attribute similarity information data table, 801: Service usage characteristics information data table, 901: Service usage impact distribution data table, 1601: Target attribute information storage unit, 1602: service aptitude information calculation unit, 1603: service aptitude information storage unit, 1604: Recommendation service calculation unit, 1701: Target attribute statistical information data table, 1801: Target attribute time series information data table, 1901: Service suitability information data table

Claims

1. A terminal used by a user, a server that performs data analysis, and a database that stores usage history data, which is data configured in a form that does not identify individuals who have used the provided service, and public data on residents, which is statistical information data related to attributes that make up the usage history data, are connected to each other via a network; The terminal transmits input information related to a service targeted by the user to the server, and outputs attribute information that is a candidate for recommending the targeted service received as an output from the server; The server a function of extracting the usage record data and the statistical information data related to the target service from the database based on the input information; a function of calculating a usage rate of the target service based on the extracted usage record data and statistical information data; a function of calculating attribute items related to the use of the target service based on the calculated utilization rate and the extracted statistical information data; a function of calculating a distribution of the influence of each attribute item on the use of the target service based on each attribute value for each attribute item; a function of calculating, based on a distribution of the influence on the use of the target service, an attribute value for which the estimated influence degree of each attribute value is a singular value as the attribute value that influences the use of the target service; a processing unit having a function of outputting the calculated attribute value that influences the use of the target service and the attribute item having the attribute value as attribute information that is a candidate for recommending the target service; and a storage unit that stores each piece of information calculated by the processing unit through each of the functions; Equipped with A collaborative data analysis system characterized by:

2. The linked data analysis system according to claim 1, The function of calculating attribute items related to the use of the target service possessed by the processing unit is a function of calculating attribute items related to the use of the target service based on the similarity of the distribution of attribute items present in the statistical information data with respect to attributes having a high usage rate of the target service among the attribute items related to the usage history included in the usage history data. A collaborative data analysis system characterized by:

3. The linked data analysis system according to claim 1, The function of the processing unit for calculating the distribution of the influence on the use of the target service is a function for calculating the distribution of the influence on the use of the target service, with respect to attributes with high usage rates extracted from the usage rate of the target service, based on the usage rate of the service and the distribution of attribute items related to the use of the target service that exist in the statistical information data. A collaborative data analysis system characterized by:

4. A terminal used by a user, a server that performs data analysis, and a database that stores usage history data, which is data configured in a form that does not identify individuals who have used the provided service, and public data on residents, which is statistical information data related to attributes that make up the usage history data, are connected to each other via a network; the terminal transmits input information related to the target service to the server via the network; The server extracting the usage record data and the statistical information data related to the target service from the database based on the input information; Calculating a usage rate of the target service based on the extracted usage history data and statistical information data; Calculating attribute items related to the use of the target service based on the calculated usage rate and the extracted statistical information data; calculating a distribution of the influence of each attribute item on the use of the target service from each attribute value for each attribute item related to the use of the target service; calculating, based on a distribution of the influence on the use of the target service, an attribute value for which the estimated influence degree of each attribute value is a singular value as the attribute value that influences the use of the target service; outputting the calculated attribute value that influences the use of the target service and the attribute item having the attribute value as attribute information that is a candidate for recommending the target service; The terminal receives, via the network, attribute information that is a candidate for recommending the target service output by the server, and presents the information to the user. A collaborative data analysis method characterized by:

5. The linked data analysis method according to claim 4, The server calculates attribute items related to the use of the target service based on a similarity between a distribution of attribute items present in the statistical information data and an attribute that has a high usage rate of the target service among the attribute items related to the usage history included in the usage history data. A collaborative data analysis method characterized by:

6. The linked data analysis method according to claim 4, The server calculates a distribution of the influence on the use of the target service, for attributes with high usage rates extracted from the usage rate of the target service, based on the usage rate of the service and a distribution of attribute items related to the use of the target service that exist in the statistical information data. A collaborative data analysis method characterized by:

Images