Attribute determination device

Abstract

An attribute determination device according to an embodiment of the present disclosure comprises an acquisition unit, a classification unit, and a first determination unit. The acquisition unit acquires a plurality of pieces of history data corresponding one-to-one to a plurality of users who have purchased a first product, and a plurality of pieces of attribute data corresponding one-to-one to the plurality of users. Each of the plurality of pieces of history data expresses the behavior history of the corresponding user. Each of the plurality of pieces of attribute data expresses an attribute of the corresponding user. The classification unit classifies the plurality of users into one or more classes by clustering the plurality of users on the basis of the plurality of pieces of history data. The one or more classes include a first class. The first determination unit determines an attribute which is common to a plurality of first users, on the basis of a plurality of pieces of first attribute data which correspond one-to-one to the plurality of first users, who belong to said first class, from among the plurality of pieces of attribute data.

Description

Attribute Determination Apparatus

[0001] The present disclosure relates to an attribute determination apparatus.

[0002] In the formulation of marketing strategies, it is common practice to analyze, based on product purchase history, what kind of user attributes characterize the purchasers of a product. In marketing strategy formulation, it is also important to appropriately set the attributes of users that a business intends to newly acquire. Conventionally, analysis of product purchase history and identification of attributes of users who have purchased the product have been performed manually, requiring a great deal of labor. For this reason, various technologies for supporting marketing strategy formulation have been proposed, and one example is the technology disclosed in Patent Document 1. Patent Document 1 discloses a technology for, in broadcast stream transmission in which a stream is transmitted via broadcasting, creating a population with common viewer attributes based on viewer attributes such as "favorite program genre, day of the week, and time slot", and analyzing the viewing behavior of this population.

[0003] Japanese Patent No. 7185091

[0004] According to the technology disclosed in Patent Document 1, although it is possible to grasp population attributes related to stream viewing, such as "frequently watched channels, days of the week, and time slots", it is not possible to grasp attributes that are not directly related to stream viewing, such as hobbies, preferences, or values, for example.

[0005] An object of the present disclosure is to provide a technology that enables a broad grasping of attributes of users who have purchased a product based on the product's purchase history.

[0006] An attribute determination device according to one aspect of the present disclosure includes: an acquisition unit that acquires a plurality of history data corresponding one-to-one to a plurality of users who have purchased a first product, and a plurality of attribute data corresponding one-to-one to the plurality of users, each of which of the plurality of history data indicates the history of the corresponding user's actions, and each of the plurality of attribute data indicates the attributes of the corresponding user; a classification unit that classifies the plurality of users into at least one class by clustering the plurality of users based on the plurality of history data, the at least one class including a first class; and a first determination unit that determines attributes common to the plurality of first users based on a plurality of first attribute data that correspond one-to-one to a plurality of first users belonging to the first class from among the plurality of attribute data.

[0007] According to this disclosure, it is possible to broadly understand the attributes of users who purchased a product based on their purchase history.

[0008] This is a block diagram showing the configuration of a system 1 including an attribute determination device 20 according to one embodiment of the present disclosure. This is a diagram illustrating a method for determining attribute data D3. This is a block diagram showing the configuration of the attribute determination device 20. This is a diagram showing an example of text T1 represented by text data input to the first learning model MDL1, a prompt P1 input to the first learning model MDL1, and a summary A1 output from the first learning model MDL1. This is a diagram showing an example of an analysis result screen G1 displayed on the display device 240 of the attribute determination device 20. This is a flowchart showing the processing flow in the first determination method executed by the processing device 230 according to program PR1. This is a flowchart showing the processing flow in the second determination method executed by the processing device 230 according to program PR1.

[0009] A. Embodiment Figure 1 is a diagram showing an example configuration of a system 1 including an attribute determination device 20 according to one embodiment of the present disclosure. System 1 is a computer system operated by a telecommunications carrier (hereinafter referred to as carrier X) that provides communication services via telecommunications lines such as the Internet. In addition to the attribute determination device 20, system 1 includes a management device 10 connected to the attribute determination device 20 via a communication network NW. The communication network NW may include the Internet, or it may be a communication network that does not include the Internet, such as a LAN (Local Area Network). In this embodiment, the management device 10 and the attribute determination device 20 are wiredly connected to the communication network NW, but they may be wirelessly connected. The attribute determination device 20 communicates with the management device 10 via the communication network NW.

[0010] The management device 10 stores multiple subscriber data D1, each corresponding to a specific user who has entered into a service agreement with carrier X regarding communication services. The subscriber data D1 includes identification data that uniquely identifies the corresponding user, and data representing the user's name, age, and gender. An example of the identification data is data representing the line number. The management device 10 also stores multiple history data D2, each corresponding to a specific user who uses the communication services provided by carrier X. Although not shown in detail in Figure 1, the history data D2 is associated with the identification data of the corresponding user. The history data D2 represents the history of the corresponding user's actions. In this embodiment, the history data D2 includes payment data showing the corresponding user's payment history, location data showing the corresponding user's location history, and log data showing the application usage history. The payment history includes the purchase history of goods. When the history data D2 represents multiple types of history, each history represents the history of actions during a common period. In other words, if the historical data D2 represents multiple types of history, the time axes of each history are the same.

[0011] Furthermore, the management device 10 stores multiple attribute data D3 that correspond one-to-one with multiple users who use the communication service provided by business operator X. In Figure 1, detailed illustrations are omitted, but the attribute data D3 is associated with the identification data of the corresponding user. In this embodiment, the attribute data D3 is text data that indicates the attributes of the corresponding user. User attributes represent at least one of the following: the user's age, gender, hobbies, preferences, and thinking tendencies. In this embodiment, the attribute data D3 is text data that expresses the user's age, gender, hobbies, and thinking tendencies in sentences. This text data is generated based on the corresponding user's history data D2.

[0012] More specifically, the management device 10 associates subscriber data D1 and history data D2 with each of several users who have already concluded a service agreement with carrier X regarding communication services, where the identification data matches. Note that multiple history data D2 may be associated with a single subscriber data D1. Next, the management device 10 analyzes one or more history data D2 associated with the subscriber data D1 to determine a word representing the user's attributes corresponding to the subscriber data D1. Existing technologies may be used as appropriate for determining the attribute words based on the history data D2. Then, the management device 10 associates the identification data with the name, age, and gender represented by the corresponding subscriber data D1, and with the words representing hobbies and thinking tendencies determined based on the corresponding history data D2, to generate the table TBL shown in Figure 2. Figure 2 shows that the person whose identification data is "124285" is named "Mejiro Shiro", is "29 years old", is "male", enjoys cooking, and has a tendency towards "honesty".

[0013] Next, the management device 10 generates attribute data D3 corresponding to the identification data "124285" based on the contents stored in table TBL, using a rule-based approach. Rule-based means, for example, applying the information represented by each data stored in table TBL in association with the identification data to S1 to S5 in template TP of a sentence such as "S1 is S2 years old and S3, and has the characteristics of hobbies S4 and S5." S1 in template TP is a variable corresponding to "name" in table TBL. S2 in template TP is a variable corresponding to "age" in table TBL. S3 in template TP is a variable corresponding to "gender" in table TBL. S4 in template TP is a variable corresponding to "hobbies" in table TBL. S5 in template TP is a variable corresponding to "thinking tendencies" in table TBL. For example, the attribute data D3 generated by the rule-based system based on the table TBL shown in Figure 2 would be: "Shiro Mejiro is a 29-year-old male with the characteristics of cooking as a hobby and honesty."

[0014] The attribute determination device 20 is a device that enables the determination of the attributes of a user who has purchased a certain product (hereinafter referred to as product A) based on a plurality of history data D2 stored in the management device 10 and a plurality of attribute data D3 stored in the management device 10. The following describes the case in which product A is purchased by a large number of users. Product A is, for example, a nutritional supplement. Product A is an example of the first product in this disclosure. A user who has purchased product A is referred to as the first user. History data representing the history of the first user's actions is referred to as the first history data. Attribute data corresponding to the first user is referred to as the first attribute data.

[0015] Figure 3 shows an example of the configuration of the attribute determination device 20. As shown in Figure 3, the attribute determination device 20 includes a communication device 210, a storage device 220, a processing device 230, a display device 240, an input device 250, and a bus 260 that mediates the exchange of data between these components.

[0016] The communication device 210 includes a communication circuit that is connected to a communication network NW wirelessly or by wire. The communication device 210 communicates with other devices connected to the communication network NW. A specific example of other communication devices that communicate with the communication device 210 via the communication network NW is the management device 10.

[0017] The display device 240 is, for example, a liquid crystal display. The display device 240 displays various images under the control of the processing device 230. The input device 250 includes a pointing device such as a mouse and multiple operators such as a numeric keypad. The input device 250 receives operations from an analyst who analyzes attributes common to multiple users who have purchased product A using the attribute determination device 20, and outputs data representing the content of said operations to the processing device 230. By outputting data representing the content of the analyst's operations from the input device 250 to the processing device 230, the content of the operations performed by the analyst is transmitted to the processing device 230.

[0018] The storage device 220 is a recording medium readable by the processing device 230. The storage device 220 includes, for example, non-volatile memory and volatile memory. Non-volatile memory includes, for example, ROM (Read Only Memory), EPROM (Erasable Programmable Read Only Memory), and EEPROM (Electrically Erasable Programmable Read Only Memory). The non-volatile memory stores a first learning model MDL1, a second learning model MDL2, and a program PR1. Program PR1 is a program that causes the processing device 230 to execute processing that prominently demonstrates the features of this disclosure. The first learning model MDL1 and the second learning model MDL2 are machine learning models used by the processing unit 230 during the execution of program PR1. Details of the first learning model MDL1 and the second learning model MDL2 will be revealed later. The volatile memory is, for example, RAM (Random Access Memory). The non-volatile memory stores the first learning model MDL1, the second learning model MDL2, and program PR1. The volatile memory is used by the processing unit 230 as a work area when executing program PR1.

[0019] The processing unit 230 includes one or more CPUs (Central Processing Units). One or more CPUs are examples of one or more processors. Each of the processors and CPUs is an example of a computer. The processing unit 230 reads program PR1 from the storage device 220, for example, when the attribute determination device 20 is powered on. By executing the read program PR1, the processing unit 230 functions as the acquisition unit 230a, classification unit 230b, first determination unit 230c, display control unit 230d, reception unit 230e, and second determination unit 230f shown in Figure 3.

[0020] The acquisition unit 230a acquires multiple historical data D2 from the management device 10 by communicating with the management device 10 using the communication device 210. The acquisition unit 230a also acquires multiple attribute data D3 by communicating with the management device 10 using the communication device 210. The multiple historical data D2 acquired by the acquisition unit 230a includes multiple first historical data. The multiple attribute data D3 acquired by the acquisition unit 230a includes multiple first attribute data.

[0021] The classification unit 230b classifies multiple users into at least one class by clustering multiple users that correspond one-to-one to the multiple historical data D2 acquired by the acquisition unit 230a. More specifically, the classification unit 230b first converts the multiple historical data acquired by the acquisition unit 230a into multiple feature vector data that correspond one-to-one to the multiple historical data using an Autoencoder. An Autoencoder is a mathematical model called a neural network, which is a technique that achieves feature extraction by first compressing the input data to identify the most important information and then discarding the parts that are not important. Each of the multiple feature vector data contains multiple elements. The classification unit 230b classifies multiple users into at least one class by clustering the multiple feature vector data in the feature space determined by the multiple elements. Since the aforementioned multiple first users have in common that they are users who purchased product A, the multiple first users are classified into one class by the classification unit 230b. A class to which multiple first users belong is called a first class.

[0022] The first determination unit 230c determines the attributes common to multiple first users based on multiple first attribute data that correspond one-to-one to multiple first users belonging to a first class, from among the multiple attribute data acquired by the acquisition unit 230a. The first determination unit 230c further determines the number of multiple first users based on the multiple first attribute data. Note that determining the number of first users is not necessarily required and may be omitted.

[0023] More specifically, the first determination unit 230c first determines common attributes by summarizing the concepts common to multiple first attribute data using the first learning model MDL1. Since attribute data is text data, the first attribute data may be referred to as first text data below. The first learning model MDL1 is a large-scale language model that accepts input of multiple text data and a prompt instructing it to summarize the concepts common to the multiple text data, and summarizes the concepts common to the multiple text data according to the prompt. Figure 4 shows an example of text T1 represented by any of the multiple first text data input to the first learning model MDL1, and an example of a prompt P1 input to the first learning model MDL1. When multiple first text data, including the text data representing text T1 shown in Figure 4, and the prompt P1 shown in Figure 4 are input, the first learning model MDL1 outputs the summary A1 shown in Figure 4.

[0024] The first decision unit 230c inputs a plurality of first text data and a prompt instructing it to summarize the common concepts of the plurality of first text data to the first learning model MDL1. The first decision unit 230c then determines common attributes by obtaining the summary of common concepts generated by the first learning model MDL1. If the number of first attribute data exceeds the threshold Th, the first decision unit 230c may select M first attribute data from the plurality of first attribute data using pseudorandom numbers or the like (2 ≤ M ≤ Th), and determine the common concepts of the selected M first attribute data using the first learning model MDL1.

[0025] The display control unit 230d displays on the display device 240 the attributes common to multiple first users and the number of such first users. For example, suppose the first determination unit 230c determines that attributes such as "teenager," "likes soccer," "often eats ramen," and "health-conscious" are common to multiple first users, and that there are 10 first users who have the attributes of "likes soccer" and "teenager," and 20 first users who have the attributes of "often eats ramen" and "health-conscious." In this case, the display control unit 230d displays the image of the analysis result screen G1 shown in Figure 5 on the display device 240. Through this analysis result screen G1, the analyst can grasp the attributes common to multiple users who purchased product A and the number of users who have those attributes. As mentioned above, product A is a nutritional supplement, but attributes such as "likes soccer" or "often eats ramen" are not directly related to nutritional supplements. On the other hand, the attribute "health-conscious" is directly related to nutritional supplements. Thus, according to this embodiment, it becomes possible to broadly grasp the attributes common to multiple users who have purchased product A, regardless of whether or not they are directly related to product A.

[0026] The reception unit 230e accepts the designation of a feature point, which is a single point within the feature space described above. Various methods can be considered for designating a feature point. For example, one method is to display an image of a three-dimensional space that mimics the feature space on the display device 240, and have the analyst designate a single point within that three-dimensional space by clicking with a mouse or the like, thereby designating a feature point within the feature space.

[0027] The second determination unit 230f determines text data that describes the user's attributes corresponding to a feature point, based on the feature vector data corresponding to the feature point specified by the reception unit 230e. The feature vector data corresponding to the feature point specified by the reception unit 230e is an example of the second feature vector data. The text data determined by the second determination unit 230f is an example of the second text data.

[0028] More specifically, the second decision unit inputs feature vector data corresponding to feature points in the feature space into the second learning model MDL2. The second learning model MDL2 is a machine learning model that has already learned the correspondence between feature vector data and text data that describes user attributes in sentences. The second learning model MDL2 outputs text data corresponding to the input feature vector data. The second decision unit 230f determines the text data output from the second learning model MDL2 as the second text data. By referring to this second text data, the analyst can understand the user attributes whose behavioral characteristics are represented by the specified feature points in the feature space.

[0029] Furthermore, the processing unit 230, which operates according to program PR1, executes either the first determination method or the second determination method described below, depending on the operation performed on the input device 250. In the first determination method, attributes common to multiple users who purchased product A are determined. In the second determination method, user attributes corresponding to feature points specified by the analyst are determined. In this embodiment, the second determination method becomes executable after the execution of the first determination method.

[0030] Figure 6 is a flowchart showing the processing flow in the first determination method. As shown in Figure 6, the first determination method includes the processing of steps SA110 to SA140. In step SA110, the processing unit 230 functions as an acquisition unit 230a. In step SA110, the processing unit 230 acquires a plurality of history data D2 and a plurality of attribute data D3 from the management device 10 by communicating with the management device 10 using the communication device 210.

[0031] In step SA120, which follows step SA110, the processing unit 230 functions as a classification unit 230b. In step SA120, the processing unit 230 classifies multiple users into at least one class by clustering multiple users that correspond one-to-one to the multiple historical data D2 acquired in step SA110.

[0032] In step SA130, which follows step SA120, the processing unit 230 functions as a first determination unit 230c. In step SA130, the processing unit 230 determines the common attributes of multiple first users based on multiple first attribute data that correspond one-to-one to multiple first users belonging to a first class, from among the multiple attribute data D3 acquired in step SA110. In step SA130, the processing unit 230 further determines the number of multiple first users based on the multiple first attribute data.

[0033] In step SA140, which follows step SA130, the processing unit 230 functions as a display control unit 230d. In step SA140, the processing unit 230 displays the analysis results screen G1 on the display device 240. Through the analysis results screen G1, the analyst can understand the attributes common to multiple users who purchased product A and the number of users who possess those attributes.

[0034] Figure 7 is a flowchart showing the processing flow in the second determination method. As shown in Figure 7, the second determination method includes the processing in steps SB110 and SB120. In step SB110, the processing unit 230 functions as a receiving unit 230e. In step SB110, the processing unit 230 receives the designation of a feature point, which is a point in the feature space.

[0035] In step SB120, which follows step SB110, the processing unit 230 functions as a second determination unit 230f. In step SB120, the processing unit 230 determines text data that describes the user's attributes corresponding to the feature points, based on the feature vector data corresponding to the feature points specified in step SB110. By referring to this text data, the analyst can understand the user's attributes whose behavioral characteristics are represented by the specified feature points in the feature space.

[0036] As described above, according to this embodiment, analysts can broadly grasp the attributes common to multiple users who purchased product A, regardless of whether they are directly related to product A or not, thereby reducing the effort required for attribute analysis. In addition, according to this embodiment, since the attributes of users whose behavioral characteristics are represented by specified feature points in the feature space can be grasped, it becomes possible to determine the attributes of new target users when selling product A based on the behavioral characteristics of those users.

[0037] B. Modifications The above embodiment can be modified as follows. Furthermore, the embodiment and each modification may be combined as appropriate. (1) The attribute determination device 20 may be manufactured or sold as a standalone unit, and the program PR1 may be manufactured or sold as a standalone unit. When selling the program PR1 as a standalone unit, specific methods of provision include, for example, writing the program PR1 to a computer-readable recording medium such as flash ROM and distributing it, or distributing the program PR1 by downloading it via a telecommunications line such as the Internet.

[0038] (2) In the above embodiment, the acquisition unit 230a, classification unit 230b, first determination unit 230c, display control unit 230d, reception unit 230e, and second determination unit 230f were software modules realized by operating a computer such as a CPU according to the program PR1. However, at least one of the classification unit 230b, first determination unit 230c, display control unit 230d, reception unit 230e, and second determination unit 230f may be a hardware module composed of an electronic circuit or the like.

[0039] (3) If it is sufficient to determine attributes common to multiple users who purchased product A, the reception unit 230e and the second determination unit 230f can be omitted, and the second determination process can also be omitted. In the above embodiment, the first learning model MDL1 and the second learning model MDL2 were stored in the storage device 220 of the attribute determination device 20. However, either the first learning model MDL1 or the second learning model MDL2 may be stored in a storage device accessible to the attribute determination device 20 via a communication network NW.

[0040] C: Other (1) In the above embodiment, ROM, RAM and the like are exemplified as the storage device 220, but the storage device 220 may also be a flexible disk, a magneto-optical disk (for example, a compact disc, a digital versatile disc, a Blu-ray (registered trademark) disc), a smart card, a flash memory device (for example, a card, a stick, a key drive), a CD-ROM (Compact Disc-ROM), a register, a removable disk, a hard disk, a floppy (registered trademark) disk, a magnetic strip, a database, a server, or any other suitable storage medium.

[0041] (2) In the above embodiment, the information, signals, etc. described may be represented by any of various different techniques. For example, data, instructions, commands, information, signals, bits, symbols, chips, etc., that may be referred to throughout the above description may be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, light fields or photons, or any combination thereof.

[0042] (3) In the above embodiment, input and output information and the like may be stored in a specific location (for example, a memory), or may be managed using a management table. Input and output information and the like can be overwritten, updated, or additionally recorded. Output information and the like may be deleted. Input information and the like may be transmitted to other devices.

[0043] (4) In the above embodiment, the determination may be made based on a value represented by 1 bit (0 or 1), may be made based on a boolean value (Boolean: true or false), or may be made based on numerical comparison (for example, comparison with a predetermined value).

[0044] (5) The processing procedures, sequences, flowcharts, etc. exemplified in the above embodiment may have their orders changed as long as there is no contradiction. For example, for the method described in the present disclosure, elements of various steps are presented using an exemplary order, and the method is not limited to the specific order presented.

[0045] (6) Each function illustrated in FIG. 3 is implemented by any combination of at least one of hardware and software. In addition, the method of implementing each functional block is not particularly limited. That is, each functional block may be implemented using one physically or logically coupled device, or may be implemented using two or more physically or logically separated devices connected directly or indirectly (for example, via wired connection, wireless connection, etc.), using the plurality of devices. A functional block may be implemented by combining software with the one device or the plurality of devices.

[0046] (7) Regardless of whether the program exemplified in the foregoing embodiment is called software, firmware, middleware, microcode, a hardware description language, or any other name, it should be broadly interpreted to mean instructions, an instruction set, code, a code segment, program code, a program, a subprogram, a software module, an application, a software application, a software package, a routine, a subroutine, an object, an executable file, an execution thread, a procedure, a function, and the like.

[0047] In addition, software, instructions, information, and the like may be transmitted and received via a transmission medium. For example, when software is transmitted from a website, a server, or another remote source using at least one of wired technology (coaxial cable, optical fiber cable, twisted pair, Digital Subscriber Line (DSL), etc.) and wireless technology (infrared, microwave, etc.), at least one of the wired technology and the wireless technology is included within the definition of a transmission medium.

[0048] (8) In each of the foregoing embodiments, the terms "system" and "network" are used interchangeably.

[0049] (9) In the present disclosure, the information, parameters, and the like described may be represented using absolute values, may be represented using relative values from a predetermined value, or may be represented using other corresponding information.

[0050] (10) In the embodiments described above, the portable device may be a Mobile Station (MS). A Mobile Station may also be referred to by those skilled in the art as a subscriber station, mobile unit, subscriber unit, wireless unit, remote unit, mobile device, wireless device, wireless communication device, remote device, mobile subscriber station, access terminal, mobile terminal, wireless terminal, remote terminal, handset, user agent, mobile client, client, or several other appropriate terms. In this disclosure, terms such as "mobile station," "user terminal," "user equipment (UE)," and "terminal" may be used interchangeably.

[0051] (11) In the embodiments described above, the terms “connected,” “coupled,” or any variation thereof means any direct or indirect connection or coupling between two or more elements, and may include the presence of one or more intermediate elements between two elements that are “connected” or “coupled” with each other. The coupling or connection between elements may be physical, logical, or a combination thereof. For example, “connection” may be read as “access.” As used in the present disclosure, two elements may be considered to be “connected” or “coupled” with each other using at least one of one or more wires, cables, and printed electrical connections, and, in some non-limiting and non-exclusive examples, electromagnetic energy having wavelengths in the radio frequency domain, microwave domain, and optical (both visible and invisible) domain.

[0052] (12) In the embodiments described above, the phrase "based on" does not mean "based solely on" unless otherwise specified. In other words, the phrase "based on" means both "based solely on" and "based at least on".

[0053] (13) The terms “determinating” and “deciding” as used in this disclosure may encompass a wide variety of actions. “Determinating” and “deciding” may include, for example, judging, calculating, computing, processing, deriving, investigating, looking up, search, inquiry (for example, searching in a table, database or other data structure), and confirming. Furthermore, "judgment" and "decision" may include considering something as a "judgment" or "decision" based on actions such as receiving (e.g., receiving information), transmitting (e.g., sending information), input, output, and access (e.g., accessing data in memory). Additionally, "judgment" and "decision" may include considering something as a "judgment" or "decision" based on actions such as resolving, selecting, choosing, establishing, and comparing. In short, "judgment" and "decision" may include considering something as a "judgment" or "decision" based on some action. Furthermore, "judgment (decision)" may be reinterpreted as "assuming," "expecting," or "considering."

[0054] (14) Where the terms “include,” “including,” and variations thereof are used in the embodiments described above, these terms are intended to be inclusive, as is the term “comprising.” Furthermore, the term “or” as used in this disclosure is not intended to be exclusive OR.

[0055] (15) In the present disclosure, if articles are added by translation, for example, a, an, and the in English, the present disclosure may include the fact that the noun following these articles is plural.

[0056] (16) In this disclosure, the term “A and B are different” may mean “A and B are different from each other.” The term may also mean “A and B are each different from C.” Terms such as “separate” and “combine” may be interpreted in the same way as “different.”

[0057] (17) Each aspect / embodiment described herein may be used individually, in combination, or switched between as needed during implementation. Furthermore, notification of certain information (e.g., notification that "it is X") is not limited to explicit notification, but may also be implicit (e.g., by not providing such notification).

[0058] 20...Attribute determination device, 210...Communication device, 220...Storage device, 230...Processing device, 230a...Acquisition unit, 230b...Classification unit, 230c...First determination unit, 230d...Display control unit, 230e...Reception unit, 230f...Second determination unit, MDL1...First learning model, MDL2...Second learning model, PR1...Program.

Claims

1. An attribute determination device comprising: an acquisition unit that acquires a plurality of historical data corresponding one-to-one to a plurality of users who have purchased a first product, and a plurality of attribute data corresponding one-to-one to the plurality of users, wherein each of the plurality of historical data indicates the behavioral history of the corresponding user, and each of the plurality of attribute data indicates the attributes of the corresponding user; a classification unit that classifies the plurality of users into at least one class by clustering the plurality of users based on the plurality of historical data, wherein the at least one class includes a first class; and a first determination unit that determines attributes common to the plurality of first users based on a plurality of first attribute data that correspond one-to-one to a plurality of first users belonging to the first class from among the plurality of attribute data.

2. The attribute determination device according to claim 1, wherein the first determination unit further determines the number of the plurality of first users based on the plurality of first attribute data.

3. Each of the plurality of attribute data is text data that expresses the attributes of the corresponding user in sentence form, the plurality of first attribute data is a plurality of first text data, and the first determination unit determines the common attribute by summarizing a concept common to the plurality of first text data, the attribute determination device according to claim 1.

4. The attribute determination device according to claim 3, wherein the first determination unit inputs the plurality of first text data and a prompt instructing the plurality of first text data to summarize a concept common to the plurality of first text data, and determines the common attribute by obtaining the summary of the common concept generated by the large-scale language model.

5. The attribute determination device according to claim 1, wherein the classification unit converts the plurality of historical data into a plurality of feature vector data that correspond one-to-one with the plurality of historical data, each of the plurality of feature vector data includes a plurality of elements, and the plurality of users are classified into at least one cluster by clustering the plurality of feature vector data in a feature space determined by the plurality of elements.

6. The attribute determination device according to claim 5, further comprising: a reception unit that receives the designation of a feature point which is a point in the feature space; and a second determination unit that determines second text data which indicates the user's attributes corresponding to the feature point in sentence form, based on feature vector data corresponding to the feature point.

7. The attribute determination device according to claim 6, wherein the second determination unit determines the second text data using a learned model that has learned the correspondence between feature vector data corresponding to feature points in the feature space and text data that describes the user's attributes in sentences.

8. The attribute determination device according to claim 1, further comprising a display control unit that displays attributes common to the plurality of first users on a display device.

9. The attribute determination device according to claim 1, wherein each of the plurality of historical data includes at least one of the following: payment data indicating the payment history of the corresponding user, location data indicating the location history of the corresponding user, and log data indicating the log of the application used by the corresponding user.

10. The attribute determination device according to claim 1, wherein each of the plurality of attribute data is generated based on the corresponding user's history data, and each of the plurality of attribute data represents at least one of the corresponding user's hobbies, preferences, and thinking tendencies.

Images