KR102537900B1 - Apparatus, method and recording medium storing commands for generating tags that indicate user characteristics - Google Patents

Abstract

According to one aspect of the present disclosure, an apparatus for generating a tag indicating a user characteristic may be provided. An apparatus according to the present disclosure includes one or more processors and one or more memories storing instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, and the one or more processors, according to the instructions , obtaining user information for each of a plurality of users and payment information for each of the plurality of users, and generating one or more primary tags for each of the plurality of users based on the user information and the payment information; The one or more primary tags indicate one or more primary user characteristics, respectively, and each of the one or more primary user characteristics is specified by at least one of the user information and the payment information - predetermined among the plurality of users. A similarity between each of the plurality of users and the sample user group is determined based on one or more primary tags for each of the sample users in the sample user group, and a secondary tag is determined for each of the plurality of users based on the degree of similarity. A tag may be created, wherein the secondary tag indicates secondary user characteristics associated with the sample user group.

Description

Recording medium on which devices, methods and commands for generating tags indicating user characteristics are recorded

The present disclosure relates to techniques for generating tags indicating user characteristics.

As part of marketing tasks, companies may search potential customers (ie, users) for their products or services in various ways. To search for potential customers, you may need to determine the characteristics of each customer. What characteristics each customer has can be determined by humans in a so-called heuristic way. However, the heuristic method may lack consistency in the decision process in that a person directly determines the customer characteristics, and accordingly, the reliability of the customer characteristic determination result may decrease.

This disclosure provides techniques for creating tags that indicate user characteristics.

According to one aspect of the present disclosure, an apparatus for generating a tag indicating a user characteristic may be provided. An apparatus according to the present disclosure may include one or more processors; and one or more memories in which commands for causing the one or more processors to perform operations when executed by the one or more processors are stored, wherein the one or more processors, according to the commands, user information for each of a plurality of users and Acquiring payment information for each of the plurality of users, generating one or more primary tags for each of the plurality of users based on the user information and the payment information, wherein the one or more primary tags include one or more 1 Primary user characteristics are each indicated, and each of the one or more primary user characteristics is specified by at least one of the user information and the payment information, one for each sample user in a predetermined sample user group among the plurality of users. Based on the above primary tag, a similarity between each of the plurality of users and the sample user group is determined, and based on the similarity, a secondary tag is generated for each of the plurality of users - the secondary tag is Indicate secondary user characteristics in relation to the sample user group - may.

In one embodiment, each of the one or more primary tags is configured to at least one selected from among the user information of each of the plurality of users, a payment destination, a payment item, a payment method, a payment time point, a number of payments, a payment frequency, and a payment amount. Each of the one or more primary user characteristics specified on the basis may be indicated.

In one embodiment, the user information for each of the plurality of users may include behavior information or credit information of each of the plurality of users.

In an embodiment, the one or more processors may determine the similarity based on a difference between a primary tag value for each of the plurality of users and a primary tag value for each sample user.

In an embodiment, the one or more processors determine the degree of similarity - the primary tag based on a difference between a primary tag value for each of the plurality of users and a primary tag reference value of the sample user group. The reference value may be a value determined based on the value of the primary tag for each of the sample users.

In one embodiment, the one or more memories further store a neural network-based learning model learned based on one or more primary tags for each of the sample users, and the one or more processors, the learning model Based on, a similarity between each of the plurality of users and the sample user group may be determined.

In an embodiment, the one or more processors may input values of one or more primary tags for each of the plurality of users to the learning model, and determine the degree of similarity based on an output of the learning model.

In one embodiment, the one or more processors determine a random number of users from among the plurality of users in an order of high similarity, determine a ratio of the sample user among the random number of users, and Based on the ratio, a value of the secondary tag for each of the plurality of users may be determined.

In one embodiment, the secondary tag may indicate whether each of the plurality of users has the secondary user characteristics or a degree of matching with the secondary user characteristics.

According to another aspect of the present disclosure, a method for generating a tag indicating a user characteristic may be provided. A method according to the present disclosure is performed in a computing device including one or more processors and one or more memories in which instructions to be executed by the one or more processors are stored, wherein the one or more processors, the user for each of a plurality of users obtaining information and payment information for each of the plurality of users; Generating, by the one or more processors, one or more primary tags for each of the plurality of users based on the user information and the payment information, the one or more primary tags indicating one or more primary user characteristics, respectively. and each of the one or more primary user characteristics is specified by at least one of the user information and the payment information; determining, by the one or more processors, a degree of similarity between each of the plurality of users and the sample user group, based on one or more primary tags for each sample user in a predetermined sample user group among the plurality of users; and generating, by the one or more processors, a secondary tag for each of the plurality of users based on the degree of similarity, wherein the secondary tag indicates a secondary user characteristic related to the sample user group. can

In one embodiment, each of the one or more primary tags is configured to at least one selected from among the user information of each of the plurality of users, a payment destination, a payment item, a payment method, a payment time point, a number of payments, a payment frequency, and a payment amount. Each of the one or more primary user characteristics specified on the basis may be indicated.

In one embodiment, the user information for each of the plurality of users may include behavior information or credit information of each of the plurality of users.

In an embodiment, the determining of the degree of similarity may include determining the degree of similarity based on a difference between a value of a primary tag for each of the plurality of users and a value of a primary tag for each of the sample users. can include

In an embodiment, the determining of the similarity may include determining the similarity based on a difference between a primary tag value for each of the plurality of users and a primary tag reference value of the sample user group - The primary tag reference value is a value determined based on the value of the primary tag for each of the sample users.

In one embodiment, the one or more memories further store a neural network-based learning model learned based on one or more primary tags for each of the sample users, and the determining of the similarity may include the learning and determining a degree of similarity between each of the plurality of users and the sample user group based on a model.

In an embodiment, the determining of the degree of similarity may include inputting one or more primary tag values for each of the plurality of users to the learning model; and determining the degree of similarity based on an output of the learning model.

In an embodiment, the generating of the secondary tag may include determining a random number of users from among the plurality of users in an order of high similarity; determining a proportion of the sample users among the random number of users; and determining a value of the secondary tag for each of the plurality of users based on the ratio.

In one embodiment, the secondary tag may indicate whether each of the plurality of users has the secondary user characteristics or a degree of matching with the secondary user characteristics.

According to another aspect of the present disclosure, a non-transitory computer-readable recording medium recording instructions to be executed on a computer to generate a tag indicating a user characteristic may be provided. When the instructions recorded on the recording medium according to the present disclosure are executed by one or more processors, the one or more processors acquire user information for each of a plurality of users and payment information for each of the plurality of users, and generate one or more primary tags for each of the plurality of users based on the user information and the payment information, wherein the one or more primary tags each indicate one or more primary user characteristics; Each feature is specified by at least one of the user information and the payment information - Based on one or more primary tags for each sample user in a predetermined sample user group among the plurality of users, each of the plurality of users and determine a similarity between the sample user groups, and generate a secondary tag for each of the plurality of users based on the similarity, the secondary tag indicating a secondary user characteristic related to the sample user group; can do.

According to various embodiments of the present disclosure, characteristics of a corresponding user may be objectively determined from various data on the user.

According to various embodiments of the present disclosure, it is possible to determine whether each of a plurality of users has a corresponding user characteristic based on a consistent criterion of similarity between each user and a sample user group related to a target user characteristic.

According to various embodiments of the present disclosure, since a user who will purchase a specific product or service is searched for based on a consistent criterion of similarity, the reliability of the search result can be improved.

1 is a diagram illustrating a process of operating a computing device according to an embodiment of the present disclosure.
2 is a diagram for explaining user information, payment information, behavior information, credit information, a primary tag set, and a secondary tag according to an embodiment of the present disclosure.
3 is a block diagram of a computing device 100 according to one embodiment of the present disclosure.
4 is a diagram for explaining a relationship between a plurality of users and a sample user group according to an embodiment of the present disclosure.
5 is a diagram for explaining a similarity determination process based on a neural network-based learning model according to an embodiment of the present disclosure.
6 is a diagram illustrating a process of determining a secondary tag based on similarity according to an embodiment of the present disclosure.
7 is a flowchart illustrating an operation of a computing device according to an embodiment of the present disclosure.
8 is a flowchart illustrating an operation of a computing device according to an embodiment of the present disclosure.

Various embodiments described in this document are illustrated for the purpose of clearly explaining the technical spirit of the present disclosure, and are not intended to be limited to specific embodiments. The technical spirit of the present disclosure includes various modifications, equivalents, and alternatives of each embodiment described in this document, and embodiments selectively combined from all or parts of each embodiment. In addition, the scope of the technical idea of the present disclosure is not limited to various embodiments or specific descriptions thereof presented below.

Terms used in this document, including technical or scientific terms, may have meanings commonly understood by those of ordinary skill in the art to which this disclosure belongs, unless otherwise defined.

Expressions such as "comprises", "may include", "has", "may have", "has", "may have", etc. used in this document refer to the target characteristics (e.g. function, operation or component), and does not preclude the presence of other additional features. That is, such expressions should be understood as open-ended terms that include the possibility of including other embodiments.

Expressions in the singular form used in this document may include plural meanings unless the context indicates otherwise, and this applies equally to expressions in the singular form in the claims.

As used herein, expressions such as "first", "second", or "first" or "second" refer to plural objects of the same kind, unless the context dictates otherwise, distinguishing one object from another. It is used to do so, and does not limit the order or importance between the objects. In one embodiment, each user included in the plurality of users according to the present disclosure may be distinguished from each other by being expressed as a “first user” or a “second user”. In one embodiment, a plurality of sample users included in the sample user group according to the present disclosure may be distinguished from each other by being expressed as "first sample user" and "second sample user". In one embodiment, one or more primary tags according to the present disclosure may be distinguished from each other by being expressed as “first primary tag” or “second primary tag”.

As used herein, "A, B, and C", "A, B, or C", "A, B, and/or C" or "at least one of A, B, and C", "A, B Expressions such as “at least one of , or C”, “at least one of A, B, and/or C” may mean each listed item or all possible combinations of the listed items. For example, “at least one of A or B” can refer to (1) at least one A, (2) at least one B, and (3) both at least one A and at least one B.

The expression "unit" used in this document may mean software or a hardware component such as a field-programmable gate array (FPGA) or an application specific integrated circuit (ASIC). However, "unit" is not limited to hardware and software. A "unit" may be configured to be stored in an addressable storage medium or configured to execute one or more processors. In one embodiment, "unit" refers to components such as software components, object-oriented software components, class components and task components, processors, functions, properties, procedures, subroutines, programs. It can include segments of code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays and variables.

As used herein, the expression "based on" is used to describe one or more factors that affect the act or operation of a decision, judgment, or action described in a phrase or sentence in which the expression is included, and the expression refers to that It does not preclude additional factors that may affect the decision, the act of judgment, or the action.

As used in this document, the expression that a certain component (eg, a first component) is "connected" or "connected" to another component (eg, a second component) refers to a component as described above. It may mean not only being directly connected or connected to another component, but also being connected or connected through another new component (eg, a third component).

The expression "configured to" as used in this document means "configured to", "having the ability to", "modified to", "made to", "to do", depending on the context. can have meanings such as "can". The expression is not limited to the meaning of "specifically designed in hardware", for example, a processor configured to perform a specific operation is a special purpose computer structured through programming to perform that specific operation. can mean

In the present disclosure, artificial intelligence (AI) may refer to a technology that imitates human learning ability, reasoning ability, perception ability, and the like, and implements it with a computer. Artificial intelligence may include machine learning or element technology using machine learning. Machine learning may refer to an algorithm that extracts at least one feature of training data in order to classify input data. In addition, technologies that use machine learning algorithms to mimic the functions of the human brain, such as recognition and judgment, can also be understood as a category of artificial intelligence. For example, technical fields such as linguistic understanding, visual understanding, inference/prediction, knowledge expression, and motion control may be included.

Hereinafter, various embodiments of the present disclosure will be described with reference to the accompanying drawings. In the accompanying drawings and description of the drawings, identical or substantially equivalent components may be given the same reference numerals. In addition, in the following description of various embodiments, overlapping descriptions of the same or corresponding components may be omitted, but this does not mean that the corresponding components are not included in the embodiments.

1 is a diagram illustrating a process of operating a computing device according to an embodiment of the present disclosure. The computing device 100 according to the present disclosure may obtain user information 210 and/or payment information 220 for each of a plurality of users from at least one other device 110 . In the present disclosure, user information 210 may include at least one piece of information about a user's personal information. For example, the user information 210 may include information about the user's gender, the user's age, the user's home address, and the user's marital status. In one embodiment, the user information 210 may include behavioral information 215 and/or credit information 217 of the corresponding user. The behavioral information 215 may include at least one piece of information generated based on data transmitted from a terminal device of a corresponding user or an entity performing marketing for the corresponding user. For example, the behavior information 215 may include website log records, application log records, TM (Telemarketing) response information, LMS (Long Message Service) response information, and the like. Here, the TM response information may be, for example, information based on whether the user has set reception rejection for the telemarketer's marketing call or the duration of the call after receiving the marketing call. In addition, the LMS response information is, for example, information based on whether the user has set reception rejection for the marketing text delivered in the LMS format or whether the user has access to the marketing target website using the marketing text in the LMS format. can be In the present disclosure, the credit information 217 may include at least one piece of information to indicate the credit level of the corresponding user. For example, the credit information 217 may include a credit score assigned to a corresponding user by a predetermined credit bureau. The payment information 220 according to the present disclosure may include at least one piece of information generated based on a user's payment behavior. For example, the payment information 220 may include information about a payment destination, payment item, payment method, payment timing, payment number, payment frequency, payment amount, and the like. In the present disclosure, user information 210 and/or payment information 220 may mean raw data in a state not processed by the computing device 100 .

In the present disclosure, a "tag" may be information that may mean, describe, indicate, or categorize a certain characteristic related to a user. The computing device 100 of FIG. 1 may generate one or more primary tags for each of a plurality of users based on the user information 210 and/or the payment information 220 . Each of these primary tags may indicate each primary user characteristic. In the present disclosure, a “primary user feature” may be a feature that is primarily derived based on raw data included in at least one of the user information 210 and the payment information 220 . That is, the primary user characteristics may be characteristics derived from user information and/or payment information of the corresponding user through a predetermined calculation process (eg, four arithmetic operations, statistical calculations, etc.). For example, the primary user characteristics (eg, the amount paid through home shopping for the past 30 days) are the number of raw data included in the user information 210 or payment information 220 of the user, the average value of the raw data, and It may be derived based on the quartile value of data, the total value of raw data, and the like. In addition, the primary user characteristics for one user may be independently derived for that user, regardless of user information, payment information, or similarity with other users' information.

In the present disclosure, the primary tag may have a value for the corresponding primary tag. The value of the primary tag is whether the user has the corresponding primary user characteristic (eg, OO mart usage age of 60 days or more) (eg, 1 (True) / 0 (False)), or The degree of conformity (eg, 300,000 won) to the primary user characteristics (eg, the amount paid for the past 30 days through home shopping) can be indicated. The computing device 100 may generate a primary tag indicating a primary user characteristic and a value of the primary tag for each user.

The computing device 100 may generate a secondary tag 250 for each of a plurality of users based on one or more primary tags. In the present disclosure, the secondary tag 250 may indicate secondary user characteristics. Secondary user characteristics may be user characteristics arbitrarily set for marketing purposes (eg, Chinese food consumption index, etc.). In one embodiment, a sample user group for a secondary user characteristic may be predetermined. The sample user group may be a set of sample users estimated to have secondary user characteristics among a plurality of users. In one embodiment, the sample user group may be referred to as a target customer group, a correct customer group, or the like. In an embodiment, the computing device 100 may determine a degree of similarity between each of a plurality of users and a sample user group based on one or more primary tags for each sample user. Based on the similarity, the computing device 100 may generate (apply) a secondary tag 250 indicating a corresponding secondary user characteristic to each of a plurality of users.

In the present disclosure, a secondary tag may have a value for the corresponding secondary tag. The value of the secondary tag is whether the user corresponds to that secondary user characteristic (e.g., whether or not he or she is a Chinese food eater) (e.g., 1 (True) / 0 (False)), or whether the corresponding secondary tag The degree of conformity (eg, 5) to user characteristics (eg, Chinese food usage index) may be indicated. The computing device 100 may generate a secondary tag indicating a secondary user characteristic and a value of the secondary tag for each user.

In the present disclosure, “secondary user characteristics” may be distinguished from primary user characteristics directly derived from raw data in that “secondary user characteristics” are characteristics that are secondarily derived based on one or more primary tags. In the present disclosure, “secondary user characteristics” for each user may be relative characteristics of the corresponding user with respect to a sample user group. For example, secondary user characteristics cannot be derived independently with only information about each user (e.g., user information, payment information), but can be relatively derived by comparing information about a standard sample user group. there is.

In the present disclosure, the relationship between "the primary tag and the value of the primary tag" or "the secondary tag and the value of the secondary tag" is a key value for data processing and a value corresponding to each key value. (value) It can be understood as a relationship of values.

As such, the computing device 100 may generate one or more primary tags and one or more secondary tags for each of a plurality of users. In one embodiment, computing device 100 may communicate one or more primary tags and one or more secondary tags to another device 110 . Through this, the computing device 100 may provide an analysis of user characteristics of each of a plurality of users to an operator (eg, a company intending to conduct marketing) of the other device 110 .

2 is a diagram for explaining user information, payment information, a primary tag set, and a secondary tag according to an embodiment of the present disclosure. In the present disclosure, a set composed of one or more primary tags may be referred to as a primary tag set 230 . As mentioned above, a primary tag may indicate a primary user characteristic. In one embodiment, the primary user characteristic may be specified by at least one of raw data of user information 210 and payment information 220 of the user. In one embodiment, the primary user characteristic may be specified based on at least one selected from user information (eg, gender, etc.) of the user, payment destination, payment item, payment method, payment time, payment number, payment frequency, and payment amount. can In addition, since the user information 210 may include the user's behavioral information 215 and/or credit information 217, the primary user characteristics include the user's web log, app log, TM response information, LMS response information, and a credit score.

According to an embodiment of the present disclosure, the first primary tag 231 included in one or more primary tags may indicate a primary user characteristic meaning "number of payments for Chinese food in the past week". In this case, the primary user characteristic meaning "the number of payments for Chinese food in the past week" may be specified based on information about the payment destination and payment time included in the payment information 220 . The computing device 100 may calculate the value of the first primary tag 231 indicating "the number of payments for Chinese food in the past week" based on the payment destination and payment time. A value of the first primary tag 231 indicating "number of payments for Chinese food in the past week" may be an integer of 0 or more representing "number of times". Specifically, among the payment actions included in the payment information for any one user among the payment information 220 acquired by the computing device 100, the "payment destination" is a "Chinese restaurant" and the "payment point" is the last one week. A value of the first primary tag 231 indicating "the number of payments for Chinese food in the past week" may be calculated based on the payment actions corresponding to within the range. For example, the value of the first primary tag 231 may be 3 or 10.

According to another embodiment of the present disclosure, the second primary tag 232 included in one or more primary tags may indicate a primary user characteristic meaning "amount paid in home shopping for the past 30 days". . In this case, the primary user characteristic meaning "amount paid in home shopping for the past 30 days" may be specified based on information about the payment destination, payment time, and payment amount included in the payment information 220 . The computing device 100 may calculate the value of the second primary tag 232 indicating "amount paid for the past 30 days in home shopping" based on the payment destination, payment time, and payment amount. A value of the second primary tag 232 indicating "amount paid in home shopping for the past 30 days" may be a real number greater than or equal to 0 representing "amount". Specifically, the computing device 100 determines that among the payment actions included in the payment information for any one user among the acquired payment information 220, the "payment destination" is a home shopping related company and the "time of payment" is within 30 days. A value of the second primary tag indicating "amount paid for the last 30 days in home shopping" may be calculated by summing the "payment amount" of the payment activities. For example, the value of the second primary tag 232 may be 100,000 or 300,000.

The computing device 100 according to an embodiment of the present disclosure, in the same or similar manner to the above examples, includes a third primary tag 233 related to a primary user characteristic meaning "the number of gas station uses", "B" The 4th primary tag 234 related to the primary user characteristic meaning "more than 60 days of mart usage" or the 5th primary tag 234 related to the primary user characteristic meaning "restaurant-related payment amount for the past 30 days" Secondary tags 235 and the like may be created, and values corresponding to respective primary tags may be calculated. In addition, the computing device 100 according to various embodiments of the present disclosure includes a primary tag meaning "the number of accesses to site A in the past 30 days" based on the web log or app log included in the behavior information 215, and behavior information. Based on the TM response information or LMS response information included in (215), the primary tag meaning "the number of marketing contact rejections" or "whether or not the credit rating has changed" based on the credit score included in the credit information (217) It is also possible to create primary tags and the like. The foregoing description of the primary tag or the value of the primary tag is only an exemplary description for description of the present disclosure and does not limit the present disclosure.

According to an embodiment of the present disclosure, the computing device 100 generates a secondary tag 250 based on at least one primary tag included in the primary tag set 230, and the corresponding secondary tag 250 value can be determined. For example, the computing device 100 sets the value of the second tag 250 indicating the "Chinese food usage index" based on the first primary tag 231 indicating "the number of payments for Chinese food in the past week". can determine As another example, the computing device 100 includes a first primary tag 231 indicating "the number of payments for Chinese food in the past week" and a fifth primary tag indicating "amount of payment related to the restaurant business in the past 30 days". Also based on (235), the value of the secondary tag 250 indicating the "Chinese food availability index" can be determined. Here, the restaurant business indicated by the fifth primary tag 235 may be limited to a Chinese food-related business. As another example, the computing device 100 may determine the value of the secondary tag 250 indicating “Chinese food availability index” based on all of the primary tags of the primary tag set.

3 is a block diagram of a computing device 100 according to one embodiment of the present disclosure. According to one embodiment of the present disclosure, the computing device 100 may include one or more processors 310 and/or one or more memories 320 as components. In some embodiments, at least one of these components of computing device 100 may be omitted or other components may be added to computing device 100 . In some embodiments, additionally or alternatively, some components may be integrated and implemented, or implemented as a single or multiple entities. In this disclosure, one or more processors 310 may be referred to as processor 310 . The expression processor 310 may mean a set of one or more processors, unless the context clearly dictates otherwise. In this disclosure, one or more memories 320 may be referred to as memory 320 . The expression memory 320 may mean a set of one or more memories, unless the context clearly dictates otherwise.

At least some of the internal and external components of the computing device 100 are connected to each other through a bus, general purpose input/output (GPIO), serial peripheral interface (SPI), or mobile industry processor interface (MIPI). , data and/or signals can be sent and received.

The processor 310 according to various embodiments may control at least one component of the computing device 100 connected to the processor 310 by driving software (eg, instructions, programs, etc.). In addition, the processor 310 may perform operations such as various calculations, processing, data generation, and processing related to the present disclosure. Also, the processor 310 may load data or the like from the memory 320 or store data in the memory 320 .

The memory 320 according to various embodiments may store various data. Data stored in the memory 320 is data acquired, processed, or used by at least one component of the computing device 100 and may include software (eg, commands, programs, etc.). Memory 320 may include volatile and/or non-volatile memory. In the present disclosure, a command or a program is software stored in the memory 320, and various functions such as an operating system for controlling resources of the computing device 100, an application, and/or an application to utilize the resources of the computing device 100. may include middleware that provides the application. In one embodiment, memory 320 may store instructions that, when executed by processor 310, cause processor 310 to perform an operation. In one embodiment, the memory 320 may store user information for each of a plurality of users, payment information for each of a plurality of users, a primary tag for a plurality of users, a secondary tag for a plurality of users, and the like. .

In one embodiment, computing device 100 may further include communication circuitry 330 . The communication circuitry 330 may perform wireless or wired communication between the computing device 100 and a server or between the computing device 100 and other devices. For example, the communication circuit 330 may include enhanced mobile broadband (eMBB), ultra reliable low-latency communications (URLLC), massive machine type communications (MMTC), long-term evolution (LTE), and LTE Advance (LTE-A). , New Radio (NR), Universal Mobile Telecommunications System (UMTS), Global System for Mobile communications (GSM), Code Division Multiple Access (CDMA), Wideband CDMA (WCDMA), Wireless Broadband (WiBro), Wireless Fidelity (WiFi), Wireless communication may be performed according to a method such as Bluetooth, Near Field Communication (NFC), Global Positioning System (GPS), or Global Navigation Satellite System (GNSS). For example, the communication circuit 330 performs wired communication according to a method such as Universal Serial Bus (USB), High Definition Multimedia Interface (HDMI), Recommended Standard-232 (RS-232), or Plain Old Telephone Service (POTS). can be done In one embodiment, communication circuitry 330 may communicate with another device 110 .

In one embodiment, the computing device 100 may further include an input unit 340 . The input unit 340 may be a component for transferring data received from the outside to at least one component included in the computing device 100 . For example, the input unit 340 may include a mouse, keyboard, or touch pad.

The computing device 100 according to various embodiments of the present disclosure may be various types of devices. For example, a computing device may be a portable communication device, a portable multimedia device, a wearable device, a consumer electronics device, or a device according to a combination of one or more of the foregoing. The computing device of the present disclosure is not limited to the devices described above.

In one embodiment, the computing device 100 may communicate with another external device 110 through a server or directly. The other device 110 may be the same or a different type of device than the computing device 100 . The computing device 100 may perform subsequent operations using information received from the other device 110 . The computing device 100 receives user information 210 and/or payment information 220 for each of a plurality of users from the other device 110, and sets a primary tag set 230 and/or 2 associated with each user. Subsequent operations to create tea tag 250 may be performed. In order to perform such a process, a method such as cloud computing or distributed computing may be used.

4 is a diagram for explaining a relationship between a plurality of users and a sample user group according to an embodiment of the present disclosure. As described above, the secondary user characteristics indicated by the secondary tag 250 are user characteristics arbitrarily set for marketing purposes, and a sample user group for the secondary user characteristics may be predetermined. Hereinafter, a method of pre-determining a sample user group from among a plurality of users will be described in detail through some embodiments.

According to an embodiment of the present disclosure, the computing device 100 may pre-determine the sample user group 430 by receiving a signal for selecting one or more users from among the plurality of users 410 from the outside. The computing device 100 may receive an input signal for selecting at least one user among the plurality of users 410 as a sample user through the input unit 340 and determine the sample user group 430 based on the input signal. The computing device 100 receives a signal for selecting one or more users among a plurality of users 410 as a sample user from an external server or other device 110 through the communication circuit 330, and based on the signal, a sample user group ( 430) may be determined.

According to one embodiment of the present disclosure, the computing device 100 may pre-determine the sample user group 430 based on one or more arbitrary conditions. Here, one condition may include a primary tag and a threshold value for the primary tag value.

For example, the computing device 100 may determine the sample user group 430 based on one condition, "the number of times of using a Chinese restaurant in the last month is 10 or more." That is, the corresponding condition may include a first tag indicating "the number of times of Chinese restaurant use in the last month" and a threshold value of 10. The computing device 100 may determine, among the plurality of users 410 , users having a value of 10 or more in a primary tag corresponding to “the number of times of using a Chinese restaurant in the last month” as sample users.

As another example, the computing device 100 selects a sample user group 430 based on the condition that "the number of times of using a Chinese restaurant in the past month is 10 or more times" and the condition that "the total amount of payment related to the restaurant business in the past month is 3 million won or more" can determine That is, the first tag indicating “the number of times of Chinese restaurant use in the last month” and its threshold value is 10, and the first tag indicating “the total number of restaurant-related payments in the last month” and its threshold value is 3,000,000. All of these can be used. In the computing device 100, among the plurality of users 410, the value of the first tag corresponding to "the number of times of using a Chinese restaurant in the last month is 10 or more" is 10 or more, and corresponds to "the total amount of payments related to the restaurant business in the last month" Users whose primary tag value of 3,000,000 or more may be determined as sample users.

The illustrative descriptions of any conditions for determining the sample user group 430 described above are not limiting of this disclosure, and computing device 100 according to various embodiments may use any conditions for each of the one or more primary tags. Based on this, a sample user group 430 may be determined from among the plurality of users 410 .

As described above, the computing device 100 determines each of the plurality of users 410 and the sample based on one or more primary tags for each of the sample users in a predetermined sample user group 430 among the plurality of users 410 . A degree of similarity between user groups 430 may be determined. Hereinafter, a method for determining the degree of similarity between each of the plurality of users 410 and the sample user group 430 will be described in detail through some embodiments.

The computing device 100 according to an embodiment of the present disclosure determines the difference between the value of the primary tag for each of the plurality of users 410 and the value of the primary tag for each sample user in the sample user group 430. Based on this, a similarity between each of the plurality of users 410 and the sample user group 430 may be determined. That is, the computing device 100 calculates the difference between the values of the primary tags in the relationship between one user (hereinafter, first user 411 hereinafter) and each sample user among the plurality of users 410 and calculates the difference value according to the calculation result. Based on this, a degree of similarity between the first user 411 and the sample user group 430 may be determined. The similarity determination method is described below based on the first user 411 , but the similarity determination method for each of the plurality of users may be determined in a manner similar to the similarity determination method for the first user 411 .

For example, the computing device 100 calculates an average value of a difference between a predetermined primary tag value of the first user 411 and a corresponding primary tag value of each sample user (hereinafter referred to as a first average error). ) can be calculated. The computing device 100 may determine a similarity of the first user 411 to the sample user group 430 based on the first average error. Specifically, the computing device 100 may determine that the first user 411 has a high degree of similarity to the sample user group 430 as the first average error decreases, and may assign a high degree of similarity value.

, 제2 평균 오차에 대해 가중치

를 각각 곱한 후 합산하여 평균 오차를 가중 합산할 수 있다. 컴퓨팅 장치(100)는 복수의 평균 오차들에 대한 가중 합산 값이 작을수록 제1 사용자(411)에 대해 높은 유사도 값을 부여할 수 있다.In a similar way, the degree of similarity may be determined based on a difference between values of two or more primary tags. The computing device 100 may calculate an average value of a difference between a predetermined primary tag value of the first user 411 and a corresponding primary tag value of each sample user (ie, a first average error). there is. In addition, the computing device 100 may calculate an average value of differences between values of other primary tags of the first user 411 and values of corresponding other primary tags of each sample user (hereinafter referred to as a second average error). can The computing device 100 may determine a degree of similarity between the first user 411 and the sample user group 430 based on the first average error and the second average error. For example, the computing device 100 may determine the degree of similarity between the first user 411 and the sample user group 430 by summing a plurality of mean errors. The computing device 100 may assign a higher similarity value to the first user 411 as the sum of the plurality of average errors is smaller. As another example, the computing device 100 determines the degree of similarity of the first user 411 to the sample user group 430 by applying a weight to each of a plurality of mean errors and performing a weighted sum operation. can Specifically, the computing device 100 weights the first mean error.

, weighted for the second mean error

The weighted sum of average errors can be obtained by multiplying and summing each. The computing device 100 may assign a higher similarity value to the first user 411 as the weighted sum of the plurality of average errors is smaller.

In determining the degree of similarity of the first user 411 to the sample user group 430 in the present disclosure, the computing device 100 may utilize various statistical analysis methods other than the method based on the above-described “average error”. For example, the computing device 100 performs operations such as average value calculation, standard deviation calculation, variance value calculation, total sum calculation, and Nth percentile calculation (N is an integer of 0 or more and 100 or less) by performing operations such as each of a plurality of users. The degree of similarity can be determined. The above examples of various computational methods for statistical analysis are for explanation only and do not limit the present disclosure.

Based on the difference between the value of the primary tag for the first user 411 and the reference value of the primary tag of the sample user group 430, the computing device 100 according to an embodiment of the present disclosure determines the first user ( 411) and the sample user group 430 may be determined. In the present disclosure, a “primary tag reference value” for a predetermined primary tag may mean a value determined based on a primary tag value of each sample user for the corresponding primary tag. The computing device 100 may determine the primary tag reference value based on the value of the primary tag for each sample user according to a predetermined method in order to set a value representing the characteristics of the sample user group 430 . For example, the computing device 100 may determine an average value of primary tag values for each sample user as the primary tag reference value of the sample user group 430 . As another example, the computing device 100 may determine a median value of primary tag values for each sample user as the primary tag reference value of the sample user group 430 . As another example, the computing device 100 may determine a mode value of the primary tag value for each sample user as the primary tag reference value of the sample user group 430 . Examples of the primary tag reference value described above do not limit the present disclosure.

Computing device 100 according to an embodiment of the present disclosure compares the value of the primary tag of the first user 411 for a predetermined primary tag with the reference value of the primary tag of the sample user group 430, As the difference value is smaller, a higher similarity value may be assigned to the first user 411 . For example, a first user 411 and another user among a plurality of users (hereinafter referred to as a second user 412) are users who are subject to similarity determination, and each user's primary tag for a specific primary tag. It is assumed that the value and the primary tag reference value are "first user 411: 6, second user 412: 1, primary tag reference value: 9". At this time, the computing device 100 calculates the difference between the value of each user's primary tag and the primary tag reference value, and as the difference between the primary tag reference value and the primary tag reference value according to the calculation result is smaller, the first user 411 and A high degree of similarity may be assigned to each of the two users 412 . In the above example, the difference from the primary tag reference value may be calculated as “3” for the first user 411 and “8” for the second user 412 . In this case, the computing device 100 may assign a higher degree of similarity to the first user 411 having a smaller difference than the second user 412 .

5 is a diagram for explaining a similarity determination process based on a neural network-based learning model according to an embodiment of the present disclosure.

In one embodiment of the present disclosure, the computing device 100 is based on a neural-network-based trained model (550), a plurality of users (eg, the first user 411, the second user ( 412), etc.) and the similarity between each sample user group 430 may be determined. Here, the learning model 550 may be learned based on the primary tag set 531 and the like for each sample user in the sample user group 430 . Each primary tag set may include one or more primary tags and their values. In one embodiment, one or more memories 320 may store neural network-based learning models.

In the present disclosure, a neural network-based learning model may be designed to implement a human brain structure on a computer, and may include a plurality of network nodes that simulate neurons of a human neural network and have weights. A plurality of network nodes may have a connection relationship between them by simulating synaptic activities of neurons that transmit and receive signals through synapses. In a neural network-based learning model, a plurality of network nodes may exchange data according to a convolutional connection relationship while being located in layers of different depths. The neural network-based learning model may be, for example, an artificial neural network model, a convolutional neural network model, and the like. In the present disclosure, a neural network-based learning model is a model learned according to a predetermined machine learning method, and may mean a model in which a weight for at least one network node included in an untrained model is determined by machine learning. Machine learning may mean that computer software improves data processing capabilities through learning using data and experiences of processing the data. A neural network-based learning model is constructed by modeling correlations between data, and the correlations may be expressed by a plurality of parameters. A neural network-based learning model derives a correlation between data by extracting and analyzing features from given data, and optimizing the parameters of a neural network-based learning model by repeating this process can be referred to as machine learning. For example, a neural network-based learning model may learn a mapping (correlation) between an input and an output with respect to data given as an input/output pair. Alternatively, the neural network-based learning model may learn the relationship by deriving a regularity between given data even when only input data is given.

Hereinafter, a process of determining the degree of similarity between each of a plurality of users and the sample user group 430 using the aforementioned neural network-based learning model 550 will be described. Although this process may be performed by various embodiments, the first and second embodiments will be described below. However, the two embodiments do not limit the scope of the present disclosure.

The first embodiment will be described below. In the first embodiment, the learning model 550 may receive primary tag sets for each of a plurality of sample users and learn common features of the plurality of primary tag sets. In one example, the learning model 550 receives raw data (ie, user information 210 and/or payment information 220) for each of a plurality of sample users, and obtains common data among the plurality of raw data. Features can also be learned. In another example, the learning model 550 receives a primary tag set and raw data for each of a plurality of sample users, and learns a common feature existing among the plurality of “primary tag sets and raw data”. may be In the following description of the first embodiment, for convenience of description, it is assumed that the learning model 550 receives a primary tag set for each of a plurality of sample users. The learning model 550 generates intermediate data by reducing the dimension of input data (ie, the plurality of primary tag sets) in order to learn common features of a plurality of primary tag sets, and converts the dimension of the generated intermediate data to Again, an operation of restoring output data having the same dimensions as the input data may be performed. Specifically, the plurality of primary tag sets input for learning include the primary tag set 531 of the first sample user, the primary tag set of the second sample user, ..., the primary tag of the nth sample user. may contain sets. The primary tag set of each sample user may include primary tags of the same type. When the primary tag set includes K primary tags (K is a natural number), the primary tag set may be converted into a vector having K elements and input to the learning model 550 . Here, K elements of the corresponding vector may correspond to values of K primary tags. The learning model 550 may convert input data having a vector size K into intermediate data having a vector size L (L is a natural number smaller than K) through an operation based on at least one network node. As the dimension of the input data is reduced, the learning model 550 may compress and preserve main information of the input primary tag set. The learning model 550 may restore output data having a vector size of K from intermediate data having a vector size of L through an operation based on at least one network node. The learning model 550 may calculate an error between input data and output data. As the error between the input data and the output data is smaller, it can be evaluated that the learning model 550 has learned well the common features of the plurality of learning data (ie, the plurality of primary tag sets). Since the learning model 550 is learned through such a data compression and restoration process, only the training data corresponding to the correct answer (primary tag sets of sample users pre-determined to have predetermined secondary user characteristics), secondary user characteristics It is possible to learn the tendency of primary user characteristics of users having . The learning model 550 may learn to well preserve the characteristics of input data of primary tag sets of predetermined sample users through the above-described data compression and restoration processes in output data. That is, the learning model 550 is a process of generating intermediate data by compressing input data or restoring intermediate data to obtain output data based on an operation of back-propagating an error between input data and output data. A weight for at least one network node involved in the creation process may be updated. As a result, the learning model 550 may determine that "the error is small (ie, the degree of similarity is high)" as the newly input input data is similar to the data for the primary tag set of predetermined sample users.

The computing device 100 may input one or more primary tags for each of a plurality of users (ie, a set of primary tags for each user) to the learning model 550 learned based on the first embodiment. According to the input, the computing device 100 may obtain a degree of similarity between each of the plurality of users and the sample user group 430 . Specifically, the computing device 100 may input input data (ie, a primary tag set of the first user 411 ) into the learning model 550 . The learning model 550 may generate output data having the same vector dimension size as input data through an operation based on at least one network node. Since the learning model 550 is learned based on the primary tag set of each sample user of the sample user group 430, the input primary tag set of the first user 411 is 1 of the sample user group 430. As the car tag sets are similar, output data having a smaller difference from input data can be generated. The learning model 550 may determine a degree of similarity between the first user 411 and the sample user group 430 based on the error between the input data and the output data. That is, the learning model 550 may assign a higher degree of similarity to the first user 411 as the difference between the input data and the output data is smaller. The computing device 100 may obtain a degree of similarity with respect to the first user 411 from the learning model 550 . In the same way, the computing device 100 may obtain a degree of similarity for each of a plurality of users from the learning model 550 .

이므로 사전 결정된 임계 거리 1을 초과하는 바, 학습 모델(550)은 학습을 위한 오차 역전파 과정에서, 거리의 차이 값(즉,

을 오차로 하여, 해당 매핑 데이터에 대한 적어도 하나의 네트워크 노드에 대한 가중치를 학습할 수 있다. 전술한 학습 모델(550)의 학습 방법에 관한 일부 구체적인 기재들은 발명의 설명을 위한 예시적 기재일 뿐, 본 개시를 한정하지 않는다.A second embodiment will be described below. In the second embodiment, the learning model 550 receives primary tag sets for each of a plurality of sample users, and learns common characteristics of the plurality of primary tag sets based on a clustering technique. can In one example, the learning model 550 receives raw data (ie, user information 210 and/or payment information 220) for each of a plurality of sample users, and based on a clustering technique, the plurality of raw data Common characteristics that exist among them can be learned. In another example, the learning model 550 receives a primary tag set and raw data for each of a plurality of sample users and, based on a clustering technique, exists among the plurality of “primary tag sets and raw data”. Common characteristics can be learned. In the following description of the second embodiment, for convenience of explanation, it is assumed that the learning model 550 receives a primary tag set for each of a plurality of sample users. For example, the learning model 550 may be trained based on a K-means clustering technique. The learning model 550 may map input data (ie, primary tag sets of a plurality of sample users) to an N-dimensional latent space, respectively, in order to cluster a plurality of primary tag sets ( N is a natural number). In the present disclosure, the latent space may be a coordinate system having an arbitrary dimension in which input data input to the learning model 550 can be expressed. When the coordinate space of the original input data is referred to as an “observation space,” the learning model 550 performs a one-to-one correspondence between a point on the observation space and a point on the latent space through a predetermined operation for converting the coordinate system ( that is, mapping). As described above, the primary tag sets of each sample user may include primary tags of the same type. The computing device 100 may generate input data representing the corresponding primary tag set in order to input the primary tag set of each sample user to the learning model 550 . For example, if the primary tag set includes K primary tags, the primary tag set is converted into input data (ie, K-dimensional coordinates) having values of each K primary tag as coordinate components. can The learning model 550 may convert K-dimensional coordinates, which are input data, into N-dimensional coordinates corresponding to one coordinate in an N-dimensional latent space through an operation based on a weight for at least one network node (where N is K natural number less than or equal to). That is, the learning model 550 may map one primary tag set to one coordinate in the N-dimensional latent space through such transformation. The learning model 550 may be trained so that each data mapped to the N-dimensional latent space is not farther than a predetermined threshold distance from a predetermined reference point. Here, the reference point is a point existing in the N-dimensional latent space, and is mapping data for a primary tag set of a sample user included in the sample user group 430 or mapping data for a virtual primary tag set. can be For example, the reference point may be mapping data for a primary tag set representing a predetermined criterion used to determine the sample user group 430 . The distance between data mapped in the N-dimensional latent space and a reference point may be calculated as, for example, a Euclidean distance. Specifically, the dimension of the latent space is 3, the coordinates of the reference point in the latent space are (0,0,0), the predetermined critical distance is 1, and the primary tag set of one user by the operation of the learning model 550 The coordinates of the mapping data may be (2,2,2). In this case, the distance between the reference point and the mapping data is

Since it exceeds the predetermined threshold distance 1, the learning model 550, in the error back-propagation process for learning, the difference value of the distance (i.e.,

A weight value for at least one network node for the corresponding mapping data may be learned by taking as an error. Some specific descriptions of the learning method of the learning model 550 described above are only exemplary descriptions for explanation of the invention, and do not limit the present disclosure.

The computing device 100 may input one or more primary tags for each of a plurality of users (ie, a set of primary tags for each user) to the learning model 550 learned based on the second embodiment described above. . According to the input, the computing device 100 may obtain a degree of similarity between each of the plurality of users and the sample user group 430 . Specifically, the computing device 100 may input input data (ie, a primary tag set for the first user 411 ) into the learning model 550 . The learning model 550 may also map corresponding input data to an N-dimensional latent space through an operation based on at least one network node. The learning model 550 learned by mapping the primary tag sets of a plurality of sample users to the N-dimensional latent space may calculate a center point on the N-dimensional latent space of the mapped primary tag sets of the plurality of sample users ( This center point is also expressed in N-dimensional coordinates). As the input primary tag set of the first user 411 is similar to the primary tag sets of the sample user group 430, the primary tag set of the first user 411 becomes a plurality of sample users in the N-dimensional space. It can be mapped to a distance close to the center point of the primary tag sets of . The learning model 550 determines the degree of similarity between the first user 411 and the sample user group 430 based on the distance between the coordinates to which the primary tag set of the first user 411 is mapped and the aforementioned center point. can That is, the learning model 550 has a higher similarity to the first user 411 as the distance between the center point of the primary tag sets of the plurality of sample users in the N-dimensional space and the primary tag set of the first user 411 is closer. can be granted. The computing device 100 may obtain a degree of similarity with respect to the first user 411 from the learning model 550 . In the same way, the computing device 100 may obtain a degree of similarity for each of a plurality of users from the learning model 550 .

The neural network-based learning model 550 of the present disclosure may be learned by the computing device 100 and stored in the memory 320 of the computing device 100 . In addition, the neural network-based learning model 550 of the present disclosure may be learned by an external server or other device 110 and then transmitted to the computing device 100 through the communication circuit 330 and stored in the memory 320. .

6 is a diagram illustrating a process of determining a secondary tag based on similarity according to an embodiment of the present disclosure. As mentioned above, secondary tags may indicate secondary user characteristics associated with a sample group of users. The computing device 100 according to an embodiment of the present disclosure may generate a secondary tag for each of the plurality of users based on the degree of similarity determined between each of the plurality of users and the sample user group. In the present disclosure, a secondary tag generated by the computing device 100 may have a corresponding tag value. That is, the computing device 100 may generate a secondary tag for each of a plurality of users and determine a value of a secondary tag corresponding to the corresponding secondary tag.

According to an embodiment of the present disclosure, the computing device 100 may determine an arbitrary number of users from among a plurality of users in an order of high similarity. At this time, the arbitrary number is a number temporarily set to determine the value of the secondary tag, and may be changed in the process of determining the value of the secondary tag. In one example, an arbitrary number may be set considering the number of potential customers to be marketed to. For example, a random number is set to a smaller number (eg, 1,000) of the number of potential customers (eg, 20,000), then a larger number (eg, 2,000, 3,000 . ..) can be increased. Also, for example, an arbitrary number may be set as a multiple of the number of potential customers. The computing device 100 may determine a ratio of a sample user among a random number of users determined in order of similarity, and determine a value of a secondary tag for each of a plurality of users based on the determined ratio.

The label set 600 shown in FIG. 6 may include a plurality of labels. Each label may have a one-to-one correspondence with each of a plurality of users. Each label in the label set 600 may be listed in order of high similarity based on a similarity between a user corresponding to the corresponding label and a sample user group. That is, the plurality of labels included in the label set 600 may be arranged so that the similarity between the user corresponding to the label and the sample user group decreases as the user number corresponding to each label increases. Since the sample user group is determined within a plurality of users, some labels included in the label set 600 may correspond to the sample user. In FIG. 6, labels corresponding to sample users (eg, user #1, user #2, user #1000, user #1600, etc.) are shaded, and labels corresponding to users other than sample users (eg, user #1000, user #1600, etc.) : User #4, User #5, User #1595, User #1602, etc.) are displayed without shading.

According to an embodiment of the present disclosure, the computing device 100 may determine a secondary tag value representing “whether or not a secondary user has a characteristic” for each of a plurality of users. The secondary tag value indicating whether or not the secondary user characteristic is present may be, for example, a tag value for binary classification. If not, it may be determined as "0". The computing device 100 may determine an arbitrary number of users in the order of high similarity among the plurality of users. For example, if the number of arbitrary users is “1000”, the computing device 100 may determine 1000 users in order of similarity. Reference number 611 in FIG. 6 indicates a label corresponding to user #1000. Computing device 100 may determine a percentage of a sample user out of an arbitrary number of users. Assuming that 900 sample users are among 1000 users, the computing device 100 may determine the sample user ratio as 90%.

The computing device 100 may determine the value of the secondary tag for each of the plurality of users based on the sample user ratio determined as described above. The computing device 100 may determine the value of the secondary tag for each of the plurality of users by comparing the determined sample user ratio with a predetermined threshold ratio. For description of various embodiments according to the present disclosure, it is assumed that an arbitrary number of users is N, a ratio of a sample user among N users is x%, and a predetermined threshold ratio is y%. At this time, if x is greater than or equal to y, the computing device 100 assigns a secondary tag value indicating that secondary user characteristics exist to the top N users based on similarity among the plurality of users, and A secondary tag value representing non-existence of secondary user characteristics may be assigned to the remaining users other than the top N users among users based on similarity. The computing device 100 may determine a secondary tag value for each of a plurality of users additionally based on an operation of sequentially increasing the number N of arbitrary users. The computing device 100 determines N users in order of similarity among a plurality of users based on the increased number of random users (N), and calculates a ratio (x%) of sample users among the determined N users. can decide The computing device 100 compares the sample user ratio (x%) with a predetermined threshold ratio (y%), and compares the sample user ratio (x%) to a random user whose ratio (x%) is less than the predetermined threshold ratio (y%) for the first time. number (N) can be determined. According to the determined number of users (N), the computing device 100 assigns a secondary tag value indicating that a secondary user characteristic exists to the top N users based on similarity among the plurality of users, and assigns a secondary tag value to the remaining users. A secondary tag value representing non-existence of user characteristics may be assigned. For example, if the threshold rate is 50%, 300 of the top 600 users based on similarity are sample users, and the remaining 300 are non-sample users, so the percentage of sample users out of the 600 users is 50%, then the 600 users A secondary tag value of “1” may be assigned to the user, and a secondary tag value of “0” may be assigned to the remaining users of 600 or less based on similarity.

According to an embodiment of the present disclosure, the computing device 100 may determine a secondary tag value indicating "a degree of conformity to secondary user characteristics" for each of a plurality of users. The value of the secondary tag may be a value for classifying users into one or more categories according to the degree of matching with secondary user characteristics. For example, secondary tag values may be differentially determined between a predetermined maximum value (eg, 10) and a predetermined minimum value (eg, 1) according to a degree of conformity to secondary user characteristics. The process of determining, by the computing device 100 according to the present disclosure, a secondary tag value representing “the degree to which a secondary user characteristic is matched” for each of a plurality of users is “whether or not the user has the secondary user characteristic” described above. It can be understood as the same as or similar to performing the process of determining the secondary tag value indicating 2 or more times.

Specifically, the computing device 100 may determine an arbitrary number of users in order of high similarity among the plurality of users. The computing device 100 may sequentially increase the number of users in order to determine a secondary tag value indicating "a degree of conformity to secondary user characteristics". The computing device 100 may determine a ratio of a sample user among random users and compare the determined sample user ratio with a plurality of predetermined threshold ratios to determine a secondary tag value for each of a plurality of users.

In order to describe the operation of the computing device 100 according to an embodiment of the present disclosure, it is assumed that the plurality of predetermined critical ratios include "first critical ratio - 90%" and "second critical ratio - 80%". . When the sample user ratio included in the random user M1 users is 90% or more, the computing device 100 sets the secondary tag value for the top M1 people based on similarity to “10” because the sample user ratio is equal to or greater than the first threshold ratio. can be determined by After that, the computing device 100 may determine the size of the number of arbitrary users as a value greater than M1 in order to determine the secondary tag value. The computing device 100 calculates similarity when the sample user ratio included in M2 users (M2 is a natural number greater than M1) is less than 90% and greater than or equal to 80%, since the sample user ratio is less than the first threshold ratio and greater than or equal to the second threshold ratio. A secondary tag value for the remaining users excluding the top M1 among the top M2 people may be determined to be "9".

For example, in FIG. 6, it is assumed that the number of sample users included in users from user #1 to user #1000 is 900 and the number of sample users included in users from user #1 to user #1600 is 1280. do. Additionally, it is assumed that the plurality of predetermined critical ratios include a first critical ratio and a second critical ratio, and the first critical ratio and the second critical ratio are determined to be 90% and 80%, respectively. The computing device 100 may determine an arbitrary number of users as “1000”. Computing device 100 may determine a percentage of sample users out of 1000 users. The computing device 100 selects labels corresponding to sample users (eg, users #1, #2, #6, #997, and #1000) among labels having the same or higher rank than the label 611 corresponding to user #1000. etc.) can be counted to determine the ratio of sample users among 1000 users. According to the assumption, since the number of sample users among 1000 users is 900, the proportion of sample users among 1000 users may be 90%. Since the ratio (90%) of sample users among 1000 users is equal to or greater than the first threshold ratio (90%), the computing device 100 determines the top 1000 users (user #1 to user #) based on similarity among a plurality of users. up to 1000) can be determined as "10".

Continuing, the computing device 100 may determine the number of arbitrary users as a value greater than “1000”. The computing device 100 may determine an arbitrary number of users as “1600”. Computing device 100 may determine a percentage of the sample users out of the 1600 users. The computing device 100 selects labels corresponding to sample users (eg, users #1, #2, #6, #997, and #1000) among labels having the same or higher rank than the label 613 corresponding to user #1600. , #1594, #1599, #1600, etc.) may be counted to determine a sample user ratio among 1600 users. According to the above assumption, the number of sample users among 1600 users is 1280, so the ratio of sample users among 1600 users may be 80%. The computing device 100 determines that the secondary tag value among the plurality of users is determined because the ratio (80%) of the sample users among the 1600 users is less than the first threshold ratio (90%) and is greater than or equal to the second threshold ratio (80%). Secondary tag values for the remaining users (users #1001 to user #1600) excluding the top 1000 users (users #1 to user #1000) based on the determined similarity may be determined to be “9”. The computing device 100 may divide the plurality of users into two or more groups by repeating the above method and assign a corresponding secondary tag value to each group.

In order to describe the operation of the computing device 100 according to an embodiment of the present disclosure, a plurality of predetermined threshold ratios are "first threshold ratio - 90%", "second threshold ratio - 80%", "third threshold ratio" ratio - 70%", "fourth threshold ratio - 60%", and "fifth threshold ratio - 50%". When the number of random users is expressed as "N_1", "N_2", "N_3", "N_4", and "N_5", respectively, the number of random users is equivalent to "N_1 < N_2 < N_3 < N_4 < N_5" can have a relationship. The computing device 100 may monotonically increase a certain number of users from 1 to determine a sample user ratio for each number of users. When the ratio of sample users included in the top N_1 users based on similarity is 90% or more, the computing device 100 may determine a secondary tag value of “10” for the top N_1 users based on similarity among the plurality of users. When the ratio of sample users included in the top N_2 users based on similarity is 80% or more, the computing device 100 selects the top N_2 users based on similarity among the plurality of users, and among the users other than the top N_1 users based on similarity, the computing device 100 The secondary tag value can be determined as "9". When the proportion of sample users included in the top N_3 users based on similarity is 70% or more, the computing device 100 selects the top N_3 users based on similarity among a plurality of users, and among the users except for the top N_2 users based on similarity, the computing device 100 The secondary tag value can be determined as "8". When the proportion of sample users included in the top N_4 users based on similarity is 60% or more, the computing device 100 selects the top N_4 users based on similarity among the plurality of users, and among the users except for the top N_3 users based on similarity, the computing device 100 The secondary tag value can be determined as "7". When the ratio of sample users included in the top N_5 users based on similarity is 50% or more, the computing device 100 selects the top N_5 users based on similarity among the plurality of users, and among the users other than the top N_4 users based on similarity, the computing device 100 The secondary tag value can be determined as "6".

The computing device 100 may also determine secondary tag values for the remaining users excluding the top N_5 users based on similarity among the plurality of users. At this time, for the remaining users excluding the top N_5 users based on similarity, unlike the above method, a method of dividing the number of remaining users into two or more equal sections and determining different secondary tag values according to each section is taken. can For example, if there are 100 users excluding the top N_5 users based on similarity, the computing device 100 assigns "5" as a secondary tag value to the top 100 users among the remaining 100 users, and assigns "5" to the next top 100 users. For , the secondary tag value can be determined in a manner such as giving "4" as the secondary tag value.

According to various embodiments of the present disclosure, the computing device 100 may generate a secondary tag for each of the plurality of users from raw data of the plurality of users through quantified steps. Raw data for a plurality of users may have low density of information in terms of individual data, so it may be difficult to convey meaningful information by itself. The present disclosure discloses a method of generating a primary tag related to “primary user characteristics” having a certain level of meaning through numerical analysis of such raw data. In addition, the present disclosure specifically discloses a method of generating a secondary tag representing a “second user characteristic” based on a plurality of primary tags indicating a plurality of “primary user characteristics,” thereby providing at least two tags from a vast amount of data. A method for effectively extracting key information through the above calculation steps is disclosed.

According to various embodiments of the present disclosure, the computing device 100 generates secondary tags for each user based on a degree of similarity between the primary tag set of each user and the primary tag set of sample users included in the sample user group. By disclosing a method of assigning a tag, a method of effectively searching for a potential customer group having the most similar pattern to a target customer group is disclosed. In addition, when the computing device 100 determines a secondary tag value indicating the degree to which the plurality of users match the secondary user characteristics of the sample user group, the computing device 100 divides the plurality of users into two or more groups. By determining secondary tag values (eg, 10, 7, 3, etc.) of different sizes for each group, marketing priority among a plurality of users can be determined according to the secondary tag value. As described above, according to the present disclosure, by disclosing a method for determining a marketing target priority among a plurality of users, the efficiency of marketing can be increased, while waste of resources required for marketing can be reduced.

According to various embodiments of the present disclosure, one or more primary tags are generated from raw data, and a similarity calculation is performed between users using the one or more primary tags to generate a secondary tag, thereby calculating similarity using the raw data. Comparatively, the calculation load on the computing device 100 may be reduced.

7 is a flowchart illustrating an operation of a computing device according to an embodiment of the present disclosure. The computing device 100 according to an embodiment of the present disclosure may obtain user information for each of a plurality of users and payment information for each of a plurality of users in operation S710. User information may include information about gender, age, home address, marital status, and the like. The payment information may include at least one piece of information about a payment destination, payment time, number of payments, frequency of payment, amount of payment, and the like.

The computing device 100 according to an embodiment of the present disclosure may generate one or more primary tags for each of a plurality of users based on user information and/or payment information in operation S720. Each of the one or more primary tags may indicate a primary user characteristic about the user. As described above, the primary user characteristics may be characteristics that are primarily derived based on raw data included in the user information 210 and/or the payment information 220 . That is, the primary user characteristics may be characteristics derived from user information and/or payment information of the corresponding user through a predetermined calculation process (eg, four arithmetic operations, statistical calculations, etc.).

The computing device 100 according to an embodiment of the present disclosure may determine a degree of similarity between each of a plurality of users and the sample user group based on one or more primary tags for each sample user in the sample user group, in operation S730. there is. As described above, the sample user group may be a set of sample users estimated to have secondary user characteristics among a plurality of users. The degree of similarity determined between each user and the sample user group may be an index indicating how similar each user is to the sample user group.

In operation S740, the computing device 100 according to an embodiment of the present disclosure may generate a secondary tag based on a degree of similarity between each of the plurality of users and the determined sample user group. A secondary tag for a user may indicate secondary user characteristics about the user. As described above, the secondary user characteristics may be user characteristics arbitrarily set for marketing purposes (eg, Chinese food usage index, etc.). Secondary user characteristics may be distinguished from primary user characteristics directly derived from raw data in that they are characteristics that are secondarily derived based on one or more primary tags.

As described above, the user information 210 for each of the plurality of users may include behavior information 215 for each of the plurality of users and/or credit information 217 for each of the plurality of users. Accordingly, computing device 100 may generate one or more primary tags for each of a plurality of users based further on behavioral information 215 and/or credit information 217 in user information 210 .

8 is a flowchart illustrating an operation of a computing device according to an embodiment of the present disclosure. Operations S810 and S820 of FIG. 8 may be performed identically or similarly to operations S710 and S720 of FIG. 7 , respectively, and descriptions of overlapping contents are omitted below.

The computing device 100 according to an embodiment of the present disclosure, in operation S830, performs a neural network-based learning model learned based on one or more primary tags (ie, primary tag groups) for each sample user in the sample user group. In, one or more primary tags for each of a plurality of users may be input. The neural network-based learning model may be a model previously learned by the computing device 100 or an external device. Since the neural network-based learning model is a model learned based on one or more primary tags for each sample user included in a sample user group, newly input data (ie, one or more primary tags of each of a plurality of users (primary tags) It may be determined whether the tag group)) is similar to the data of the sample user group.

The computing device 100 according to an embodiment of the present disclosure may determine a degree of similarity between each of a plurality of users and a sample user group based on a neural network-based learning model in operation S840. The computing device 100 may determine the similarity of each user by directly using the similarity value output from the neural network-based learning model. The computing device 100 may determine the similarity of each user by performing at least one additional process on the similarity value output from the neural network-based learning model.

The computing device 100 according to an embodiment of the present disclosure may generate a secondary tag indicating a secondary user characteristic related to a sample user group for each of a plurality of users based on the degree of similarity in operation S850.

Although each step of the method or algorithm according to the present disclosure has been described in a sequential order in each flowchart shown in this disclosure, each step may be performed in an order that may be arbitrarily combined according to the present disclosure, in addition to being performed sequentially. may be The description according to this flowchart does not exclude changes or modifications to the method or algorithm, and does not imply that any step is necessary or desirable. In one embodiment, at least some of the steps may be performed in parallel, iteratively or heuristically. In one embodiment, at least some steps may be omitted or other steps may be added.

Various embodiments of the present disclosure may be implemented as software in a machine-readable storage medium. The software may be software for implementing various embodiments of the present disclosure. Software can be inferred from various embodiments of this disclosure by programmers skilled in the art. For example, software may be a program containing machine-readable instructions (eg, code or code segments). A device is a device capable of operating according to a command called from a storage medium, and may be, for example, a computer. In one embodiment, the device may be a computing device 100 according to embodiments of the present disclosure. In one embodiment, the processor of the device may execute the called command, so that components of the device perform functions corresponding to the command. In one embodiment, the processor may be processor 310 according to embodiments of the present disclosure. The storage medium may mean any type of recording medium in which data that can be read by a device is stored. The storage medium may include, for example, ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, and the like. In one embodiment, the storage medium may be memory 320 . In one embodiment, the storage medium may be implemented in a distributed form such as a computer system connected to a network. The software may be distributed, stored, and executed on a computer system or the like. The storage medium may be a non-transitory storage medium. A non-transitory storage medium means a tangible medium regardless of whether data is stored semi-permanently or temporarily, and does not include a signal propagated temporarily.

Although the technical idea of the present disclosure has been described by various embodiments, the technical idea of the present disclosure includes various substitutions, modifications, and changes that can be made within the scope understandable by those skilled in the art to which the present disclosure belongs. include It is also to be understood that such substitutions, modifications and alterations may be included within the scope of the appended claims.

Claims (19)

one or more processors; and
one or more memories storing instructions that, when executed by the one or more processors, cause the one or more processors to perform operations;
The one or more processors, according to the instructions,
Acquiring user information for each of a plurality of users and payment information for each of the plurality of users;
generate one or more primary tags for each of the plurality of users based on the user information and the payment information, wherein the one or more primary tags each indicate one or more primary user characteristics; Each user characteristic is specified by at least one of the user information and the payment information.
determining a degree of similarity between each of the plurality of users and the sample user group based on one or more primary tags for each sample user in a predetermined sample user group among the plurality of users;
Based on the degree of similarity, generate a secondary tag for each of the plurality of users, wherein the secondary tag indicates a secondary user characteristic associated with the sample user group, the secondary user characteristic being associated with the one or more primary users. specified by the tag - ,
determining a predetermined number of users from among the plurality of users in order of the similarity;
determine a proportion of the sample users among the predetermined number of users;
based on the ratio, determine a value of the secondary tag for each of the plurality of users;
The value of the secondary tag indicates whether each of the plurality of users has the secondary user characteristics or a degree of matching with the secondary user characteristics.
According to claim 1,
Each of the one or more primary tags may include one or more tags specified based on at least one selected from among the user information, payment destination, payment item, payment method, payment time, payment number, payment frequency, and payment amount of each of the plurality of users. A device that indicates each of the primary user characteristics.
According to claim 1,
The user information for each of the plurality of users includes behavior information or credit information of each of the plurality of users.
According to claim 1,
The one or more processors,
and determining the degree of similarity based on a difference between a value of a primary tag for each of the plurality of users and a value of a primary tag for each of the sample users.
According to claim 1,
The one or more processors,
determining the degree of similarity based on a difference between the value of the primary tag for each of the plurality of users and the primary tag reference value of the sample user group - the primary tag reference value is the primary tag for each of the sample users - It is a value determined based on the value of the device.
According to claim 1,
The one or more memories,
Further storing a learning model based on a neural network learned based on one or more primary tags for each of the sample users;
The one or more processors,
and determining a degree of similarity between each of the plurality of users and the sample user group based on the learning model.
According to claim 6,
The one or more processors,
Entering a value of one or more primary tags for each of the plurality of users into the learning model;
and determining the degree of similarity based on an output of the learning model.
delete delete A method performed in a computing device comprising one or more processors and one or more memories storing instructions for execution by the one or more processors,
obtaining, by the one or more processors, user information for each of a plurality of users and payment information for each of the plurality of users;
Generating, by the one or more processors, one or more primary tags for each of the plurality of users based on the user information and the payment information, the one or more primary tags indicating one or more primary user characteristics, respectively. and each of the one or more primary user characteristics is specified by at least one of the user information and the payment information;
determining, by the one or more processors, a degree of similarity between each of the plurality of users and the sample user group, based on one or more primary tags for each sample user in a predetermined sample user group among the plurality of users;
generating, by the one or more processors, a secondary tag for each of the plurality of users based on the degree of similarity, wherein the secondary tag indicates a secondary user characteristic related to the sample user group, and the secondary user Characteristics specified by the one or more primary tags;
determining a predetermined number of users from among the plurality of users in an order of high similarity;
determining a proportion of the sample users among the predetermined number of users; and
based on the ratio, determining a value of the secondary tag for each of the plurality of users;
The value of the secondary tag indicates whether each of the plurality of users has the secondary user characteristics or a degree of matching with the secondary user characteristics.
According to claim 10,
Each of the one or more primary tags may include one or more tags specified based on at least one selected from among the user information, payment destination, payment item, payment method, payment time, payment number, payment frequency, and payment amount of each of the plurality of users. Indicating each of the primary user characteristics.
According to claim 10,
The user information for each of the plurality of users includes behavior information or credit information of each of the plurality of users.
According to claim 10,
wherein the determining of the degree of similarity includes determining the degree of similarity based on a difference between a value of a primary tag for each of the plurality of users and a value of a primary tag for each of the sample users.
According to claim 10,
The determining of the similarity may include determining the similarity based on a difference between a primary tag value for each of the plurality of users and a primary tag reference value of the sample user group - the primary tag reference value is and is a value determined based on a value of a primary tag for each of the sample users.
According to claim 10,
The one or more memories further store a neural network-based learning model learned based on one or more primary tags for each of the sample users;
Wherein the determining of the degree of similarity comprises determining a degree of similarity between each of the plurality of users and the sample user group based on the learning model.
According to claim 15,
The step of determining the degree of similarity is,
inputting a value of one or more primary tags for each of the plurality of users into the learning model; and
determining the degree of similarity based on an output of the learned model.
delete delete In a non-transitory computer-readable recording medium recording instructions to be executed on a computer,
The instructions, when executed by one or more processors, cause the one or more processors to:
Acquiring user information for each of a plurality of users and payment information for each of the plurality of users;
generate one or more primary tags for each of the plurality of users based on the user information and the payment information, wherein the one or more primary tags each indicate one or more primary user characteristics; Each user characteristic is specified by at least one of the user information and the payment information.
determining a degree of similarity between each of the plurality of users and the sample user group based on one or more primary tags for each sample user in a predetermined sample user group among the plurality of users;
Based on the degree of similarity, generate a secondary tag for each of the plurality of users, wherein the secondary tag indicates a secondary user characteristic associated with the sample user group, the secondary user characteristic being associated with the one or more primary users. specified by the tag - ,
determining a predetermined number of users from among the plurality of users in order of the similarity;
determine a proportion of the sample users among the predetermined number of users;
determine a value of the secondary tag for each of the plurality of users based on the ratio;
The value of the secondary tag indicates whether each of the plurality of users has the secondary user characteristics or a degree of matching with the secondary user characteristics.

Images