KR102290213B1 - Method for generating customer profile using card usage information and appratus for generating customer profile - Google Patents

Abstract

A method for creating a customer profile is disclosed. The method for generating the customer profile includes: obtaining customer information and card usage information for the customers; preference fields included in the customer profile for each of the customers based on the customer information and the card usage information; calculating a preference score for , and generating the customer profile using the calculated preference score.

Description

Method for creating a customer profile using card usage information and an apparatus for creating a customer profile

Embodiments of the present invention relate to a method for generating a customer profile and an apparatus for generating a customer profile.

With the development of the Internet, various contents are being delivered to customers through the Internet. In addition, the amount of content delivered to customers is also increasing. Accordingly, it is more effective for the customer to not only want to be provided with the content that he or she is interested in from among the variously provided content, but also from the perspective of the business operator to provide the content corresponding to the customer's field of interest. However, there are existing content recommendation methods such as collaborative filtering and content-based filtering, but since these are simply based on a customer's search keyword or purchase history, there is a problem in that the accuracy is lowered.

The problem to be solved by the present invention is to provide a method for generating a customer profile that well represents the customer's propensity, which is the basis of content recommendation, in order to increase the accuracy of content recommendation. In particular, the customer's customer information and card use information An object of the present invention is to provide a method for generating a customer profile using

A method for generating a customer profile for customers according to embodiments of the present invention includes: obtaining customer information and card use information for the customers; based on the customer information and the card use information, for each of the customers calculating a preference score for preference fields included in the customer profile; and generating the customer profile using the calculated preference score.

An apparatus for generating a customer profile for customers according to embodiments of the present invention obtains customer information and card use information about the customers, and based on the customer information and the card use information, for each of the customers A preference score is calculated for preference fields included in the customer profile, and the customer profile is generated using the calculated preference score.

A non-transitory computer readable medium according to embodiments of the present invention stores a computer program including instructions for executing a method for generating a customer profile.

The device for generating a customer profile according to embodiments of the present invention has an effect of generating a customer profile that can indicate characteristics of each customer based on customer information and card use information.

According to the content recommendation method according to the embodiments of the present invention, there is an effect that the accuracy of determining the recommended content for the recommendation target customer is effectively improved.

1 illustrates a content recommendation system for each individual according to embodiments of the present invention.
2 illustrates an apparatus for generating a customer profile according to embodiments of the present invention.
3 is a flowchart illustrating a method of generating a customer profile according to embodiments of the present invention.
4 illustrates a content recommendation apparatus according to embodiments of the present invention.
5 is a diagram for explaining a content recommendation method according to embodiments of the present invention.
6 is a flowchart illustrating a content recommendation method according to embodiments of the present invention.
7 is a flowchart illustrating a content recommendation method according to embodiments of the present invention.
8 is a flowchart illustrating a content recommendation method according to embodiments of the present invention.

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

1 illustrates a content recommendation system for each individual (hereinafter, a content recommendation system) according to embodiments of the present invention. The content recommendation system 10 may include a customer profile generating device 100 , a content recommendation device, and a customer terminal 300 .

Meanwhile, in the present specification, the term "content" refers to data that provides arbitrary information. That is, the contents collectively refer to any type of data including specific information such as events, movies, information, music, advertisements, programs, images, or texts. On the other hand, the content may be one or two or more.

The customer profile generating apparatus 100 may generate a customer profile PRO_CTM for each customer. The generated customer profile PRO_CTM may be stored in the first database 410 .

The customer profile PRO_CTM may be an indicator indicating a customer attribute or preference. The customer profile PRO_CTM may include a plurality of preference fields F1 to Fn; n is a natural number), and a preference score may be assigned to each of the preference fields F1 to Fn.

For example, in the case of the customer profile PRO_CTM for the first customer CTM1, the preference score of the first preference field F1 may be 87 points, and the preference score of the nth preference field Fn may be 95 points. In the case of the customer profile PRO_CTM for the second customer CTM2, the preference score of the first preference field F1 may be 84 points and the nth preference field Fn may be 75 points, but these numbers are only examples.

Each of the preference fields F1 to Fn may include one or more items. According to embodiments, the preference fields F1 to Fn may include at least one of a preferred affiliated store field, a shopping pattern field, a preferred area field, and a preferred item field.

The customer profile generating apparatus 100 may generate the customer profile PRO_CTM based on the customer information CTMI and the card use information CDUI. The customer information CTMI may include general information about the customer, such as the customer's gender, age, preferred industry, occupation, and residence. Card use information (CDUI) is information on the customer's card use (ie, payment history), and may include card use merchants, card use amount, card use items, card use industry, card use region, card use propensity, etc. there is.

Customer information (CTMI) and card use information (CDUI) may exist for each customer.

The customer profile generating apparatus 100 may provide the generated customer profile PRO_CTM to the content recommendation apparatus 200 . According to embodiments, the customer profile generating apparatus 100 stores the generated customer profile PRO_CTM in the first database DB1 410 , and in response to the request of the content recommendation device 200 in the first database 410 . The stored customer profile PRO_CTM may be transmitted to the content recommendation device 200 .

The first database 410 is a database managed by the customer profile generating apparatus 100 , and may store and manage a customer profile PRO_CTM for each customer. According to embodiments, the customer profile PRO_CTM may be stored as a table for each customer.

The content recommendation apparatus 200 may acquire a customer profile PRO_CTM. According to embodiments, the content recommendation apparatus 200 may acquire the customer profile PRO_CTM from the customer profile generating apparatus 100 , but may also directly access the first database 410 to obtain the customer profile PRO_CTM there is.

The content recommendation apparatus 200 may determine the recommended content RCNT for each customer based on the customer profile PRO_CTM. According to embodiments, the content recommendation apparatus 200 may determine the recommended content RCNT for the recommendation target customer and provide the recommended content RCNT to the recommendation target customer's customer terminal 300 .

Meanwhile, in this specification, when the content recommendation device 200 provides the recommended content RCNT to the customer terminal 300 , the content recommendation device 200 transmits the recommended content RCNT itself to the customer terminal 300 . as well as providing an accessible address (such as a link or URL) to the recommended content (RCNT).

According to embodiments, the content recommendation device 200 determines a customer group (hereinafter, similar customer group) having a customer profile PRO_CTM most similar to the customer profile PRO_CTM of the recommended customer, and recommends content to which the similar customer group responded. It can be determined as content (RCNT).

According to embodiments, the content recommendation apparatus 200 generates content profiles (PRO_CNT) of the content, and recommends content corresponding to the content profile (PRO_CNT) most similar to the customer profile (PRO_CTM) of the recommended target customer (RCNT) can be determined as Meanwhile, the content recommendation apparatus 200 may store the generated content profile PRO_CNT in the second database 420 .

According to embodiments, the content profile PRO_CNT may have the same structure as the customer profile PRO_CTM. For example, the content profile PRO_CNT may include preference fields and a preference score constituting the customer profile PRO_CTM.

The customer terminal 300 may be a terminal of a recommended target customer. That is, the recommendation target customer may be a user of the customer terminal 300 .

The customer terminal 300 may receive the recommended content RCNT from the content recommendation device 200 . As described above, the customer terminal 300 may receive the recommended content RCNT itself or an address accessible to the recommended content RCNT from the content recommendation device 200 .

According to embodiments, the customer terminal 300 may detect a reaction of the recommended target customer to the provided recommended content RCNT, and transmit the detection result to the content recommendation apparatus 200 . In this case, the reaction refers to all actions of the customer terminal 300 in response to the reception of the recommended content RCNT, such as execution of the recommended content RCNT and access to the recommended content RCNT.

For example, the customer terminal 300 may be a mobile device, such as a smart phone, PC, tablet PC, or PDA, or a wearable device, but is not limited thereto.

2 illustrates an apparatus for generating a customer profile according to embodiments of the present invention. 1 and 2 , the customer profile generating apparatus 100 may include a preference score calculation module 110 and a customer profile generating module 120 .

Hereinafter, a module referred to in this specification means hardware capable of performing at least one function or hardware including software. That is, a specific module described in this specification may mean a device or circuit capable of performing a corresponding function, or may mean a device in which software capable of performing the above function is executed.

The preference score calculation module 110 analyzes customer information (CTMI) and card use information (CDUI), and preference fields included in customer profile (PRO_CTM) based on customer information (CTMI) and card use information (CDUI) A preference score for (F1 to Fn) can be calculated.

According to embodiments, the preference score calculation module 110 extracts variables from customer information CTMI and card use information CDUI, and preference scores for preference fields F1 to Fn based on the extracted variables can be calculated.

According to embodiments, the preference score calculation module 110 uses techniques such as collaborative filtering, user behavior prediction model, artificial neural network, geographic information system (GIS) spatial analysis, Bayesian analysis, vector dot product, and regression model to provide customer information ( CTMI) and card usage information (CDUI) may determine each customer's preference for preference fields F1 to Fn, and a preference score may be calculated based on the determined preference.

According to embodiments, the preference score calculation module 110 uses cooperative filtering based on the card usage pattern for each affiliated store, such as the number of uses for each affiliated store and the amount used for each affiliated store included in the card use information (CDUI), for each customer for the affiliated store using collaborative filtering. may determine the preference of , and calculate a preference score for the preferred merchant field accordingly.

According to embodiments, the preference score calculation module 110 applies user behavior prediction modeling to customer information (CTMI) and card use information (CDUI) to determine a preference for each consumption pattern of customers, and accordingly, a shopping pattern field It is possible to calculate a preference score for

According to embodiments, the preference score calculation module 110 models the economic activity radius of the customer using the customer information (CTMI) and the card use information (CDUI) to determine a preference for each consumption region of the customer, and according to this, the preference A preference score can be calculated for the local field.

The preference score calculation module 110 may calculate a preference score corresponding to the preference by standardizing (or normalizing) the determined preference.

Since the preference for each of the preference fields F1 to Fn determined by the preference score calculation module 110 may have different standards or units, the preference score calculation module 110 provides the preference fields F1 to Fn. It is possible to standardize the preference for the user according to a common criterion, and calculate a preference score corresponding to the preference according to the standardization result. According to embodiments, the preference score calculation module 110 may calculate a preference score corresponding to a preference by applying an inverse user frequency (IUF) algorithm.

The customer profile generation module 120 may generate the customer profile PRO_CTM using the preference scores of the preference fields F1 to Fn calculated by the preference score calculation module 110 . According to embodiments, the customer profile generating module 120 may generate a customer profile PRO_CTM including preference scores of the calculated preference fields F1 to Fn for each customer.

The customer profile generation module 120 may store the generated customer profile PRO_CTM in the first database 410 . According to embodiments, the customer profile generating module 120 may index and store the customer profile PRO_CTM for each customer.

3 is a flowchart illustrating a method of generating a customer profile according to embodiments of the present invention. The customer profile generating method to be described with reference to FIG. 3 may include operations of the customer profile generating apparatus 100 described with reference to FIGS. 1 to 2 .

1 to 3 , the customer profile generating apparatus 100 may acquire customer information CTMI and card use information CDUI (S110). According to embodiments, the customer profile generating apparatus 100 may acquire customer information CTMI and card use information CDUI by accessing an arbitrary database.

The customer profile generating apparatus 100 may determine a preference for the preference fields F1 to Fn for each customer based on the customer information CTMI and the card use information CDUI ( S120 ). According to embodiments, the customer profile generating apparatus 100 uses techniques such as collaborative filtering, contents based filtering (CBF), user behavior prediction model, artificial neural network, GIS spatial analysis, Bayesian analysis, vector dot product, regression model, etc. The customer's preference for the preference fields F1 to Fn may be determined from the information CTMI and the card use information CDUI.

The customer profile generating apparatus 100 may calculate a preference score for each preference field F1 to Fn based on the determined preference ( S130 ). According to embodiments, the customer profile generating apparatus 100 may calculate a preference score corresponding to the preference by standardizing (or normalizing) the determined preference.

The customer profile generating apparatus 100 may generate a customer profile PRO_CTM using the calculated preference scores for each of the preference fields F1 to Fn ( S140 ). According to embodiments, the customer profile generating apparatus 100 may store the generated customer profile PRO_CTM in the first database 410 for each customer.

The customer profile generating apparatus according to the embodiments of the present invention can generate a customer profile (PRO_CTM) that can indicate characteristics (eg, preferences) of each customer based on customer information (CTMI) and card use information (CDUI) It works.

4 shows a content recommendation apparatus according to embodiments of the present invention, and FIG. 5 is a diagram for explaining a content recommendation method according to embodiments of the present invention.

1 to 5 , the content recommendation apparatus 200 may include a first content recommendation module 210 and a second content recommendation module 220 .

The first content recommendation module 210 determines a customer group (ie, a similar customer group) having a customer profile (PRO_CTM) most similar to the customer profile (PRO_CTM) of the recommended customer, and recommends content (RCNT) to the content that the similar customer group responded to. ) can be determined as

The second content recommendation module 220 generates content profiles (PRO_CNT) of the content, and determines the content corresponding to the content profile (PRO_CNT) most similar to the customer profile (PRO_CTM) of the recommended target customer as the recommended content (RCNT). there is. In some embodiments, the second content recommendation module 220 may determine the recommended content RCNT by calculating a similarity between the customer profile PRO_CTM and the content profile PRO_CNT of the customer.

First, the first content recommendation module 210 will be described.

The first content recommendation module 210 sets response customer groups including customers who responded to each content based on the content usage information (CNTUI), and selects a similar customer group having a similar preference to the recommended target customer from among the response customer groups determined, and content corresponding to the similar customer group (ie, the content responded to by the similar customer group) may be determined as the recommended content RCNT.

According to embodiments, the first content recommendation module 210 uses a customer profile (PRO_CTM) of each customer included in the response customer group and a customer profile (PRO_CTM) of the recommendation target customer, a similar customer group having a similar preference to the recommendation target customer can be decided

For example, the first content recommendation module 210 may generate a profile of the reactive customer group by using the customer profile PRO_CTM of each customer included in the reactive customer group. The profile of the reactive customer group may be generated based on preference scores of the customer profile PRO_CTM of each customer.

The first content recommendation module 210 may calculate a similarity between the response customer group and the recommendation target customer by using the profile of the response customer group and the customer profile PRO_CTM of the recommendation target customer. According to embodiments, the first content recommendation module 210 uses the preference scores of the preference fields of the profile of the response customer group and the preference scores of the preference fields of the customer profile (PRO_CTM) of the recommendation target customer to be between the response customer group and the recommendation target customer. We can calculate the similarity of

For example, the similarity calculated by the first content recommendation module 210 may have a specific value, and the first content recommendation module 210 may include collaborative filtering, contents based filtering (CBF), user behavior prediction model, artificial neural network, The similarity may be calculated using at least one of a GIS spatial analysis, a Bayesian analysis, a vector dot product, and a regression model.

Thereafter, the first content recommendation module 210 may determine a similar customer group based on the calculated similarity.

The similar customer group may be a response customer group having a similarity greater than or equal to a reference value, and may be a response customer group having a maximum similarity, but is not limited thereto.

According to embodiments, the first content recommendation module 210 may determine the similar customer group by additionally considering the calculated similarity and preference scores for the preference fields F1 to Fn of the profile. For example, the first content recommendation module 210 may set, as a similar customer group, a response customer group that is adjacent to a recommendation target customer and a preferred region among response customer groups having a similarity level equal to or greater than a reference value.

The first content recommendation module 210 may determine the content reacted by the determined similar customer group as the recommended content RCNT. According to embodiments, the first content recommendation module 210 may store the recommended content RCNT of the recommendation target customer or may provide the recommended content RCNT to the recommendation target customer.

According to embodiments, the first content recommendation module 210 monitors the interaction between the recommendation target customer and the recommended content (RCNT), and when there is the interaction, the recommendation target customer responds to the recommended content (RCNT) customer group can be added to The similarity may be calculated again based on the added reactive customer group, and through this, the accuracy of the content recommendation method (or model) may be increased.

According to embodiments, the first content recommendation module 210 determines whether the content usage information (CNTUI) is updated, determines whether a customer group responding to each content is updated, and, if the change exists, a changed response customer group It is possible to determine the recommended content (RCNT) again based on .

Now, the second content recommendation module 220 will be described.

The second content recommendation module 220 may determine content similar to the recommendation target customer as the recommended content RCNT.

The second content recommendation module 220 may generate the content profile PRO_CNT using the content use information CNTUI and the customer profile PRO_CTM. The content usage information (CNTUI) may include information on the use of each of the customers' content. For example, the content usage information (CNTUI) may include, for each content, information about a customer who has interacted (eg, used or reacted) with the corresponding content.

The second content recommendation module 220 may set a reactive customer group including customers who responded to each content based on the content use information (CNTUI), and select the customer profile (PRO_CTM) of the reactive customers included in each response customer group. content profiles (PRO_CNTs) of the corresponding content can be created using According to embodiments, the second content recommendation module 220 may generate a content profile (PRO_CNT) by using the preference scores of the preference fields included in the customer profile (PRO_CTM) of the response customers included in the response customer group for each content. there is.

A process of generating the content profile PRO_CNT will be described with reference to the example shown in FIG. 5 . Referring to FIG. 5 , the second content recommendation module 220 performs interactions with the first content CNT1 based on the content use information CNTUI, customers CTM1-1, CTM1-2, ... ) may be determined as the first response customer group (RG1), and customers (CTM2-1, CTM2-2, ...) who have interacted with the second content (CNT2) may be determined as the second response customer group (RG2). can In this case, each of the first content CNT1 and the second content CNT2 may be referred to as non-responsive content for each of the first responding customer group RG1 and the second responding customer group RG2 .

That is, since the customers who responded to the specific content express their preference or interest in the specific content, the specific content may be content that meets the customer's preference.

The second content recommendation module 220 is based on the customer profiles (PRO_CTM) of the customers (CTM1-1, CTM1-2, ...) included in the first response customer group (RG1) to the first content (CNT1) You can create a content profile (PRO_CNT) for

The second content recommendation module 220 may determine content having a similar preference to the recommendation target customer as the recommended content RCNT by using the content profile PRO_CNT of the content and the customer profile PRO_CTM of the recommendation target customer.

The second content recommendation module 220 may calculate a similarity between each content and the recommendation target customer by using the content profiles PRO_CNTs of the content and the customer profile PRO_CTM of the recommendation target customer. According to embodiments, the second content recommendation module 220 recommends each content using the preference scores of the preference fields of the content profile (PRO_CNT) of the content and the preference scores of the preference fields of the customer profile (PRO_CTM) of the recommended customer. The similarity between target customers can be calculated.

For example, the similarity calculated by the second content recommendation module 220 may have a specific value, and the second content recommendation module 220 may include collaborative filtering, contents based filtering (CBF), user behavior prediction model, artificial neural network, The similarity may be calculated using at least one of a GIS spatial analysis, a Bayesian analysis, a vector dot product, and a regression model.

Thereafter, the second content recommendation module 220 may determine a recommended content (RCNT) among the content based on the calculated similarity. Also, the second content recommendation module 220 may set a content recommendation ranking based on the calculated similarity.

For example, the second content recommendation module 220 may determine content having the highest similarity or content having a similarity greater than or equal to a reference value as the recommended content (RCNT), but is not limited thereto.

According to embodiments, the second content recommendation module 220 may determine the similar customer group by additionally considering the calculated similarity and the preference scores for the preference fields F1 to Fn of the content profile. For example, the second content recommendation module 220 may determine, as the recommended content (RCNT), the content most similar to the preferred affiliated store of the recommended customer from among the content having a similarity greater than or equal to the reference value.

According to embodiments, the second content recommendation module 220 determines whether the content use information (CNTUI) is updated, determines whether a customer group responding to each content is updated, and, if the change exists, a changed response customer group It is possible to create a content profile (PRO_CNT) of the content based on the .

For example, the second content recommendation module 220 feeds back the history of content use information (CNTUI) to update the response customer group by applying the feedback result to the response customer group for each content, and based on the updated response customer group, You can create a content profile (PRO_CNT).

Profiles according to embodiments of the present invention are universal because they are based on a standardized score calculated by standardizing preferences, and are expandable because various preference scores can be calculated as the preference field is changed.

6 is a flowchart illustrating a content recommendation method according to embodiments of the present invention. The content recommendation method to be described with reference to FIG. 6 may include operations of the content recommendation apparatus 200 described with reference to FIGS. 1 to 5 .

1 to 6 , the content recommendation apparatus 200 may set reactive customer groups for each content ( S210 ). According to embodiments, the content recommendation apparatus 200 may select content having a use history based on content use information CNTUI, and may set a reactive customer group for each content having a use history.

The content recommendation apparatus 200 may generate a profile of the reactive customer group by using the customer profile PRO_CTM of the customers included in the reactive customer group ( S220 ).

The content recommendation apparatus 200 may calculate a similarity (eg, first similarity) between the reaction customer group and the recommendation target customer based on the profile of the response customer group and the customer profile PRO_CTM of the recommendation target customer ( S230 ).

The content recommendation apparatus 200 may determine the recommended content (RCNT) for the recommendation target customer based on the calculated similarity ( S240 ). According to embodiments, the content recommendation apparatus 200 may determine, as the recommended content (RCNT), content to which a customer group most similar to the recommendation target customer responds.

The content recommendation apparatus 200 according to embodiments of the present invention has an effect of calculating a similarity between a recommendation target customer and a customer group responding to the content, and determining the recommended content RCNT based on the calculated similarity.

7 is a flowchart illustrating a content recommendation method according to embodiments of the present invention. The content recommendation method to be described with reference to FIG. 7 may include operations of the content recommendation apparatus 200 described with reference to FIGS. 1 to 5 .

1 to 7 , the content recommendation apparatus 200 may set response customer groups for each content ( S310 ). According to embodiments, the content recommendation apparatus 200 may select content having a use history based on content use information CNTUI, and may set a reactive customer group for each content having a use history.

The content recommendation apparatus 200 may generate a content profile PRO_CNT for each content based on the customer profile PRO_CTM of customers included in the reactive customer group ( S320 ).

The content recommendation apparatus 200 may calculate a similarity (eg, second similarity) between the content and the recommendation target customer based on the content profile PRO_CNT of the content and the customer profile PRO_CTM of the recommendation target customer ( S330 ).

The content recommendation apparatus 200 may determine the recommended content RCNT for the recommendation target customer based on the calculated similarity ( S340 ). According to embodiments, the content recommendation apparatus 200 may determine the content most similar to the recommended target customer as the recommended content RCNT.

The content recommendation apparatus 200 according to embodiments of the present invention may determine the characteristics of the content (ie, the content profile) based on the characteristics (ie, the customer profile) of the customers who have interacted with the content, and There is an effect that can determine recommended content based on the content.

8 is a flowchart illustrating a content recommendation method according to embodiments of the present invention. The content recommendation method to be described with reference to FIG. 8 may include operations of the content recommendation apparatus 200 described with reference to FIGS. 1 to 5 .

1 to 8 , the content recommendation apparatus 200 may set response customer groups for each content ( S410 ). According to embodiments, the content recommendation apparatus 200 may select content having a use history based on content use information CNTUI, and may set a reactive customer group for each content having a use history.

The content recommendation apparatus 200 may generate a content profile PRO_CNT for each content based on the customer profile PRO_CTM of customers included in the reactive customer group ( S420 ).

The content recommendation apparatus 200 may generate a profile of the reactive customer group by using the customer profile PRO_CTM of the customers included in the reactive customer group ( S430 ).

The content recommendation apparatus 200 may calculate a similarity (ie, the first similarity) between the reaction customer group and the recommendation target customer based on the profile of the response customer group and the customer profile PRO_CTM of the recommendation target customer ( S440 ).

The content recommendation apparatus 200 may determine a similar customer group for the recommendation target customer based on the similarity between the reaction customer group and the recommendation target customer ( S450 ). According to embodiments, the similar customer group may mean a customer group most similar to the recommended target customer among the response customer groups, but is not limited thereto.

The content recommendation device 200 is configured to provide a similarity (eg, the third similarity) can be calculated (S460).

On the other hand, the difference between step S460 described with reference to FIG. 8 and step S330 described with reference to FIG. 7 is that, in step S460, a similarity group similar to the recommendation target customer responds by calculating the similarity between the content and the recommendation target customer, In step S330, the similarity between the content and the recommendation target customer is calculated by the general reaction customer group regardless of whether the recommendation target customer is similar.

The content recommendation apparatus 200 according to embodiments of the present invention generates a content profile of the content based on the customer's response to the content, and recommends it from among the content that a similar customer group similar to the recommendation target customer among the response customer groups responded to the content. There is an effect that content similar to the target customer can be determined as the recommended content (RCNT).

Methods according to embodiments of the present invention may be implemented as instructions that are stored in a computer-readable storage medium and executed by a processor.

A storage medium, whether directly and/or indirectly, in a raw, formatted, organized or any other accessible state, may include a relational database, a non-relational database, an in-memory database, Alternatively, it may include a database, including a distributed one, such as any other suitable database capable of storing data and allowing access to such data through a storage controller. In addition, the storage medium includes a primary storage device (storage), a secondary storage device, a tertiary storage device, an offline storage device, a volatile storage device, a non-volatile storage device, a semiconductor storage device, a magnetic storage device, an optical storage device, a flash. It may include any type of storage device, such as a storage device, a hard disk drive storage device, a floppy disk drive, magnetic tape, or other suitable data storage medium.

As used herein, instructions are assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state setting data, or object-oriented such as Smalltalk, C++, etc. It may be either source code or object code written in any combination of a programming language and one or more programming languages, including conventional procedural programming languages such as "C" programming languages or similar programming languages.

Although the present invention has been described with reference to the embodiment shown in the drawings, this is merely exemplary, and those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. Accordingly, the true technical protection scope of the present invention should be determined by the technical spirit of the appended claims.

10: Content Recommendation System
100: customer profile generating device
200: content recommendation device
300: customer terminal
410: first database
420: second database
CTMI: Customer Information
CDUI: Card Usage Information
PRO_CTM: Customer Profile
CNTUI: Content Usage Information
PRO_CNT: content profile
RCNT: Recommended Content

Claims (10)

A method for generating a customer profile for customers by a customer profile generating device, the method comprising:
obtaining customer information and card use information for the customers;
calculating a preference score for preference fields included in the customer profile for each of the customers based on the customer information and the card usage information; and
generating the customer profile using the calculated preference score;
The preference fields include a preference merchant preference field, a shopping pattern preference field, a preference area preference field, and a preference item field,
Calculating a preference score for the preference fields includes:
modeling a radius of economic activity of the customers through a GIS spatial analysis technique using the customer information and the card use information;
determining a preference for each consumption region of the customers according to the modeling result; and
calculating a preference score for the preferred area preference field based on a preference for each of the customers' consumption areas;
How to create a customer profile. The method of claim 1, wherein calculating the preference score comprises:
determining a customer's preference for the preference fields from the customer information and the card usage information using at least one of contents based filtering (CBF), collaborative filtering, user behavior prediction model, and GIS spatial analysis; and
Normalizing the preference for the preference fields according to a common criterion, and calculating a preference score corresponding to the preference according to the standardization result,
How to create a customer profile. The method of claim 2, wherein calculating the preference score comprises:
Comprising the step of calculating the preference score by normalizing the determined preference using an inverse user frequency (IUF) algorithm,
How to create a customer profile. According to claim 1,
The preference fields included in the customer profile include at least one of a preference merchant preference field, a shopping pattern preference field, a preference area preference field, and a preference item field,
How to create a customer profile. According to claim 1,
The preference fields included in the customer profile include a preferred merchant preference field,
Calculating a preference score for the preference fields includes:
determining each customer's preference for an affiliated store by using collaborative filtering based on at least one of the number of times of use by affiliated store, the amount of use by affiliated store, and card use pattern by affiliated store included in the card use information; and
calculating a preference score for the preferred merchant preference field based on each customer's preference for the merchant;
How to create a customer profile. According to claim 1,
The preference fields included in the customer profile include a shopping pattern preference field,
Calculating a preference score for the preference fields includes:
determining a preference for each customer's consumption pattern by applying user behavior prediction modeling to the customer information and the card use information; and
calculating a preference score for a shopping pattern preference field based on a preference for each of the customers' consumption patterns;
How to create a customer profile. delete According to claim 1,
Further comprising the step of storing the created customer profile in a database for each customer,
How to create a customer profile. A non-transitory computer readable storage medium storing a computer program comprising instructions for executing the method for generating a customer profile according to claim 1 . A computer program stored on a non-transitory computer readable storage medium comprising instructions for executing the method for generating a customer profile according to claim 1 .

Images