KR101661689B1 - Method for recommending card goods and card company server - Google Patents

Abstract

According to an embodiment of the present invention, there is provided a method for recommending a card product to a user in a card issuer server, the method comprising the steps of: calculating a weight CF of a Category of business frequency (CF) generating a first vector table including at least one of a weight value, an inverse goods benefit frequency (IGF) value, a weight value, and a normalization value; A second vector table including at least one of a weight value of the industry frequency, a weight value, and the normalization value using the service benefit information of each card product, Generating for each of the card merchandise; Calculating a similarity value between the first vector table and the second vector table; And a step of selecting a recommended card product based on the calculated similarity value.

Description

METHOD FOR RECOMMENDING CARD GOODS AND CARD COMPANY SERVER "

The present invention relates to a method of recommending a card product and a card company server, which calculates a similarity value between a first vector table concerning card usage pattern information of a user and a second vector table related to service benefit information of a card product, A method of recommending a card product based on a value, and a card company server recommending a card product.

In the past, when recommending a card product, it was simply a method of recommending a newly introduced card product by sorting the number of times of use and amount of settlement for each business type, recommending a card product specialized for the business type concerned.

Accordingly, there is a problem in that the cost of the card issuer is wasted because it is highly likely that the card product or the newly released card product specific to a specific business type is derived and issued and then gradually excluded from actual use and strategized as a human card.

In addition, since many card products provide various kinds of benefit services rather than being specialized in a specific business type, it is difficult to recommend a user-customized card product through a simple business type usage amount.

Therefore, there is an urgent need for a method of recommending a user-customized card product through a service recommending a similar card product suitable for a practical consumption life.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems of the conventional art, and it is an object of the present invention to calculate a similarity value between a first vector table concerning card usage pattern information of a user and a second vector table concerning service benefit information of a card product, A method of recommending a card product based on the similarity value, and a card company server recommending the card product.

The objects of the present invention are not limited to the above-mentioned objects, and other objects not mentioned can be clearly understood from the following description.

According to an aspect of the present invention, there is provided a method for recommending a card product to a user in a card company server, the method comprising: generating a first vector table including at least one of a weight value of a business frequency (CF weight), an inverse goods benefit frequency (IGF) value, a weight value, and a normalization value; A second vector table including at least one of a weight value of the industry frequency, a weight value, and the normalization value using the service benefit information of each card product, Generating for each of the card merchandise; Calculating a similarity value between the first vector table and the second vector table; And a step of selecting a recommended card product based on the calculated similarity value.

The first vector table generation step may include a step of generating the first vector table by multiplying the weight value of the industry frequency by an inverse product frequency (IGF) 1 Vector table can be created.

The generating of the second vector table may generate the second vector table including the weight value of the sector frequency, which is identical to the CF weight value of each industry type.

The first vector table or the second vector table may include a weight value of the industry frequency (CF weight) calculated by weighting the industry frequency value by logarithmic.

The first vector table or the second vector table may be calculated by summing the squares of the weight values for each of a plurality of types of business and calculating a query length Length) value.

The first vector table or the second vector table may include the normalization value calculated for each type of business by associating the weight value with the query length value .

The similarity value calculation step calculates the similarity value by summing the calculated vector values through the operation of multiplying the normalization value included in the first vector table and the normalization value included in the second vector table by business type , The similarity value can be calculated.

Wherein the similarity value calculating step compares the first vector table and a second vector table corresponding to each of the plurality of card products to calculate respective similarity values, The card merchandise for the second vector table corresponding to the highest similarity value can be selected as the recommended card merchandise.

According to another embodiment of the present invention, a CF weight value of a category frequency of a business (CF) value, an inverse product frequency (IGF) A first vector table including at least one of an inverse goods benefit frequency value, a weight value, and a normalization value, and generates a first vector table including weight values of the industry frequency A vector table generating unit for generating a second vector table including at least one of a CF weight value, a weight value, and the normalization value for each of a plurality of card merchandise; A similarity calculating unit for calculating a similarity value between the first vector table and the second vector table; And a card merchandise recommending section for selecting a recommended card merchandise based on the calculated similarity value.

According to the embodiment of the present invention, since the card merchandise can be recommended by using a complex element, the influence of the simple card type of each business type can be reduced and the accuracy of the card merchandise recommendation service can be improved.

In addition, according to an embodiment of the present invention, it is possible to use a CF weight value of a business frequency frequency obtained by weighting a CF frequency value by log, Can be solved.

Also, according to an embodiment of the present invention, it is possible to provide a card product which is expected to benefit substantially according to the use of the card, by appropriately excluding the arbitrary usage amount of the business type or the arbitrary judgment of the card company employee, Thus, it is possible to satisfy the customer's needs, reduce the ratio of the human card, and reduce the card issuing cost.

Also, according to an embodiment of the present invention, consumption patterns of customers are vectorized, and the similarity degree of consumption among various customers can be grouped according to conditions such as area, age, and time, So that the advertising effect can be increased.

It should be understood that the effects of the present invention are not limited to the above effects and include all effects that can be deduced from the detailed description of the present invention or the configuration of the invention described in the claims.

1 is a diagram illustrating a card company server according to an embodiment of the present invention.
FIG. 2 is a graph showing a comparison between a business frequency (CF) value and a business frequency weight (CF weight) value according to an exemplary embodiment of the present invention.
3 is a view illustrating a process of selecting a recommended card product according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described with reference to the accompanying drawings. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In order to clearly illustrate the present invention, parts not related to the description are omitted, and similar parts are denoted by like reference characters throughout the specification.

Throughout the specification, when a part is referred to as being "connected" to another part, it includes not only "directly connected" but also "indirectly connected" . Also, when an element is referred to as "comprising ", it means that it can include other elements, not excluding other elements unless specifically stated otherwise.

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

1 is a diagram illustrating a card company server according to an embodiment of the present invention.

Referring to FIG. 1, the card issuer server 100 may include a vector table generating unit 110, a similarity calculating unit 120, and a card product recommending unit 130.

First, the card issuer server 100 can store the card usage pattern information for each business type in a database and manage it. Here, the card usage pattern information may include information related to a user's card payment history (for example, a payment card, a payment amount, a payment business type, a payment date, payment benefit information, and the like).

Table 1 shows the card usage pattern information of the user according to an exemplary embodiment of the present invention.

Payment card Amount of payment Payment business Payment Date Billing Benefits A card $ 20 coffee 2014.08.01 5% off A card 30,000 won food 2014.08.02 5% Earn B card $ 15 traffic 2014.08.03 10% off B card $ 24 coffee 2014.08.04 -

Specifically, when the user's card is used at the merchant's office, the card issuer server 100 may store the card usage pattern information including the card payment details after performing settlement through the card, If it is stored, the card usage pattern information can be updated by adding the card payment details.

Upon updating the card usage pattern information, the card issuer server 100 can update the card usage pattern information in real time in accordance with the use of the card, and collectively use the card usage pattern information in accordance with a predetermined period (for example, one week) It can also be updated.

The card issuer server 100 can store and manage the service benefit information for each type of card merchandise in the database. Here, the service benefit information may include card merchandise, benefit category, benefit service information, and the like.

Table 2 shows the service benefit information of the card merchandise according to an embodiment of the present invention, and may be only a part for convenience of explanation.

Card goods Benefit industry Benefit Services A card coffee 5% off A card food 5% Earn B card coffee 10% off B card traffic 10% Earned

Specifically, when the card company 100 releases a card product from a card company, it can store the service benefit information of the newly released card product. If the benefit details of the already released card product are changed, Information can be updated.

The card issuer server 100 can manage the above-described card usage pattern information and service benefit information separately for each type of business (e.g., coffee, food, traffic, movie, hospital, shopping, etc.).

The vector table generating unit 110 can generate the first vector table using the card usage pattern information of the user. Here, the first vector table includes a CF weight value, an Inverse Goods benefit frequency (IGF) value, a weight value, and a normalization value of a category of business frequency (CF) And may include at least one. In this regard, a detailed description will be given with reference to Table 3.

Table 3 is a first vector table according to an embodiment of the present invention, and may be only a part for convenience of explanation.

Sectors CF weight value IGF value Weight value Normalization value coffee 2.18 2.00 4.35 0.47 food 2.51 2.30 5.76 0.62 Travel 1.00 3.00 3.00 0.32 traffic 1.90 2.70 5.14 0.55

The vector table generating unit 110 may weight the category frequency of the business frequency by a logarithmic value to calculate the CF weight value of the business frequency. Here, the category frequency (CF) may indicate how many card usage counts for a certain type of business are within a predetermined period (for example, the last one month), and the larger the business type frequency (CF) The proportion of card use in products can be high.

According to one embodiment, the formula for calculating the weight value of the industry frequency (CF weight) is as follows.

For example, the vector table generating unit 110 may generate 1 + log (15) as the CF weight value for the coffee industry if the CF frequency value for the coffee industry is 15, (32), ie, about 2.51, can be calculated as the CF weight value for the food industry if the CF for the food industry is 32 (1), ie, 1, can be calculated as the CF weight value of the industry frequency for the travel industry if the CF frequency value for the travel industry is 1, If the CF value is 8, 1 + log (8), that is, about 1.90, can be calculated as the CF weight value for the traffic type.

The vector table generation unit 110 may generate a first vector table including a weight value of a business frequency (CF weight) calculated for each of a plurality of types of business.

As described above, according to the embodiment of the present invention, by using the weight value of the industry frequency (CF weight) calculated by weighting the industry frequency (CF) value by logarithm, the difference of the industry frequency (CF) Can be prevented. In this regard, a detailed description will be made with reference to Fig.

FIG. 2 is a graph showing a comparison between a business frequency (CF) value and a business frequency weight (CF weight) value according to an exemplary embodiment of the present invention.

The abscissa of the first graph and the abscissa of the second graph are the CF values, the vertical axis of the first graph is the CF value, and the vertical axis of the second graph is the CF weight value.

First, as shown in the first graph, the difference between the frequency values of the industry frequency (CF) can be large because it is a direct proportional form in which the frequency of the industry frequency (CF) of the vertical axis increases as the frequency of the industry frequency (CF) of the horizontal axis becomes larger.

The difference between the values of the weight values of the industry frequency (CF weight) can be expressed by the following equation (2): < EMI ID = It may not appear large.

For example, when the industry frequency (CF) value for the food industry is 32 and the industry frequency (CF) value for the travel industry is 1, the difference in the industry frequency (CF) value is 31, The difference of the CF weight values may not be as large as 1.51.

Thus, according to an embodiment of the present invention, the difference in the frequency of business frequency (CF) values is shown to be large for each type of business by using the weight value (CF weight) of the business frequency frequency obtained by weighting the business frequency The problem can be solved.

In addition, according to the embodiment of the present invention, since the influence of the simple card type for each type of business can be reduced, card items can be recommended using a composite element such as a weight value and a normalization value to be described later It is possible to improve the accuracy of the card merchandise recommendation service.

The vector table generating unit 110 may calculate an inverse goods benefit frequency (IGF) value.

According to one embodiment, the formula for calculating the inverse product frequency (IGF) value is as follows.

Here, N is the number of card merchandise items, and GF is the goods benefit frequency (GF). The number of card items (N) represents the total number of card merchandise that the card company has released. The product frequency (GF) can indicate the number of benefits provided when using a card for a specific business type. This can mean that benefits are often provided within an industry.

For example, if the number of card items (N) is 1,000,000, the product frequency (GF) value for the coffee industry is 10,000, the product frequency (GF) value for the food industry is 5,000, ) Value is 1,000 and the product frequency (GF) value for the traffic industry is 2,000, the vector table generation unit 110 can calculate 2 as the inverse product frequency (IGF) value for the coffee industry, (IGF) value of about 2.30, and IGF value of travel industry can be calculated as 3, and IGF value of transportation industry is about 2.70 can be calculated.

The vector table generating unit 110 may generate a first vector table including an inverse product frequency (IGF) value calculated for each of a plurality of types of business.

The vector table generation unit 110 may calculate a weight value.

According to one embodiment, the formula for calculating the weight value is as follows.

The vector table generating unit 110 may calculate a weight value by multiplying the weight value of the industry frequency (CF weight) by the value of the inverse product frequency (IGF).

For example, the vector table generation unit 110 multiplies the value of the industry weight frequency (CF weight) of the coffee industry by about 2.18 and the value of the inverse product frequency (IGF) of the coffee industry to 2 to calculate a weight value ) Value. As described above, the operation of multiplying the CF weight value by the inverse product frequency (IGF) value is performed for each type of industry to determine the weight value 5.76, weight value for travel industry, and weight value for traffic industry, 5.14.

The vector table generating unit 110 may generate a first vector table including a weight value calculated for each of a plurality of types of business.

The vector table generation unit 110 may calculate a query length value.

According to one embodiment, a formula for calculating a query length value is as follows.

The vector table generation unit 110 may calculate a query length value by summing squares of weight values of a plurality of types of business and performing a root operation on the sum value.

, 즉, 약 9.36을 쿼리 길이(Query Length) 값으로 산출할 수 있다.For example, as described above, the weight value for the coffee industry is about 4.35, the weight value for the food industry is about 5.76, the weight value for the travel industry is 3, the weight for the traffic industry Weight value 5.14, the vector table generation unit 110 generates

, That is, about 9.36 can be calculated as a query length value.

The vector table generating unit 110 may generate a first vector table including a query length value and may temporarily store the vector table for use in calculating a normalization value to be described later.

The vector table generating unit 110 may calculate a normalization value.

According to one embodiment, the formula for calculating the normalization value is as follows.

The vector table generation unit 110 may calculate a normalization value by dividing a weight value by a query length value.

For example, the vector table generating unit 110 may calculate a normalization value of 0.47 for the coffee industry by dividing the CF weight value (CF weight) value of 4.35 by the query length value of 9.36 As described above, the computation for dividing the CF weight value for each business type by the query length value is performed for each type of business, and the normalization value of the food industry is about 0.62, The normalization value of about 0.32 and the normalization value of 0.55 for the traffic industry, respectively.

The vector table generation unit 110 may generate a first vector table including a normalization value calculated for each of a plurality of types of business.

The vector table generating unit 110 can generate the second vector table using the service benefit information for each type of card merchandise. At this time, the vector table generating unit 110 can generate the second vector table for each of the plurality of card merchandise. For example, when the number of card merchandise is 1,000, 1000 second vector tables can be generated have.

Here, the second vector table may include at least one of a CF weight value, a weight value, and a normalization value of a category of business frequency (CF). In this regard, a detailed description will be described with reference to Table 4, and duplicated description of the first vector table in the description of the second vector table will be omitted.

Table 4 is a second vector table for one card product according to an embodiment of the present invention, and may be only a part of the second vector table for convenience of explanation.

Sectors CF weight value Weight value Normalization value coffee 1.00 1.00 0.36 food 1.48 1.48 0.78 traffic 1.00 1.00 0.36 movie 1.00 1.00 0.36 hospital 1.30 1.30 0.60 Online Shopping 1.00 1.00 0.36

The vector table generation unit 110 may calculate the weight value of the industry frequency (CF weight).

For example, if the industry frequency (CF) value for the coffee, traffic, movie, and online shopping industries is 1, then the vector table generation unit 110 generates a vector weight value of 1 (CF) value for the food industry can be calculated to be about 1.48 as the CF weight value for the food industry, and the frequency of the industry for the hospital industry (CF ) Value of 2, it is possible to calculate about 1.30 as the CF weight value for the hospital industry.

The vector table generating unit 110 may generate a second vector table including a weight value of the industry frequency (CF weight) calculated for each of a plurality of types of business.

The vector table generation unit 110 may generate a second vector table including a weight value corresponding to the industry weight frequency (CF weight) value and industry type, respectively. That is, the formula for obtaining the weight value included in the first vector table and the weight value included in the second vector table may be different.

, 즉, 약 2.81을 쿼리 길이(Query Length) 값으로 산출할 수 있다.The vector table generation unit 110 uses the weight value of each type of business,

, That is, about 2.81 can be calculated as a query length value.

The vector table generation unit 110 may calculate a normalization value by dividing the weight value of each type of business by the value of the query length. That is, the vector table generating unit 110 can calculate about 0.36 as the normalization value for the coffee, traffic, movie, and online shopping business, and calculates about 0.78 as the normalization value for the food business type And can calculate about 0.60 as the normalization value for the hospital industry.

The similarity calculating unit 120 may calculate a similarity value between the first vector table and the second vector table. At this time, when a first unit vector and a second vector table are generated, a sector type unit vector is generated through a normalization value. Therefore, the similarity calculation unit 120 uses the normalization value, Value can be calculated.

Specifically, the similarity calculation unit 120 calculates the sum of the calculated vector values by multiplying the normalization value included in the first vector table and the normalization value included in the second vector table for each type of business The similarity value can be calculated.

According to one embodiment, the formula for calculating the similarity value is as follows.

With respect to the method of calculating the similarity value, a detailed description will be described with reference to Table 5. [

Table 5 is a table including a first vector table, a second vector table, and vector values according to an embodiment of the present invention, and may be only a part for convenience of explanation.

Sectors The first vector table The second vector table
Vector value CF weight value IGF value Weight value Normalization value CF weight value Weight value Normalization value coffee 2.18 2.00 4.35 0.47 1.00 1.00 0.36 0.17 food 2.51 2.30 5.76 0.62 1.48 1.48 0.78 0.48 Travel 1.00 3.00 3.00 0.32 0 0 0 - traffic 1.90 2.70 5.14 0.55 1.00 1.00 0.36 0.20 movie 0 2.00 0 0 1.00 1.00 0.36 - hospital 0 2.30 0 0 1.30 1.30 0.60 - Online Shopping 0 1.82 0 0 1.00 1.00 0.36 -

Referring to Table 5, the normalization value for the coffee industry in the first vector table is about 0.47, and the normalization value for the coffee industry in the second vector table is about 0.36.

In addition, the normalization value for the food industry of the first vector table is about 0.62, and the normalization value for the food industry of the second vector table is about 0.78.

In addition, the normalization value of the first vector table for the traffic industry is about 0.55, and the normalization value for the traffic industry of the second vector table is about 0.36.

The similarity calculation unit 120 performs an operation of multiplying the normalization value 0.47 for the coffee industry in the first vector table and the normalization value 0.36 for the coffee industry in the second vector table to obtain a vector value of about 0.17 And can calculate a vector value of about 0.48 by performing an operation of multiplying the normalization values for the food industry and performing an operation of multiplying the normalization values for the traffic industry to obtain vectors Value can be calculated to be about 0.20, and the similarity value of 0.85 can be calculated by summing the vector values 0.17, 0.48 and 0.20 respectively calculated for each type of industry.

In this case, since the value of either the first vector table or the second vector table is 0 in the travel, movie, hospital, and online shopping industry, the vector value may be zero There is no influence on the calculation of the similarity value.

As described above, the similarity calculation unit 120 can calculate the card usage pattern of the user and the similarity value of one card product. If the card product is a plurality of cards, The similarity value can be calculated.

For example, the vector table generation unit 110 may generate a first vector table for a card usage pattern of a user, and generate a second vector table corresponding to each of the A card product, the B card product, and the C card product have.

Thereafter, the similarity calculation unit 120 may compare the first vector table and the second vector table corresponding to each of the plurality of card products, and calculate the similarity value.

Specifically, the similarity calculation unit 120 calculates the similarity value between the first vector table and the second vector table corresponding to the A-card product, the similarity value between the first vector table and the second vector table corresponding to the B-card product, The similarity value between the vector table and the second vector table corresponding to the C-card product can be respectively calculated.

The card merchandise recommendation unit 130 can select the recommended card product based on the similarity value calculated by the similarity degree calculation unit 120. [

For example, the similarity calculating unit 120 compares the card usage pattern of the user with the benefits of the A-card merchandise, the B-card merchandise, and the C-card merchandise to calculate 0.7 as the similarity value corresponding to the A-card merchandise, 0.5 is calculated as the similarity value corresponding to the card merchandise, and 0.9 can be calculated as the similarity value corresponding to the C card merchandise.

Then, the card merchandise recommendation unit 130 compares the similarity values corresponding to the A card merchandise, the B card merchandise, and the C card merchandise, respectively, and selects the C card merchandise corresponding to the highest similarity value as the recommended card merchandise .

3 is a diagram illustrating a process of selecting a recommended card product according to an embodiment of the present invention.

First, the card issuer server 100 can store the card usage pattern information of the user and the service benefit information of the card merchandise in a database for each type of business and manage it.

In step S301, the card issuer server 100 can generate the first vector table using the card usage pattern information of the user. At this time, the card issuer server 100 can generate the first vector table using only the usage pattern information of the specific card among the plurality of cards issued to the user, and can use the card usage pattern within the predetermined period (last one month) It is also possible to generate the first vector table using only the information.

In step S302, the card issuer server 100 can generate the second vector table using the service benefit information for each type of card merchandise. At this time, the card issuer server 100 may repeat step S302 for the number of card merchandise issued by the credit card company to generate a second vector table corresponding to a plurality of card merchandise.

In step S303, the card issuer server 100 can calculate the similarity value between the first vector table and the second vector table. At this time, the card issuer server 100 can compare the first vector table and each of the plurality of second vector tables, and calculate the similarity value for each of them.

In step S304, the card issuer server 100 can select a recommended card product based on the similarity value calculated in step S303. At this time, the card issuer server 100 compares the similarity values of the first vector table and the plurality of second vector tables, and selects the card product for the second vector table having the highest similarity value as the recommended card product .

Thereafter, the card issuer server 100 can provide information on the recommended card product selected in step S304 to the user terminal (not shown). At this time, the information on the recommended card product is provided in the form of a notification message through the PUSH method can do.

As described above, according to the embodiment of the present invention, it is possible to provide a card product which is expected to benefit substantially according to the use of the card, by simply analyzing the actual consumption pattern of the user, It is possible to satisfy the customer's needs, reduce the ratio of the human card, and reduce the card issuing cost.

Also, according to an embodiment of the present invention, consumption patterns of customers are vectorized, and the similarity degree of consumption among various customers can be grouped according to conditions such as area, age, and time, So that the advertising effect can be increased.

It will be understood by those skilled in the art that the foregoing description of the present invention is for illustrative purposes only and that those of ordinary skill in the art can readily understand that various changes and modifications may be made without departing from the spirit or essential characteristics of the present invention. will be. It is therefore to be understood that the above-described embodiments are illustrative in all aspects and not restrictive. For example, each component described as a single entity may be distributed and implemented, and components described as being distributed may also be implemented in a combined form.

The scope of the present invention is defined by the appended claims, and all changes or modifications derived from the meaning and scope of the claims and their equivalents should be construed as being included within the scope of the present invention.

100: Card issuer server
110: vector table generation unit
120:
130: Card product recommendation section

Claims (9)

A method for recommending a card product to a user in a card company server,
(CF) value, an inverse goods benefit frequency (IGF) value, a weight value, and a normalized value of a category of business frequency (CF) Generating a first vector table including at least one of a normalization value and a normalization value;
(CF weight) value, the weight value, the normalization value, and the weight value of the industry frequency (CF weight) value of the business frequency using the service benefit information of the card product, Generating a second vector table including a weight value corresponding to each business type for each of a plurality of card merchandise;
Calculating a similarity value between the first vector table and the second vector table; And
And selecting a recommended card product based on the calculated similarity value. The method according to claim 1,
Wherein the first vector table generation step comprises:
And generating the first vector table including the weight value, which is calculated for each industry type by an operation of multiplying the CF weight value of the industry frequency and the IGF value, Recommended method. delete The method according to claim 1,
Wherein the first vector table or the second vector table comprises:
(CF weight) value of the industry frequency, which is calculated by weighting the industry frequency value by a logarithmic value. The method according to claim 1,
Wherein the first vector table or the second vector table comprises:
Further comprising a query length value calculated by summing squares of each of the weight values for a plurality of types of business and calculating a root (?) Operation on the sum. 6. The method of claim 5,
Wherein the first vector table or the second vector table comprises:
And the normalization value calculated for each type of business by associating the weight value with the query length value. The method according to claim 1,
The similarity value calculating step may include:
Calculating the similarity value by summing the calculated vector values through an operation of multiplying the normalization value included in the first vector table and the normalization value included in the second vector table for each type of business How to recommend card products. The method according to claim 1,
The similarity value calculating step may include:
Compares the first vector table with a second vector table corresponding to each of the plurality of card merchandise, calculates each similarity value,
The recommendation card product selection step includes:
And selects a card product for a second vector table corresponding to the highest similarity value among the calculated similarity values as a recommended card product. (CF) value, an inverse goods benefit frequency (IGF) value, a weight value, and a normalization (IGF) value of a category of business frequency (CF) A weight value of the business frequency, a weight value, and a normalization value of the business card using the service benefit information for each business type of card merchandise, And a weight value corresponding to a type of a business type, for each of a plurality of card merchandise items, for each of a plurality of card merchandise items, ;
A similarity calculating unit for calculating a similarity value between the first vector table and the second vector table; And
And a card product recommendation unit for selecting a recommended card product based on the calculated similarity value.

Images