WO2023136492A1

WO2023136492A1 - Method for determining e-commerce data-based credit assessment scorecard and device performing same method

Info

Publication number: WO2023136492A1
Application number: PCT/KR2022/020815
Authority: WO
Inventors: 강정석
Original assignee: 주식회사 에이젠글로벌
Priority date: 2022-01-12
Filing date: 2022-12-20
Publication date: 2023-07-20
Also published as: KR20230109070A; KR102433934B1

Abstract

The present invention relates to a method for determining an e-commerce data-based credit assessment scorecard and a device performing the method. The method for determining an e-commerce data-based credit assessment scorecard may include the steps in which: a primary candidate scorecard determination unit combines multiple pieces of credit assessment basic data to determine multiple primary candidate scorecards including multiple pieces of different credit assessment basic data; a primary candidate artificial intelligence engine generation unit generates primary candidate artificial intelligence engines trained on the basis of the multiple primary candidate scorecards; and a secondary candidate artificial intelligence engine generation unit determines an artificial intelligence engine having reliability greater than or equal to critical reliability among the primary candidate artificial intelligence engines as a secondary candidate artificial intelligence engine.

Description

Credit rating score card determination method based on e-commerce data and apparatus for performing the method

The present invention relates to a method for determining a credit evaluation score card based on e-commerce data and an apparatus for performing the method. More particularly, it relates to a method for determining a credit rating scorecard for credit rating of a seller based on e-commerce data and an apparatus for performing the method.

As we enter the intelligent information society triggered by the 4th industrial revolution, the possibility of unlimited utilization of data is bringing about changes in the data industry. With the arrival of the data era, the level of the future data industry will determine the difference in competitiveness between countries.

In particular, the establishment of big data infrastructure in the financial market is not only very urgent, but soon it has become an important asset enough to determine the direction of the country's data industry. The financial big data infrastructure consists of big data open systems, data exchanges, and data specialized institutions.

Research on new financial products based on these big data-based user financial data is needed. It is possible to analyze various risks through artificial intelligence-based learning of user financial products, and based on risk analysis, new financial services that have not been available before can be provided to users.

Therefore, it is necessary to study specific methods for utilizing user's financial data and providing various financial services based on the user's financial data.

The object of the present invention is to solve all of the above problems.

In addition, an object of the present invention is to provide a customized financial service to a seller by performing a more accurate credit evaluation of a seller through e-commerce data-based credit evaluation score card determination.

In addition, the present invention provides a customized financial service tailored to the seller by performing a step-by-step scorecard determination step in consideration of seller characteristic information and weight adjustment, scale adjustment, or creation time adjustment for a plurality of credit evaluation basic data included in the scorecard. intended to provide

Representative configurations of the present invention for achieving the above object are as follows.

According to an embodiment of the present invention, a method for determining a credit rating scorecard based on e-commerce data includes a plurality of first candidate scorecard decision units including a plurality of different credit rating base data by combining a plurality of credit rating base data. Determining a second candidate scorecard, a first candidate artificial intelligence engine generating unit generating a first candidate artificial intelligence engine learned based on the plurality of first candidate scorecards, a second candidate artificial intelligence engine generating unit Determining an artificial intelligence engine whose reliability is equal to or greater than a critical reliability level among the primary candidate artificial intelligence engines as a secondary candidate artificial intelligence engine, wherein a secondary candidate scorecard generator generates the primary candidate scorecard corresponding to the secondary candidate artificial intelligence engine. Generating a second candidate scorecard by performing weight adjustment, scale adjustment, or creation time adjustment for the plurality of credit evaluation basic data included in ; It may include the step of determining.

On the other hand, in the weight adjustment, considering the importance of each of the plurality of credit evaluation basic data included in the primary candidate scorecard, a relatively larger weight is assigned to the credit evaluation basic data having a relatively greater influence on the financial service result. setting to determine the secondary candidate scorecard, and the scale adjustment may determine the secondary candidate scorecard based on adjustment of a range scale for classifying the plurality of credit evaluation base data.

In addition, the generation time adjustment determines the secondary candidate scorecard in consideration of the generation time of the plurality of credit evaluation basic data, and the generation time adjustment includes the observation period, scoring time, and operation of the plurality of credit evaluation basic data. A plurality of scoring time points may be set by classifying into periods and setting the observation period and the operation period differently.

According to another embodiment of the present invention, a credit evaluation device for determining a credit evaluation scorecard based on e-commerce data combines a plurality of credit evaluation basic data and a plurality of primary candidate scorecards including a plurality of different credit evaluation basic data. A primary candidate scorecard determining unit implemented to determine a primary candidate scorecard, a primary candidate artificial intelligence engine generating unit implemented to generate a primary candidate artificial intelligence engine learned based on the plurality of primary candidate scorecards, and the primary candidate scorecard. A secondary candidate artificial intelligence engine generation unit configured to determine, among artificial intelligence engines, an artificial intelligence engine whose reliability is equal to or higher than a critical reliability level as a secondary candidate artificial intelligence engine; and the primary candidate scorecard corresponding to the secondary candidate artificial intelligence engine. A second candidate scorecard generation unit implemented to generate a second candidate scorecard by performing weight adjustment, scale adjustment, or creation time adjustment for the plurality of included credit rating base data, and a final one of the second candidate scorecard and a scorecard determination unit implemented to determine the scorecard to be used.

According to the present invention, a more accurate seller's credit evaluation can be performed through e-commerce data-based credit evaluation score card determination, so that a customized financial service can be provided to the seller.

In addition, according to the present invention, a customized financial service tailored to the seller is performed by performing a step-by-step scorecard determination step in consideration of seller characteristic information and weight adjustment, scale adjustment, or generation timing adjustment for a plurality of credit evaluation basic data included in the scorecard. may be provided.

1 is a conceptual diagram illustrating a credit evaluation apparatus according to an embodiment of the present invention.

2 is a conceptual diagram illustrating a sales management platform and a method of collecting basic credit evaluation data through the sales management platform according to an embodiment of the present invention.

3 is a conceptual diagram illustrating the operation of a credit basic data pre-processing unit according to an embodiment of the present invention.

4 is a conceptual diagram illustrating a first preprocessing according to an embodiment of the present invention.

5 is a conceptual diagram illustrating a first preprocessing (a seller) according to an embodiment of the present invention.

6 is a conceptual diagram illustrating a first preprocessing according to an embodiment of the present invention.

7 is a conceptual diagram illustrating a first preprocessing according to an embodiment of the present invention.

8 is a conceptual diagram illustrating a method of performing a first preprocessing according to an embodiment of the present invention.

9 is a conceptual diagram illustrating the operation of a credit evaluation unit according to an embodiment of the present invention.

10 is a conceptual diagram illustrating a scorecard determination method according to an embodiment of the present invention.

11 is a conceptual diagram illustrating a scorecard determination method according to an embodiment of the present invention.

12 is a conceptual diagram illustrating a method for determining a scorecard according to an embodiment of the present invention.

13 is a conceptual diagram illustrating a credit evaluation device that performs scorecard determination according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The detailed description of the present invention which follows refers to the accompanying drawings which illustrate, by way of illustration, specific embodiments in which the present invention may be practiced. These embodiments are described in sufficient detail to enable any person skilled in the art to practice the present invention. It should be understood that the various embodiments of the present invention are different from each other but are not necessarily mutually exclusive. For example, specific shapes, structures, and characteristics described herein may be implemented from one embodiment to another without departing from the spirit and scope of the present invention. It should also be understood that the location or arrangement of individual components within each embodiment may be changed without departing from the spirit and scope of the present invention. Therefore, the detailed description to be described later is not performed in a limiting sense, and the scope of the present invention should be taken as encompassing the scope claimed by the claims and all scopes equivalent thereto. Like reference numbers in the drawings indicate the same or similar elements throughout the various aspects.

Hereinafter, various preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings in order to enable those skilled in the art to easily practice the present invention.

In FIG. 1, a credit evaluation device for performing a credit evaluation on a seller based on data collected through an e-commerce sales and distribution management system platform is disclosed.

Referring to FIG. 1 , the credit evaluation apparatus includes a credit evaluation basic data collection unit 110, a credit evaluation basic data pre-processing unit 120, a credit evaluation basic data learning unit 130, a credit evaluation unit 140, and a financial service unit. 150 and a processor 160.

The credit evaluation basic data collection unit 110 may be implemented to collect credit evaluation basic data for a seller's credit evaluation. Sellers can sell products through various platforms for selling and distributing products through e-commerce. An e-commerce sales and distribution management system platform for managing various data related to seller's product sales, product distribution, and product payment may be expressed as a sales management platform 100 . The credit evaluation basic data collection unit 110 may be implemented to collect credit evaluation basic data for seller's credit evaluation through various sales management platforms. A specific sales management platform 100 will be described later. The term product used in the present invention may be used as a meaning including a service provided by a seller as one product.

The credit evaluation basic data preprocessing unit 120 may be implemented to preprocess the collected credit evaluation basic data. The basic credit evaluation data may be preprocessed and used for learning of an artificial intelligence engine for credit evaluation, or may be used for credit evaluation of a seller.

The credit evaluation basic data for learning of the artificial intelligence engine for credit evaluation may be transmitted to the credit evaluation basic data learning unit 130 through the first pre-processing. Credit evaluation basic data for learning of the artificial intelligence engine for credit evaluation of the seller may be transmitted to the credit evaluation unit 140 through the second pre-processing.

The credit evaluation basic data learning unit 130 may be implemented for artificial intelligence learning for seller's credit evaluation. The credit evaluation basic data learning unit 130 includes a plurality of artificial intelligence engines for credit evaluation of the seller, and each of the plurality of artificial intelligence engines may be implemented to determine a lower credit evaluation factor for the seller's credit evaluation. .

The credit evaluation unit 140 may be implemented to evaluate the seller's credit and determine the seller's credit evaluation data. The credit evaluation unit 140 may determine the seller's credit evaluation data based on a plurality of sub-level credit evaluation factors determined by each of a plurality of artificial intelligence engines of the credit evaluation basic data learning unit. In addition, the credit evaluation unit 140 may determine the seller's credit evaluation data based on a separate algorithm rather than an artificial intelligence engine.

The financial service unit 150 may be implemented to provide financial services to the seller based on the seller's credit evaluation data.

The processor 160 operates the credit evaluation basic data collection unit 110, the credit evaluation basic data pre-processing unit 120, the credit evaluation basic data learning unit 130, the credit evaluation unit 140, and the financial service unit 150. can be implemented to control

2 discloses a sales management platform and credit evaluation basic data collected from the sales management platform.

Referring to FIG. 2, the sales management platform includes an order management system (OMS) 210, an enterprise resource planning (ERP) 220, a warehouse management system (WMS) 230, and an E-commerce solution (ECS) 240. etc. may be included. OMS (210), ERP (220), WMS (230), ECS (240), as an example, may be a sales management platform for various other entities related to product sales of sellers.

OMS (210) is a platform for product order management of the seller.

The OMS 210 is a computer system through which a seller who sells a product through a plurality of sales channels can integrally process a series of sales processes. The seller may check the status of products ordered through the plurality of sales channels through the OMS 210 and may collectively process payment confirmation, delivery, order cancellation, return, and the like.

In detail, on the OMS 210, functions such as batch product registration modification, order collection, invoice registration and transmission, inventory management, and the like may be provided. In addition, the OMS 210 may provide functions for managing payment information, sales information, settlement information for sales, return information, refund information due to return, and inventory information on a plurality of sales channels.

The ERP 220 may be a sales management platform for managing information such as product production (purchase), logistics, finance, accounting, sales, purchase, and inventory of a seller as enterprise resource management.

WMS 230 is a warehouse management system and a sales management platform for supporting and optimizing warehouse or distribution center management. The WMS 230 may integrate and manage logistics processes such as warehousing, stocking, stocking, picking, and shipping of the seller's products as a whole.

The ECS 240 may be a sales management platform for creating and managing an online mall for sale by a seller. The ECS 240 may be implemented to create an online shopping mall, manage data generated on the online shopping mall, and perform marketing for product sales.

Credit evaluation basic data collection unit OMS (210), ERP (220), WMS (230), ECS (240), such as sales management platform in association with the credit evaluation basic data may be collected.

For example, the credit evaluation basic data collection unit converts product registration information, inventory information, order information, return information, payment information, sales information, settlement information, refund information, etc. generated in the OMS 210 into the seller's credit evaluation basic data. can be collected

The credit evaluation basic data collection unit may collect product storage information, product stock information, product release information, and product delivery information generated by the WMS 230 as the seller's credit evaluation basic data.

The credit evaluation basic data collection unit may collect the product marketing information generated by the ECS 240 as the seller's credit evaluation basic data.

3 discloses a method of preprocessing credit evaluation basic data in a credit basic data preprocessing unit.

Referring to FIG. 3 , the credit evaluation basic data 300 may be transmitted to the credit evaluation basic data learning unit 360 as the first preprocessing credit evaluation basic data 320 through the first preprocessing 310 . In addition, the credit evaluation basic data 300 may be transmitted to the credit evaluation unit 370 as the second preprocessing credit evaluation basic data 355 through the second preprocessing 350 .

The first pre-processing 310 may be pre-processing for learning in an artificial intelligence engine.

The first pre-processing 310 may be performed in consideration of characteristics of the sales management platform that generated the credit evaluation basic data 300 . In the present invention, since the financial service is provided considering seller characteristics and supply chain characteristics, the first pre-processing 310 may be performed to learn the artificial intelligence engine considering the seller characteristics and supply chain characteristics.

Among the first preprocessing 310 , preprocessing considering supply chain characteristics may be performed in consideration of a supply chain step corresponding to the credit evaluation basic data 300 . Among the first preprocessing 310 , preprocessing considering supply chain characteristics may be expressed as a first preprocessing (supply chain) 313 .

For example, when a supply chain is divided into a production (or purchase) stage, a distribution stage, and a sales stage, the credit evaluation basic data 300 is primarily credit evaluation basic data (production phase) based on the stage at which the data was acquired. ), credit evaluation basic data (distribution phase), and credit evaluation basic data (sale phase), and may be generated as first pre-processed credit evaluation basic data 320 .

Also, among the first preprocessing 310 , the preprocessing considering seller characteristics may be performed through seller data classification and seller data augmentation based on seller characteristics. Among the first pre-processing, the pre-processing considering seller characteristics may be expressed as a first pre-processing (seller) 316 .

The second pre-processing 350 may be performed for credit evaluation of the seller based on an artificial intelligence engine included in the credit evaluation unit. The second pre-processed credit evaluation base data 355 may be input to an artificial intelligence engine and used to determine lower credit evaluation factors. Accordingly, the second pre-processing 350 may be performed in consideration of the input data format of the artificial intelligence engine. Prediction on different credit rating base data is performed for each artificial intelligence engine, and each artificial intelligence engine may have a different data format. A second pre-processing 350 may be performed in consideration of at least one artificial intelligence engine that may be used for credit evaluation of the seller.

In FIG. 4 , a first pre-processing (seller) and a first pre-processing (supply chain) applied to credit evaluation basic data are disclosed.

Referring to FIG. 4 , credit evaluation basic data classified by seller through the first pre-processing (seller) 400 is pre-processed for each supply chain step through the first pre-processing (supply chain) 450, and the first pre-processed basic credit data (490).

The first pre-processing (supply chain) 450 may generate the first pre-processing credit basic data 490 in consideration of the sales management platform, which is the subject of data transmission, and the data format transmitted from the sales management platform.

Through the first pre-processing (supply chain) 450 that considers the information included in the supply chain stage managed by the sales management platform and the credit rating basic data generated in the sales management platform, the credit rating basic data is converted into credit rating basic data (produced). ) 460, credit evaluation basic data (distribution) 470, and credit evaluation basic data (sales) 480, and may be generated as first pre-processed credit evaluation basic data.

In addition, the first pre-processing (supply chain) 450 may generate first pre-processing credit evaluation basic data 490 by performing redundant data processing on the credit evaluation basic data transmitted through the sales management platform. When credit rating base data for the same product is duplicated in a plurality of sales management platforms, duplicate data processing may be performed. For example, when a seller purchases a specific product for sale, the product may be registered on the OMS and the product may be placed on the WMS. That is, an act of purchasing a specific product by a seller is performed only once, but data on product registration and product placement resulting from such a purchase action are generated for each sales management platform, which may result in duplication of basic credit evaluation data.

The first pre-processing (supply chain) 450 determines the redundancy of the transmitted credit rating basic data in consideration of the data generation time of the credit rating basic data, the information included in the credit rating basic data, and the later transmitted credit evaluation basic data information. Thus, first pre-processing credit evaluation basic data 490 may be generated. When duplication of credit evaluation basic data occurs, only data of one sales management platform is used through the first preprocessing (supply chain) 450, or duplicated credit evaluation basic data is filtered out and duplicate credit Only other credit evaluation basic data including information included in the evaluation basic data may be used.

Also, the first pre-processing (supply chain) 450 may be pre-processing of credit evaluation basic data considering time.

The seller's credit rating and the data underlying the seller's credit rating may change over time. Therefore, setting a time scale for credit evaluation basic data for learning may greatly affect the performance of the artificial intelligence engine. Accordingly, in the present invention, after setting a time scale for the obtained basic credit evaluation data, the first preprocessed basic credit data 490 may be generated by pre-processing the basic credit evaluation data considering the time scale. A time scale for preprocessing may be set for each credit evaluation basic data.

In FIG. 5, a method for augmenting processing of credit evaluation base data for learning of an artificial intelligence engine through a first preprocessing (seller) is disclosed. In particular, a method in which basic credit evaluation data is divided into sub-level credit evaluation basic data and used as learning data among data augmentation methods is disclosed.

Referring to FIG. 5 , a method of performing more accurate artificial intelligence engine learning by augmenting specific credit evaluation base data 500 is disclosed. For example, the credit evaluation basic data 500 may include seasonality, transaction size, delivery cycle, sales trend, return rate, sales product, inventory size, operational information, and the like. Among the credit evaluation basic data 500 , specific credit evaluation basic data may be augmented and generated as a plurality of lower credit evaluation basic data 540 .

When the return rate data is increased, data such as the size of the return, the discard rate, the average return rate, the stability of the change in the return rate, and the number of times the return rate exceeds the MAX may be generated as the lower credit evaluation basic data 540 .

In an embodiment of the present invention, when data augmentation is required to make credit evaluation more accurate, learning may be performed by augmenting the credit evaluation basic data 500 with lower credit evaluation basic data 540 through a first preprocessing. Data augmentation of the first pre-processing (seller) may be expressed in terms of a lower data augmentation 520 .

In FIG. 6, a method of augmenting processing of credit evaluation basic data for learning of an artificial intelligence engine on a first preprocessing (seller) is disclosed. In particular, among data augmentation methods, a method of augmenting data by analyzing data based on credit evaluation on a time scale and a method of augmenting data through a statistical method are disclosed.

6(a) is a method of augmenting data by analyzing the credit evaluation base data 600 on a time scale among data augmentation methods. Among data augmentation methods, a method of augmenting data by analyzing credit evaluation base data on a time scale may be expressed as a time scale data augmentation 610 .

For example, when the credit evaluation basic data 600 is a return rate, data on the number of sellers with a monthly return rate of 5% or more for 36 months may be augmented and generated based on the monthly return rate of 5% or more. As another example, based on the average monthly return rate, data on the number of sellers with an average return rate for 36 months may be augmented and generated.

(b) of FIG. 6 is a method of augmenting data by statistically analyzing the credit evaluation base data 650 among data augmentation methods. Among the data augmentation methods, a method of statistically analyzing the credit evaluation base data 650 to augment data may be expressed as a statistical data augmentation 660 . For example, if the credit evaluation basic data 650 is the return rate, data augmentation may be performed by increasing the return rate in various ways through a statistical method such as average, standard deviation, maximum, or more than a specific range of return rates for each customer.

In FIG. 7 , a method of augmenting processing of credit evaluation basic data for learning of an artificial intelligence engine on a first preprocessing (seller) is disclosed. In particular, a method of augmenting data by two-dimensionally analyzing credit evaluation basic data among data augmentation methods is disclosed.

Referring to FIG. 7 , a method of augmenting credit evaluation basic data 700 as two-dimensional data among data augmentation methods is disclosed. Among data augmentation methods, credit evaluation basic data 700 is divided into a plurality of dimensions to augment data. The method may be expressed in terms of multidimensional data augmentation 710 .

For example, when the credit evaluation basic data 700 is a return rate, the entire return rate may be divided into two-dimensional data and increased. The first dimension may be the average return rate for 36 months, and the second dimension may be the number of months in which the return rate exceeds 5% for 36 months. Through this two-dimensional analysis, a more accurate assessment of the seller's credit evaluation basic data 700 may be possible.

8 discloses a method of selectively performing a first preprocessing (seller) for learning of a credit evaluation basic data learning unit.

Referring to FIG. 8 , the first preprocessing (seller) may use lower data augmentation 810, time scale data augmentation 820, statistical data augmentation 830, and multidimensional data augmentation 840.

A first pre-processing (seller) may be selectively performed to learn the plurality of artificial intelligence engines included in the credit evaluation basic data learning unit.

For example, the sub-data augmentation 810 may be used for training of an artificial intelligence engine to generate a specialized result through specific analysis of characteristic credit evaluation data among artificial intelligence engines. For example, sub-data augmentation for the return rate may be performed for learning of an artificial intelligence engine that generates credit evaluation data with more weight on the return rate.

Time scale data augmentation 820 can be used to train an artificial intelligence engine to predict changes in credit rating data over time.

Statistical data augmentation 830 may be used for training of an artificial intelligence engine for predicting credit evaluation data according to separately preset specific criteria.

Multi-dimensional data augmentation 840 can be used to train an artificial intelligence engine to predict credit rating data based on set criteria for two dimensions.

In an embodiment of the present invention, various first preprocessing (seller) may be performed according to the properties of predicted credit evaluation data, and various artificial intelligence models may be generated.

9 discloses a method of generating credit evaluation data of a seller based on an artificial intelligence engine in a credit evaluation unit.

Referring to FIG. 9 , the credit evaluation unit may generate credit evaluation data through a seller's credit evaluation based on at least one artificial intelligence engine.

The credit evaluation unit may generate the seller's credit evaluation data based on one artificial intelligence engine, but the credit evaluation unit adaptively determines the artificial intelligence engine applicable to the seller based on the seller characteristic information 900, and the determined artificial intelligence engine. Credit evaluation data 950 may be generated based on an intelligence engine.

For example, the target artificial intelligence engine 920 for credit evaluation most suitable for the seller characteristic information 900 based on seller information such as seller's sales product, seller's product sales platform, seller's sales, seller's net profit, etc. can be determined

The credit evaluation unit may determine the reliability of the artificial intelligence engine according to the seller characteristic information based on the feedback information of each of the plurality of artificial intelligence engines. In addition, the credit evaluation unit may determine a reliability level for each seller characteristic information for each of the plurality of artificial intelligence engines. Specifically, seller characteristic information may be vectorized and expressed on a space based on each sub-seller characteristic information, seller groups may be formed based on seller characteristic information through distance information between spaces, and credit evaluation data for each seller group The reliability of each seller group of the artificial intelligence engine may be determined by comparing the result data of the financial service and the financial service result data. The reliability level may be determined in consideration of statistical characteristics of reliability for seller groups for each artificial intelligence engine.

The credit evaluation unit may determine an artificial intelligence engine having a relatively high reliability level as the target artificial intelligence engine 920 based on the seller characteristic information and generate credit evaluation data 950 for the seller.

10 discloses a method for determining a scorecard for determining a seller's credit rating data.

Referring to FIG. 10 , the scorecard 1000 may be a card including a plurality of basic credit evaluation data for determining credit evaluation data of a seller. A plurality of credit evaluation base data included in the scorecard 1000 may be preprocessed and then input to the artificial intelligence engine 1020 to determine credit evaluation data.

For example, the plurality of credit evaluation basic data included in the scorecard 1000 is information such as residence status, residence period, occupation, job retention period, bank records, card use records, existing loan records, and the like. can

The scorecard 1000 includes a combination of a plurality of basic credit evaluation data, and can be generated in various types according to the combination of a plurality of basic credit evaluation data. A combination of a plurality of credit rating base data included in the scorecard 1000 may be pre-processed and input to the artificial intelligence engine 1020 for determining the seller's credit rating data.

In the embodiment of the present invention, not only can there be a plurality of scorecards 1000 composed of combinations of a plurality of various credit evaluation base data, but also the scorecard 1000 including the same credit evaluation base data includes learning and In order to generate credit evaluation data, it may be defined as a different scorecard 1000 according to a weight applied to each of a plurality of credit evaluation basic data, a scale applied to each of a plurality of credit evaluation basic data, and the like, and input to the artificial intelligence engine 1020. there is.

That is, in the present invention, a plurality of scorecards 1000 may be input to and learned from different artificial intelligence engines 1020, and thus different credit evaluation data 1040 may be generated. Among the plurality of scorecards 1000, a specific scorecard 1000 may be selectively used. In the present invention, various scorecard determination methods may be used to determine the scorecard 1000 with the highest accuracy.

Referring to FIG. 11, a plurality of primary candidate scorecards including a plurality of different credit evaluation basic data are variously combined with a plurality of credit evaluation basic data to determine a scorecard that generates highly accurate credit evaluation data. (1100) can be determined.

After the plurality of primary candidate scorecards 1100 are determined, credit evaluation base data corresponding to the plurality of primary candidate scorecards 1100 is input to each of a plurality of artificial intelligence engines, and each of the plurality of artificial intelligence engines learns. this can be done

A plurality of artificial intelligence engines that are trained based on credit evaluation basic data corresponding to the plurality of primary candidate scorecards 1100 may be expressed as a term of primary candidate artificial intelligence engines 1120 .

If the reliability of the primary candidate artificial intelligence engine 1120 is greater than or equal to the threshold reliability, the corresponding artificial intelligence engine may be determined as the secondary candidate artificial intelligence engine 1160 .

A plurality of credit evaluation basic data included in the primary candidate scorecard 1100 corresponding to the secondary candidate artificial intelligence engine 1160 having a threshold reliability or higher is weighted, scaled, or generated by adjusting the secondary candidate scorecard 1100. Scorecard 1140 may include.

Credit evaluation base data corresponding to the secondary candidate scorecard 1140 is input to the secondary candidate artificial intelligence engine 1160, and the secondary candidate artificial intelligence engine 1160 may be trained.

Among the secondary candidate artificial intelligence engines 1160 , at least one artificial intelligence engine whose reliability is equal to or higher than a critical reliability level or an artificial intelligence engine having the highest reliability may be finally determined as an artificial intelligence engine to be used in the credit evaluation unit. In addition, the secondary candidate scorecard 1140 used in the finally determined artificial intelligence engine may be determined as the final scorecard to be used.

The above scorecard and artificial intelligence engine determining operation may be performed for each seller group in consideration of seller characteristic information, an artificial intelligence engine may be determined for each seller group, and a scorecard to be used for each seller group may be determined. That is, for each seller group, the first candidate scorecard 1000, the first candidate artificial intelligence engine 1120, the second candidate scorecard 1140, and the second candidate artificial intelligence engine 1160 may be determined.

12 discloses a method of determining a secondary candidate scorecard by performing weight adjustment, scale adjustment, or generation timing adjustment.

Referring to FIG. 12 , the weight adjustment 1200 may be set in consideration of the importance of each of the basic credit evaluation data included in the scorecard. For credit evaluation basic data that has a greater impact on actual financial service results, a larger weight can be set to perform learning on the artificial intelligence engine.

Scale adjustment 1210 may be an adjustment of a range scale for classifying credit rating base data. For example, in the case of the number of years of service at work, it can be classified into n categories, and the scale on which the basic credit evaluation data is classified and learned determines whether the credit evaluation data of the artificial intelligence engine will reflect the actual financial service results. can affect what you can do. Accordingly, optimal artificial intelligence engine learning may be performed through scaling 1210 for each of a plurality of credit evaluation base data included in the secondary candidate scorecard 1240 .

The creation time point adjustment 1220 (or scoring time point adjustment) may perform learning of the artificial intelligence engine in consideration of the generation time point (or scoring time point) of the credit evaluation base data. A plurality of credit evaluation basic data included in the scorecard may be grouped and input to the artificial intelligence engine. Accordingly, it is possible to determine which learning data is generated by setting a generation time point of the credit evaluation basic data. Therefore, more accurate learning of the artificial intelligence engine can be performed through the adjustment of the generation time point 1220 of the credit evaluation base data.

In addition, the creation time point adjustment 1220 may reduce a score error generated according to the generation time point by setting a plurality of creation time points. In the basic credit evaluation data, the observation period 1250, the scoring time 1260, and the operation period 1270 are separately classified, and the observation period 1250 and the operation period 1270 are set differently to set a plurality of scoring times 1260. can be set By setting a plurality of scoring time points 1260, it is possible to reduce a score error according to the time of creation, which may occur according to a seller's product, such as a seasonal change, and to reflect the score change according to time.

Referring to FIG. 13 , the credit evaluation device may include a scorecard determining unit.

A scorecard determination unit may be implemented to determine a scorecard. The scorecard determining unit includes a first candidate scorecard determining unit 1310, a first candidate artificial intelligence engine generating unit 1320, a second candidate scorecard determining unit 1330, a second candidate artificial intelligence engine generating unit 1340, and A scorecard determining unit 1350 may be included.

The primary candidate scorecard determining unit 1310 variously combines a plurality of credit evaluation basic data to determine a scorecard for determining highly accurate credit evaluation data, and includes a plurality of different credit evaluation basic data. A primary candidate scorecard can be determined.

The first candidate artificial intelligence engine generation unit 1320 may be implemented to generate the first candidate artificial intelligence engine. After determining the plurality of primary candidate scorecards, the credit evaluation basic data corresponding to the plurality of primary candidate scorecards is input to each of the plurality of artificial intelligence engines, learning of each of the plurality of artificial intelligence engines is performed, and the primary A candidate artificial intelligence engine may be determined.

The secondary candidate artificial intelligence engine generating unit 1340 may determine the corresponding artificial intelligence engine as a secondary candidate artificial intelligence engine when reliability among the primary candidate artificial intelligence engines is greater than or equal to a critical reliability level.

The secondary candidate scorecard generating unit 1330 performs weight adjustment, scale adjustment, or creation time adjustment on a plurality of credit evaluation base data included in the primary candidate scorecard corresponding to the secondary candidate artificial intelligence engine to generate secondary candidate scorecards. A candidate scorecard can be determined.

The scorecard determination unit 1350 may be implemented to determine a final scorecard to be used among secondary candidate scorecards.

Embodiments according to the present invention described above may be implemented in the form of program instructions that can be executed through various computer components and recorded on a computer-readable recording medium. The computer readable recording medium may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on the computer-readable recording medium may be specially designed and configured for the present invention, or may be known and usable to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and magneto-optical media such as floptical disks. medium), and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include high-level language codes that can be executed by a computer using an interpreter or the like as well as machine language codes generated by a compiler. A hardware device may be modified with one or more software modules to perform processing according to the present invention and vice versa.

Although the present invention has been described above with specific details such as specific components and limited embodiments and drawings, these are only provided to help a more general understanding of the present invention, and the present invention is not limited to the above embodiments, and the present invention Those with ordinary knowledge in the technical field to which the invention belongs may seek various modifications and changes from these descriptions.

Therefore, the spirit of the present invention should not be limited to the above-described embodiments and should not be determined, and all scopes equivalent to or equivalently changed from the claims as well as the claims described below are within the scope of the spirit of the present invention. will be said to belong to

Claims

The e-commerce data-based credit rating scorecard determination method,

determining a plurality of primary candidate scorecards including a plurality of different credit evaluation base data by combining a plurality of credit evaluation base data, by a primary candidate scorecard determination unit;

generating a first candidate artificial intelligence engine learned by a first candidate artificial intelligence engine generation unit based on the plurality of first candidate scorecards;

determining, by a secondary candidate artificial intelligence engine generation unit, an artificial intelligence engine having a reliability of which is greater than or equal to a critical reliability among the primary candidate artificial intelligence engines as a secondary candidate artificial intelligence engine;

A second candidate scorecard generation unit performs weight adjustment, scale adjustment, or creation time adjustment on the plurality of credit evaluation base data included in the first candidate scorecard corresponding to the second candidate artificial intelligence engine to generate a second candidate scorecard. generating a scorecard; and

and a scorecard determination unit determining a scorecard to be finally used among the secondary candidate scorecards.
According to claim 1,

The weight adjustment takes into account the importance of each of the plurality of credit evaluation basic data included in the primary candidate scorecard, and sets a relatively larger weight for credit evaluation basic data that has a relatively greater impact on the financial service result determining the secondary candidate scorecard;

Wherein the scale adjustment determines the secondary candidate scorecard based on the adjustment of a range scale for classifying the plurality of credit rating base data.
According to claim 1,

The generation time adjustment determines the secondary candidate scorecard in consideration of the generation time of the plurality of credit evaluation base data;

The creation time point adjustment comprises classifying the plurality of credit evaluation basic data into an observation period, a scoring time point, and an operation period, and setting the plurality of scoring time points by setting the observation period and the operation period differently.
The credit evaluation device that determines the e-commerce data-based credit evaluation score card,

a primary candidate scorecard determining unit, configured to combine a plurality of credit rating base data to determine a plurality of primary candidate scorecards including a plurality of different credit rating base data;

a first candidate artificial intelligence engine generation unit configured to generate a first candidate artificial intelligence engine learned based on the plurality of first candidate scorecards;

a second candidate artificial intelligence engine generating unit configured to determine an artificial intelligence engine whose reliability is equal to or higher than a critical reliability level among the first candidate artificial intelligence engines as a secondary candidate artificial intelligence engine;

Implemented to generate a secondary candidate scorecard by performing weight adjustment, scale adjustment, or creation time adjustment for the plurality of credit evaluation base data included in the primary candidate scorecard corresponding to the secondary candidate artificial intelligence engine. a secondary candidate scorecard generator; and

and a scorecard determination unit configured to determine a scorecard to be finally used among the secondary candidate scorecards.
According to claim 4,

The weight adjustment takes into account the importance of each of the plurality of credit evaluation basic data included in the primary candidate scorecard, and sets a relatively larger weight for credit evaluation basic data that has a relatively greater impact on the financial service result determining the secondary candidate scorecard;

The scale adjustment determines the secondary candidate scorecard based on the adjustment of a range scale for classifying the plurality of credit evaluation base data.
According to claim 4,

The generation time adjustment determines the secondary candidate scorecard in consideration of the generation time of the plurality of credit evaluation base data;

The creation time point adjustment comprises classifying the plurality of credit evaluation base data into an observation period, a scoring time point, and an operation period, and setting the observation period and the operation period differently to set a plurality of scoring time points. Device.