KR102588210B1 - Apparatus and method for alternatively rating credit through the application of statistical tests to open data - Google Patents

Abstract

An apparatus and method for providing corporate credit evaluation information are disclosed. An apparatus for providing corporate credit evaluation information according to an embodiment collects open data including a plurality of items related to a company, extracts log data according to time from the open data, and generates log derived data, Extract valid data by verifying the significance of the log-derived data, calculate the probability of occurrence of a potential risk of the company based on the log-derived data corresponding to the valid data, and calculate the probability of occurrence of the potential risk of the company based on the probability of occurrence of the potential risk. It may include a processor that generates a credit evaluation result by determining an evaluation level.

Description

Alternative credit rating method and device through application of statistical tests to open data {APPARATUS AND METHOD FOR ALTERNATIVELY RATING CREDIT THROUGH THE APPLICATION OF STATISTICAL TESTS TO OPEN DATA}

Embodiments relate to alternative credit assessment methods and devices through the application of statistical tests to open data.

Credit rating information may refer to information that evaluates the creditworthiness of the counterparty in commercial transactions such as financial transactions, and financial institutions use this information in the process of providing credit-based credit services (e.g., loans, credit card issuance, etc.). The borrower's credit rating information is used to evaluate the possibility of damage that may be caused by the company's insolvency, and this can be used to determine whether to approve a loan, the interest rate limit, etc.

In the case of models that generate existing credit rating information, financial data is mainly used to generate credit rating information, but in the case of companies where the company's financial data is absent or insignificant, there may be a problem of low reliability of credit rating information. there is.

Accordingly, there is a need to improve the reliability of credit rating information by predicting credit risk using data other than the borrower's financial data.

An apparatus for providing corporate credit evaluation information according to an embodiment collects open data including a plurality of items related to a company, generates log derived data by extracting log data according to time of the open data, and , extract valid data by verifying the significance of the log-derived data, calculate the probability of occurrence of a potential risk of the company based on the log-derived data corresponding to the valid data, and calculate the probability of occurrence of a potential risk of the company based on the probability of occurrence of a potential risk of the company. It may include a processor that generates a credit evaluation result by determining a credit rating.

According to one embodiment, for each of the log-derived data, the processor obtains a first risk occurrence probability of a company in which an item corresponding to the log-derived data has been generated, and a second risk occurrence probability of a company in which the item has not been generated. The valid data can be determined by obtaining the risk occurrence probability and applying a statistical test method to the difference between the first risk occurrence probability and the second risk occurrence probability.

According to one embodiment, the processor performs a statistical test on the average value of the log-derived data for a group of companies in which a risk occurred and the average value of the log-derived data for a group of companies in which a risk did not occur, for each of the log-derived data. The method can be applied to determine the valid data.

According to one embodiment, the processor calculates the average risk occurrence probability corresponding to the log-derived data based on the distribution of the average risk occurrence probability according to the log-derived data, and calculates the average risk occurrence probability corresponding to the log-derived data and calculates the average risk occurrence probability based on the weight for the average default rate. The potential default rate can be calculated.

According to one embodiment, the processor generates first log derived data indicating the recency of the item by obtaining the last occurrence time of the item corresponding to the log data from the log data, and the log data Generate second log derived data indicating the frequency of occurrence of the item by obtaining all dates on which the log data occurred from the log data, and determine the magnitude of occurrence of the item by obtaining the number of occurrences of the log data from the log data. Third log derived data representing can be generated.

According to one embodiment, the processor, based on the distribution of the average risk occurrence probability according to the log derived data, provides a first average risk occurrence probability corresponding to the first log derived data, and a first average risk occurrence probability corresponding to the second log derived data. Calculating a second average risk occurrence probability and a third average risk occurrence probability corresponding to the third log derived data, the first average risk occurrence probability, the second average risk occurrence probability, and the third average risk occurrence probability, The potential risk occurrence probability is determined based on a first weight corresponding to the first average risk occurrence probability, a second weight corresponding to the second average risk occurrence probability, and a third weight corresponding to the third average risk occurrence probability. It can be calculated.

According to one embodiment, the processor inputs the first log-derived data, the second log-derived data, and the third log-derived data into a machine learning model to determine the first weight, the second weight, and the third weight. Weights can be determined.

A method of providing corporate credit evaluation information performed by an apparatus for providing corporate credit evaluation information according to an embodiment includes the steps of collecting open data related to a corporation; Generating log derived data by extracting log data according to time of the open data; Extracting valid data by verifying significance of the log derived data; calculating a probability of occurrence of a potential risk of the company based on the log-derived data; It may include generating a credit evaluation result by determining the credit rating of the company based on the probability of occurrence of the potential risk.

Figure 1 shows a schematic block diagram of an apparatus for providing credit evaluation information according to an embodiment.
FIG. 2 shows an example of implementation of the credit rating information providing device shown in FIG. 1.
FIG. 3 is a diagram for explaining the pulse algorithm for providing credit evaluation information shown in FIG. 1.
FIG. 4 is a diagram for explaining the operation of the sample inlet shown in FIG. 2.
FIG. 5 is a diagram for explaining the operation of the data collector shown in FIG. 2.
FIG. 6 is a diagram for explaining the operation of the log extractor shown in FIG. 2.
FIG. 7 is a diagram for explaining the operation of the significance inspector shown in FIG. 2.
FIG. 8 is a diagram for explaining the operation of the score calculator shown in FIG. 2.
FIG. 9 is a diagram for explaining the operation of the grade designer shown in FIG. 2.
FIG. 10 is a diagram for explaining the risk occurrence probability calculation operation of the performance evaluator shown in FIG. 2.
FIG. 11 shows an example of performance evaluation results of the performance evaluator shown in FIG. 2.
FIG. 12 shows a flowchart of the operation of the credit rating information providing device shown in FIG. 1.

Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be changed and implemented in various forms. Accordingly, the actual implementation form is not limited to the specific disclosed embodiments, and the scope of the present specification includes changes, equivalents, or substitutes included in the technical idea described in the embodiments.

Terms such as first or second may be used to describe various components, but these terms should be interpreted only for the purpose of distinguishing one component from another component. For example, a first component may be named a second component, and similarly, the second component may also be named a first component.

When a component is referred to as being “connected” to another component, it should be understood that it may be directly connected or connected to the other component, but that other components may exist in between.

Singular expressions include plural expressions unless the context clearly dictates otherwise. As used herein, “A or B”, “at least one of A and B”, “at least one of A or B”, “A, B or C”, “at least one of A, B and C”, and “A Each of phrases such as “at least one of , B, or C” may include any one of the items listed together in the corresponding phrase, or any possible combination thereof. In this specification, terms such as “comprise” or “have” are intended to designate the presence of the described features, numbers, steps, operations, components, parts, or combinations thereof, and are intended to indicate the presence of one or more other features or numbers, It should be understood that this does not exclude in advance the possibility of the presence or addition of steps, operations, components, parts, or combinations thereof.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person of ordinary skill in the art. Terms as defined in commonly used dictionaries should be interpreted as having meanings consistent with the meanings they have in the context of the related technology, and unless clearly defined in this specification, should not be interpreted in an idealized or overly formal sense. No.

The term “module” used in this document may include a unit implemented in hardware, software, or firmware, and may be used interchangeably with terms such as logic, logic block, component, or circuit, for example. A module may be an integrated part or a minimum unit of the parts or a part thereof that performs one or more functions. For example, according to one embodiment, the module may be implemented in the form of an application-specific integrated circuit (ASIC).

The term '~unit' used in this document refers to software or hardware components such as FPGA or ASIC, and '~unit' performs certain roles. However, '~part' is not limited to software or hardware. The '~ part' may be configured to reside in an addressable storage medium and may be configured to reproduce on one or more processors. For example, '~part' refers to software components, object-oriented software components, components such as class components and task components, processes, functions, properties, procedures, May include subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays, and variables. The functions provided within the components and 'parts' may be combined into a smaller number of components and 'parts' or may be further separated into additional components and 'parts'. Additionally, components and 'parts' may be implemented to regenerate one or more CPUs within a device or a secure multimedia card. Additionally, '~ part' may include one or more processors.

Hereinafter, embodiments will be described in detail with reference to the attached drawings. In the description with reference to the accompanying drawings, identical components will be assigned the same reference numerals regardless of the reference numerals, and overlapping descriptions thereof will be omitted.

Figure 1 shows a schematic block diagram of an apparatus for providing credit evaluation information according to an embodiment.

Referring to FIG. 1, the credit rating information providing device 10 can generate and provide credit rating information for a company. A company's credit evaluation information relates to information that evaluates and quantifies the company's credit risk, and credit risk may include arbitrary risks of the company that may arise during financial transactions, such as bankruptcy, delinquency, and business closure.

The credit evaluation information providing device according to one embodiment may generate and provide credit evaluation information mainly for corporations, but the subject of application of the present invention is not limited to corporations, and can be expanded to various objects. Technicians will understand.

The credit evaluation information providing device 10 generates and provides credit evaluation information about a company's credit risk, thereby providing a means for managing the company's credit risk in financial transactions. The credit evaluation information providing device 10 may generate and provide credit evaluation information based on the company's loan-related information and open data.

Open data is a concept that includes public data, which is information that is open to the public, such as a company's business details, a company's financial information, a company's employment information, investment information about the company, information about the technology possessed by the company, and the company's business. It may contain arbitrary information such as authentication information and/or company promotional material.

The credit evaluation information providing device 10 may process the company's loan-related information and open data to provide the company's credit evaluation information. In this process, the credit evaluation information providing device 10 can calculate the probability of occurrence of a company's risk based on the company's open data. The risk occurrence probability may refer to the probability of the company's credit risk occurring as explained above, and the default rate may be used as an example, but this is not limited to this and is related to arbitrary credit risks such as delinquency rate and business closure rate. It will be understood by those skilled in the art that elements may be utilized.

The corporate risk management device 10 can classify information about the probability of occurrence of corporate risks using an artificial intelligence algorithm or machine learning model. Artificial intelligence can refer to computer systems with capabilities such as learning, reasoning, or judgment. For example, an artificial intelligence algorithm may be implemented using a neural network, but this is only an example. The artificial function algorithm is not limited to this, and may be implemented with any statistical-based algorithm. will be understood by those skilled in the art.

Neural networks (or artificial neural networks) can include statistical learning algorithms that mimic neurons in biology in machine learning and cognitive science. A neural network can refer to an overall model in which artificial neurons (nodes), which form a network through the combination of synapses, change the strength of the synapse connection through learning and have problem-solving capabilities.

Neurons in a neural network can contain combinations of weights or biases. A neural network may include one or more layers consisting of one or more neurons or nodes. Neural networks can infer the results they want to predict from arbitrary inputs by changing the weights of neurons through learning.

Neural networks may include deep neural networks. Neural networks include CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), perceptron, multilayer perceptron, FF (Feed Forward), RBF (Radial Basis Network), DFF (Deep Feed Forward), and LSTM. (Long Short Term Memory), GRU (Gated Recurrent Unit), AE (Auto Encoder), VAE (Variational Auto Encoder), DAE (Denoising Auto Encoder), SAE (Sparse Auto Encoder), MC (Markov Chain), HN (Hopfield) Network), BM (Boltzmann Machine), RBM (Restricted Boltzmann Machine), DBN (Depp Belief Network), DCN (Deep Convolutional Network), DN (Deconvolutional Network), DCIGN (Deep Convolutional Inverse Graphics Network), GAN (Generative Adversarial Network) ), Liquid State Machine (LSM), Extreme Learning Machine (ELM), Echo State Network (ESN), Deep Residual Network (DRN), Differential Neural Computer (DNC), Neural Turning Machine (NTM), Capsule Network (CN), It may include a Kohonen Network (KN), a transformer, and an Attention Network (AN).

The credit rating information providing device 10 may be implemented with a printed circuit board (PCB) such as a motherboard, an integrated circuit (IC), or a system on chip (SoC). . The enterprise risk management device 10 may be implemented with an application processor.

Additionally, the credit evaluation information providing device 10 may be implemented in a personal computer (PC), a data server, or a portable device.

Portable devices include laptop computers, mobile phones, smart phones, tablet PCs, mobile internet devices (MIDs), personal digital assistants (PDAs), and enterprise digital assistants (EDAs). , digital still camera, digital video camera, portable multimedia player (PMP), personal navigation device or portable navigation device (PND), handheld game console, e-book ( It can be implemented as an e-book) or a smart device. A smart device may be implemented as a smart watch, smart band, or smart ring.

The credit evaluation information providing device 10 includes a processor 100 and a memory 200.

The processor 100 may process data stored in the memory 200. The processor 100 may execute computer-readable code (eg, software) stored in the memory 200 and instructions triggered by the processor 100 .

The processor 100 may be a data processing device implemented in hardware that has a circuit with a physical structure for executing desired operations. For example, the intended operations may include code or instructions included in the program.

For example, data processing devices implemented in hardware include microprocessors, central processing units, processor cores, multi-core processors, and multiprocessors. , ASIC (Application-Specific Integrated Circuit), and FPGA (Field Programmable Gate Array).

Processor 100 may include a receiving interface. Processor 100 may receive data from external sources or memory 200.

The processor 100 can collect open data of a company.

Illustratively, the processor 100 may include, among open data, the company's business details, the company's financial information, the company's employment information, the company's investment information, information about the technology possessed by the company, the company's corporate authentication information, and/or the company's information. Promotional materials can be collected.

Depending on the embodiment, the processor 100 may select some of the items included in the open data and proceed with the preceding log data extraction procedure. For example, the processor 100 may select certain items based on whether they are easy to process, whether the attributes are suitable for a technology verification method, and whether they meet the purpose of credit evaluation and perform the following procedures. However, those skilled in the art will understand that the specific method of selecting some items from open data is not limited to the presented embodiment.

The processor 100 may generate log derived data by extracting log data according to time from open data. Hereinafter, the log-derived data is exemplified as predicting the probability of occurrence of a company's potential risk based on three types of log-derived data including first log-derived data, second log-derived data, and/or third log-derived data. However, those skilled in the art will understand that the scope of rights of the embodiments of the present invention is not limited to the number and examples of log derived data presented later. Log data extracted from open data can be standardized in the form of (presence or absence of occurrence, time of occurrence, scale of occurrence (total)) and later used to generate log derived data, and the standardization format is not limited to the examples presented. Technicians will understand.

The processor 100 may extract valid data by verifying the significance of log-derived data.

The processor 100 may obtain, for each piece of log-derived data, the first risk occurrence probability of the company in which the item corresponding to the log-derived data occurred. The processor 100 may obtain the probability of occurrence of a second risk for a company in which an item has not occurred. The processor 100 may determine valid data by applying a statistical test method to the difference between the probability of occurrence of the first risk and the probability of occurrence of the second risk. For example, the processor 100 applies a chi-squared test to the difference between the probability of occurrence of a first risk and the probability of occurrence of a second risk and converts log-derived data corresponding to items in which the difference is significant into valid data. can be decided.

According to another embodiment, the processor 100 may determine whether the difference between the probability of occurrence of the first risk and the probability of occurrence of the second risk is greater than a predetermined threshold. The processor 100 may determine valid data based on the result of determining whether the data is greater than the threshold.

According to another embodiment, the processor 100 uses a statistical test method (for example, For example, valid data can be determined by applying the T-test. More specifically, the processor 100, for each log derived data, if there is a statistically significant difference between the average value of companies that have defaulted and the average value that has not defaulted (for example, the t statistic is 1.64 or more, p -value is less than 0.05), the log-derived data can be determined as valid data.

Below, a method of predicting the probability of occurrence of a company's potential risk using log-derived data is explained as an example.

The processor 100 may generate first log derived data indicating the recency of the item corresponding to the log data by obtaining the last occurrence time of the item corresponding to the log data from the log data extracted from the open data. there is. The processor 100 may generate second log derived data indicating the frequency of occurrence of the item by obtaining all dates on which the corresponding item occurred during a specific period from the log data. The processor 100 may generate third log derived data indicating the magnitude of occurrence of the item by obtaining the number of occurrences of the item corresponding to the log data from the log data.

The processor 100 may calculate the probability of occurrence of a company's potential risk based on log-derived data. The processor 100 provides a first average risk occurrence probability corresponding to the first log derived data, a second average risk occurrence probability corresponding to the second log derived data, and a second average risk occurrence probability corresponding to the first log derived data, based on the distribution of the average risk occurrence probability according to the log derived data. 3 The third average risk occurrence probability corresponding to the log-derived data can be calculated.

The processor 100 includes a first average risk occurrence probability, a second average risk occurrence probability, a third average risk occurrence probability, a first weight corresponding to the first average risk occurrence probability, and a second average risk occurrence probability. The probability of occurrence of a potential risk may be calculated based on the third weight corresponding to the weight and the third average probability of risk occurrence.

The processor 100 may predict the probability of occurrence of a company's potential risk by inputting the first log-derived data, the second log-derived data, and the third log-derived data into an artificial intelligence algorithm. To this end, artificial intelligence algorithms can be trained in advance to predict the probability of a company's potential risk occurrence by inputting log-derived data. In a further implementation, the processor 100 may input the first log-derived data, the second log-derived data, and the third log-derived data into an artificial intelligence algorithm to determine the first weight, the second weight, and the third weight.

The processor 100 can learn an artificial intelligence algorithm. The processor 100 may determine the weight based on a learned artificial intelligence algorithm, and depending on the embodiment, may determine the weight through a statistical algorithm.

The processor 100 may generate a credit evaluation result by determining a company's credit rating based on the probability of occurrence of a potential risk.

The memory 200 may store data for calculation or calculation results. The memory 200 may store instructions (or programs) that can be executed by a processor. For example, the instructions may include instructions for executing the operation of the processor and/or the operation of each component of the processor.

The memory 200 may be implemented as a volatile memory device or a non-volatile memory device.

Volatile memory devices may be implemented as dynamic random access memory (DRAM), static random access memory (SRAM), thyristor RAM (T-RAM), zero capacitor RAM (Z-RAM), or twin transistor RAM (TTRAM).

Non-volatile memory devices include EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory, MRAM (Magnetic RAM), Spin-Transfer Torque (STT)-MRAM (MRAM), and Conductive Bridging RAM (CBRAM). , FeRAM (Ferroelectric RAM), PRAM (Phase change RAM), Resistive RAM (RRAM), Nanotube RRAM (Nanotube RRAM), Polymer RAM (PoRAM), Nano Floating Gate Memory (NFGM), holographic memory, molecular electronic memory device, or insulation resistance change memory.

FIG. 2 shows an example of implementation of the credit rating information providing device shown in FIG. 1, and FIG. 3 is a diagram for explaining a pulse algorithm for providing credit rating information shown in FIG. 1.

2 and 3, the processor (e.g., processor 100 in FIG. 1) includes a sample inlet 110, a data collector 120, a log extractor 130, a significance checker 140, and a score calculator ( 150), a rating designer 160, and a performance evaluator 170.

The memory (eg, memory 200 in FIG. 1) may include a DB (DataBase) 210.

Sample inputter 110, data collector 120, log extractor 130, significance checker 140, score calculator 150, rating designer 160, and performance evaluator 170 are from DB 210. Data can be received or data can be stored in DB 210.

The operations of the sample inlet 110, data collector 120, log detector 130, significance checker 140, score calculator 150, rating designer 160, and performance evaluator 170 are shown in Figures 4 to 4. This will be described in detail with reference to FIG. 11.

The processor 100 may predict corporate risks using a pulse algorithm, logistic regression, decision tree, and/or random forest. The processor 100 collects various open data about the company and determines the company's credit based on the recency (310), frequency (330), and magnitude (350) of the collected open data. It can provide credit assessment information about risk.

Recency can refer to the time from the time an event corresponding to open data occurred to the present. Frequency may refer to the cycle (or total number of occurrence dates) in which events corresponding to open data occur. Scale can refer to the number of incidents corresponding to open data.

The processor 100 can determine the recency of open data and classify events corresponding to the data as recent events or old events. The processor 100 can use recency to verify the effectiveness of each company's business activities from a current perspective.

The processor 100 can identify the frequency of open data and distinguish whether events corresponding to the data occur frequently or occasionally. The processor 100 can quantitatively derive the continuity or permanence of a company's business activities based on frequency.

The processor 100 can determine the size of open data and distinguish whether an event corresponding to the data occurred in large quantities or in small quantities. The processor 100 can check the influence or ripple effect of the business activities of individual companies in the market based on size. In the presented example, the processor 100 is illustrated as utilizing the recency, frequency, and scale of open data, but the present invention predicts the probability of occurrence of a company's potential risk by utilizing arbitrary time-series indicators generated by reprocessing open data. A person skilled in the art will understand that it can be implemented in this way.

FIG. 4 is a diagram for explaining the operation of the sample inlet shown in FIG. 2.

Referring to FIG. 4, a sample inlet device (e.g., sample inlet device 110 of FIG. 2) may collect information about a company from a plurality of organizations. The plurality of institutions may include financial (or credit) institutions.

The sample inflow device 110 may collect information related to a company's loan to be used as reference data from a plurality of institutions. Information related to the loan may include company sample, identification number, reference point, and default status. In addition, those skilled in the art will understand that the information related to the loan may include arbitrary information related to the corporate loan, such as whether the company is delinquent or closed, in addition to whether it is bankrupt.

A corporate sample can reveal information about the sample range of borrowers of the companies used as a sample. The borrower sample range may include all or part of the borrowers within credit institutions such as banks, credit card companies, savings banks, and credit rating agencies. Borrowers may include incorporated companies.

The identification number may represent the company's identification information. Identification information may include a corporate registration number or business registration number that can be used as a standard for inflow of open data.

The reference point may indicate the point of measurement of data to manage a company's credit risk. The reference point may include loan execution for each borrower or monitoring time after loan execution to measure default.

Bankruptcy can indicate whether the sample company is bankrupt. Whether or not a default occurs may include whether an actual default occurs within a specific period (e.g., 1 year or 2 years) from the reference point.

The sample inlet 110 can define whether default has occurred. The sample inflow unit 110 includes card delinquency, bond sale, corporate closure, provision for specific loan losses, debt recovery, card repayment, general write-off, card write-off, corporate workout, issuance of asset-backed securities, and /Or, it is possible to define whether a company goes bankrupt based on bankruptcy application information.

The sample inflow device 110 can transmit the defined default criteria to a plurality of financial institutions so that they can be used as criteria for determining default.

In the presented example, the sample inflow device 110 is explained in a way that defines whether the company is bankrupt through loan-related information and uses this to calculate the risk occurrence probability for credit risk, but this is not an exemplary implementation of the present invention. However, depending on the embodiment, the present invention may be implemented in the direction of calculating/predicting the probability of occurrence of a company's credit risk through not only information on bankruptcy, but also arbitrary loan-related information such as delinquency and business closure. A person skilled in the art will understand that this is possible.

FIG. 5 is a diagram for explaining the operation of the data collector shown in FIG. 2.

Referring to FIG. 5, a data collector (e.g., data collector 120 in FIG. 2) processes open data (data disclosed to the public) about companies collected from various organizations into a form suitable for risk management and stores it in a database (e.g. : Can be stored in DB 210 of FIG. 2).

For example, the data collector 120 includes company identification information (511), organization information (512), authentication information (513), geographic information (514), intellectual property rights information (515), and fund attraction information (516) among open data. ), technology development information (517), business information (518), press release information (519), financial information (520), and other information (521) can be collected.

The data collector 120 can collect customized data for a company using an open application programming interface (API), scraping, and conversion.

Identification information (511) may include the company's business name, representative name, corporate registration number, and business registration number. The data collector 120 may, for example, collect identification information 511 from the Supreme Court Registry and the National Tax Service.

Organizational information 512 may include information on monthly employment, monthly employees joining/leaving the company, average wages, and employee turnover rate. The data collector 120 may, for example, collect organizational information 512 from the National Pension Service and the Korea Employment Information Service.

Authentication information 513 may include venture companies, Innobiz, main biz, small giant companies, (preliminary) social enterprises, etc. The data collector 120 may exemplarily collect authentication information 513 from the Ministry of SMEs and Startups and the Social Enterprise Promotion Agency.

Geographic information 514 may include registered address, publicly announced land price, head office relocation, past address, publicly announced land price history, etc. The data collector 120 may, for example, collect geographic information 514 from the Ministry of Land, Infrastructure and Transport and the Supreme Court registry office.

Intellectual property rights information 515 may include application/registration/disclosure information of patents, utility models, trademarks, and designs, copyright information, etc. The data collector 120 may, for example, collect intellectual property rights information 515 from the Korean Intellectual Property Office and the Korea Copyright Commission.

Fund attraction information 516 may include investment attraction for each company, investment institution, investment stage, investment period, government contribution fund information, etc. The data collector 120 may, by way of example, collect fund attraction information 516 from the Korea Securities Depository and the Korea Press Foundation.

Technology development information 517 may include information on R&D contribution funds, performance period, business classification, supervising organization, development details, etc. The data collector 120 may, by way of example, collect technology development information 517 from the Ministry of SMEs and Startups and the Ministry of Science and ICT.

Business information (518) can collect information on R&D technology development content, Public Procurement Service bid/success, homepage text, etc. The data collector 120 may, for example, collect business information 518 from the Ministry of Science and ICT, the Public Procurement Service, etc.

Media report information 519 may include information about media and video reports, number of reports, time of occurrence, media company, report contents, etc. The data collector 120 may, for example, collect press release information 519 from the Korea Press Foundation and search engine API.

Financial information 520 may include detailed item information related to the income statement and statement of financial position. The data collector 120 may collect financial information 520 from the Financial Supervisory Service, the Ministry of Strategy and Finance, etc.

Other information 521 may include information on large amounts and habitual delinquency details, total amount of delinquency, representative taxes, payment date, summary of delinquency, etc. The data collector 120 may collect other information 521 from the National Tax Service, the Korea Customs Service, etc.

Those skilled in the art will understand that the types and collection sources of open data described above are not limited to the examples presented, and that the types and collection sources of open data collected by the data collector 120 can be expanded.

FIG. 6 is a diagram for explaining the operation of the log extractor shown in FIG. 2.

Referring to FIG. 6, a log extractor (e.g., log extractor 130 of FIG. 2) may generate log derived data by extracting log data according to time of open data. In the example of FIG. 6 , log extractor 130 may extract log data for each piece of open data (e.g., data on news, employment, patents, or certification items). Log data may include data that describes the time at which an item occurred in a time series manner.

The log extractor 130 may process log data to generate log derived data. For example, the log derived data may include first log derived data, second log derived data, and/or third log derived data that processes the log data based on the relevant time.

The log extractor 130 may generate first log derived data indicating the recency of each item by obtaining the final occurrence time of each item corresponding to the log data from the log data. In the example of Figure 6, the first log derived data may include a data elapsed period.

The log extractor 130 may obtain from log data all dates on which each item corresponding to the log data occurred, thereby generating second log derived data indicating the frequency of occurrence of the item. In the example of FIG. 6, the second log derived data may indicate the frequency of occurrence of items corresponding to each log data. The frequency of occurrence of data may be calculated based on the number of data occurrences during a predetermined period.

The log extractor 130 may generate third log derived data indicating the scale of occurrence of the item by obtaining the number of occurrences of the item corresponding to the log data from the log data. In the example of FIG. 6, the third log derived data may include the total number of occurrences of each log data.

The first log-derived data and the second log-derived data presented above are only examples, and those skilled in the art will understand that log-derived data can be generated from log data through various types of data processing.

FIG. 7 is a diagram for explaining an example of the operation of the significance inspector shown in FIG. 2.

Referring to FIG. 7, a significance checker (e.g., significance checker 140 of FIG. 2) can extract valid data by verifying the significance of log-derived data.

The significance checker 140 can verify the significance of each log derived data extracted from the collected open data. The significance checker 140 can distinguish between companies that have generated items corresponding to each log-derived data (e.g., export record data, investment attraction data, patent registration data, etc.) and companies that have not generated items for each item. . The significance checker 140 may calculate the difference in probability of risk occurrence (e.g., default rate) between a group of companies in which an item occurs and a group of companies in which an item does not occur, and extract meaningful data as valid data based on the difference in probability of risk occurrence. The default rate presented as an example in the above process can be calculated based on information related to loans obtained through the sample inflow period described above.

The significance checker 140 can distinguish between companies in which an item corresponding to log-derived data extracted from the collected data (e.g., data 6 in FIG. 7) occurred and companies in which it did not. For example, when the target open data item is patent data, the significance checker 140 can classify companies that generate patent data and companies that do not. The significance checker 140 corresponds to valid data through the difference between the probability of occurrence of a first risk (e.g., first default rate) of a company in which patent data occurred and the probability of occurrence of a second risk (e.g., second default rate) of a company in which patent data did not occur. Log derived data can be determined.

According to one embodiment, the significance checker 140 may determine valid data by applying a statistical test method to the difference between the probability of occurrence of the first risk and the probability of occurrence of the second risk described above. For example, the significance checker 140 applies a Chi-squared test to the difference between the probability of occurrence of the first risk and the probability of occurrence of the second risk, and converts the log-derived data determined to be significant into valid data. You can decide. The test method presented illustratively above is only an example of implementation, and those skilled in the art will understand that significance can be determined through a test using an arbitrary statistical method.

According to another embodiment, the significance checker 140 may determine whether the difference in risk occurrence probability (e.g., default rate) is greater than a threshold as shown in FIG. 7 and determine valid data based on the judgment result. there is. For example, if the difference in default rates is greater than the threshold, the significance checker 140 may determine the log-derived data corresponding to the item to be meaningful data and determine it as valid data, or if the difference in default rates is less than the threshold. In the same case, the log-derived data corresponding to the item may be judged as meaningless data and determined not to be valid data. In the example of FIG. 7, the threshold may be 0.01. In this case, if the difference in the default rate is greater than 0.01, the significance checker 140 may determine that the log-derived data corresponding to data 6 is significant data and determine this as valid data.

Those skilled in the art will understand that the types and thresholds of each item included in the open data presented above are merely examples, and the present invention is not limited thereto.

According to another embodiment, the processor 100 uses a statistical test method ( For example, valid data can be determined by applying a T-test. More specifically, the processor 100, for each log derived data, if there is a statistically significant difference between the average value of companies that have defaulted and the average value that has not defaulted (for example, the t statistic is 1.64 or more, p -value is less than 0.05), the log-derived data can be determined as valid data.

FIG. 8 is a diagram for explaining the operation of the score calculator shown in FIG. 2.

Referring to FIG. 8, a score calculator (e.g., score calculator 150 of FIG. 2) may calculate the probability of occurrence of a potential risk (e.g., potential default rate) of a company based on log-derived data determined as valid data. For example, the score calculator 150 may determine a first average risk occurrence probability (e.g., , a first average default rate), a second average risk occurrence probability corresponding to the second log-derived data (e.g., a second average default rate), and a third average risk occurrence probability corresponding to the third log-derived data (e.g. , the second average default rate) can be calculated.

The score calculator 150 includes a first average risk occurrence probability, a second average risk occurrence probability, a third average risk occurrence probability, a first weight corresponding to the first average risk occurrence probability, and a second average risk occurrence probability. The probability of occurrence of a potential risk can be calculated based on the third weight corresponding to the second weight and the third average probability of risk occurrence.

In the example of FIG. 8 , the score calculator 150 may calculate a first average default rate (eg, 0.3%) based on the distribution of average default rates for recency. The score calculator 150 may calculate a second average default rate (eg, 0.9%) based on the distribution of the average default rate with respect to frequency. The score calculator 150 may calculate a third average default rate (eg, 0.1%) based on the distribution of average default rates for scale.

The example in FIG. 8 may exemplarily represent a case where all weights are 1. Depending on the embodiment, the weight may not be 1 and may have different values depending on the average default rate.

The score calculator 150 may determine the first weight, the second weight, and the third weight by inputting the first log-derived data, the second log-derived data, and the third log-derived data into an artificial intelligence algorithm.

The score calculator 150 may input log data or log-derived data into an artificial intelligence algorithm to determine the distribution of the average default rate and calculate the potential default rate based on the average default rate distribution.

In the example presented above, it is illustrated that the potential default rate is calculated based on the first average default rate to the third average default rate, but this is only an example of the present invention, and the number of average default rates calculated according to the method of calculating log derived data, etc. can be implemented in various ways.

FIG. 9 is a diagram for explaining the operation of the grade designer shown in FIG. 2.

Referring to FIG. 9, a rating designer (e.g., the rating designer 160 of FIG. 2) may determine a company's credit rating based on the probability of occurrence of a potential risk (e.g., a potential default rate).

The rating designer 160 may calculate the expected risk occurrence probability (eg, expected default rate) for the entire group of companies in the evaluation sample. The rating designer 160 may classify credit evaluation ratings based on the distribution of expected risk occurrence probability. The grade designer 160 can classify management grades based on an equal distribution method.

The rating designer 160 may determine the company's credit rating based on the expected risk occurrence probability of the entire company group and the calculated potential risk occurrence probability.

A credit rating may consist of multiple ratings. The example of FIG. 9 shows the result of dividing the credit rating into 10 levels, but depending on the embodiment, the credit rating may be less than 10 or more than 10.

The rating designer 160 may classify a company as having a high risk occurrence probability (e.g., default rate) if the company has a high credit rating rating, and may classify the company as having a high risk occurrence probability (e.g., default rate), and if the company has a low credit rating rating, the rating designer 160 may classify the company as having a high risk occurrence probability (e.g., default rate). It can be classified as a company with low probability.

FIG. 10 is a diagram for explaining the risk occurrence probability calculation operation of the performance evaluator shown in FIG. 2. Figure 11 shows an example of a credit evaluation result of the performance evaluator shown in Figure 2.

Referring to FIGS. 10 and 11 , a performance evaluator (eg, performance evaluator 170 of FIG. 2 ) may generate a credit evaluation result based on the credit evaluation grade. The performance evaluator 170 may evaluate the performance of the credit evaluation result.

The performance evaluator 170 may evaluate performance based on AR (Accuracy Ratio). For example, the performance evaluator 170 may calculate AR based on data regarding actual default and the default rate according to the grade calculated by the rating designer (e.g., the rating designer 160 of FIG. 2).

In the example of Figure 11, it can be seen that AR increases as the cumulative composition ratio of all borrowers (e.g., companies) increases.

FIG. 12 shows a flowchart of the operation of the credit rating information providing device shown in FIG. 1.

Referring to FIG. 12, a processor (e.g., processor 100 of FIG. 1) may collect company-related information (1210). For example, information related to a company may be open data, and open data is a series of information that is open to the public, such as the company's business details, the company's financial information, the company's employment information, the company's investment information, and the company's information. It may include individual items corresponding to information about the technology possessed, the company's corporate certification information, and/or the company's promotional materials. Those skilled in the art will understand that the types of items included in open data are not limited to the presented examples.

The processor 100 may generate log derived data by extracting log data according to time from open data (1230). Exemplarily, the log-derived data may include first log-derived data, second log-derived data, and/or third log-derived data.

The processor 100 may generate first log derived data indicating the recency of each item by obtaining the final occurrence time of each item corresponding to the log data from the log data. The processor 100 may obtain from the log data all dates on which items corresponding to the log data occurred, thereby generating second log derived data indicating the frequency of occurrence of the items. The processor 100 may generate third log derived data indicating the scale of occurrence of the item by obtaining the number of occurrences of the item corresponding to the log data from the log data.

The processor 100 may extract valid data by verifying the significance of the log-derived data (1250). More specifically, the processor 100 may obtain valid data by verifying the significance of the item corresponding to each log derived data, which may be as described with reference to FIG. 7 .

For example, for each log-derived data, the processor 100 may obtain the first risk occurrence probability (eg, first default rate) of the company in which the item corresponding to the log-derived data was generated. The processor 100 may obtain the second risk occurrence probability (second default rate) of the company in which no item has occurred. The processor 100 may apply a statistical test method to the difference between the probability of occurrence of the first risk and the probability of occurrence of the second risk, or may determine valid data based on whether the difference is greater than a predetermined threshold. The specific method of determining valid data may be the same as previously described in FIG. 7.

The processor 100 may calculate the probability of occurrence of a company's potential risk based on log-derived data (1270). The processor 100 generates a first average risk occurrence probability corresponding to the first log derived data and a second log derivative based on the distribution of the average risk occurrence probability according to the log derived data (e.g., average default rate for each log derived data). A second average risk occurrence probability corresponding to the data and a third average risk occurrence probability corresponding to the third log-derived data may be calculated.

The processor 100 includes a first average risk occurrence probability, a second average risk occurrence probability, a third average risk occurrence probability, a first weight corresponding to the first average risk occurrence probability, and a second average risk occurrence probability. The probability of occurrence of a potential risk (potential default rate) can be calculated based on the third weight corresponding to the weight and the third average probability of risk occurrence.

The processor 100 may determine the first weight, the second weight, and the third weight by inputting the first log-derived data, the second log-derived data, and the third log-derived data into the neural network. Depending on the embodiment, the processor 100 may directly calculate the probability of occurrence of a potential risk by inputting the first log-derived data, the second log-derived data, and the third log-derived data into an artificial intelligence algorithm.

The processor 100 can learn an artificial intelligence algorithm. The processor 100 may determine the weight based on a learned artificial intelligence algorithm or directly calculate the potential default rate.

The processor 100 may generate a credit evaluation result by determining the company's credit rating based on the probability of occurrence of a potential risk (1290).

The embodiments described above may be implemented with hardware components, software components, and/or a combination of hardware components and software components. For example, the devices, methods, and components described in the embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, and a field programmable gate (FPGA). It may be implemented using a general-purpose computer or a special-purpose computer, such as an array, programmable logic unit (PLU), microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and software applications running on the operating system. Additionally, a processing device may access, store, manipulate, process, and generate data in response to the execution of software. For ease of understanding, a single processing device may be described as being used; however, those skilled in the art will understand that a processing device includes multiple processing elements and/or multiple types of processing elements. It can be seen that it may include. For example, a processing device may include multiple processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are possible.

Software may include a computer program, code, instructions, or a combination of one or more of these, which may configure a processing unit to operate as desired, or may be processed independently or collectively. You can command the device. Software and/or data may be used on any type of machine, component, physical device, virtual equipment, computer storage medium or device to be interpreted by or to provide instructions or data to a processing device. , or may be permanently or temporarily embodied in a transmitted signal wave. Software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored on a computer-readable recording medium.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. A computer-readable medium may store program instructions, data files, data structures, etc., singly or in combination, and the program instructions recorded on the medium may be specially designed and constructed for the embodiment or may be known and available to those skilled in the art of computer software. there is. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -Includes optical media (magneto-optical media) and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, etc. Examples of program instructions include machine language code, such as that produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc.

The hardware devices described above may be configured to operate as one or multiple software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described with limited drawings, those skilled in the art can apply various technical modifications and variations based on this. For example, the described techniques are performed in a different order than the described method, and/or components of the described system, structure, device, circuit, etc. are combined or combined in a different form than the described method, or other components are used. Alternatively, appropriate results may be achieved even if substituted or substituted by an equivalent.

Therefore, other implementations, other embodiments, and equivalents of the claims also fall within the scope of the claims described below.

Claims (6)

Collect public data, including intellectual property information, authentication information, and organizational information related to the company;
By extracting log data over time for each of the plurality of items included in the public data, log derived data corresponding to each of the plurality of items is generated,
Extract valid data by verifying the significance of the log-derived data,
Calculate the probability of occurrence of a potential risk of the company based on log-derived data corresponding to the valid data,
A processor that generates credit evaluation results by determining the credit rating of the company based on the probability of occurrence of the potential risk.
Including,
The processor,
For each of the log-derived data, calculate a first average value of the log-derived data for the first group of companies in which a risk occurred and a second average value of the log-derived data for a second group of companies in which a risk did not occur,
A T-test is applied to the difference between the first average value and the second average value to determine whether the data is valid, and the log-derived data determined to have a significant difference between the first average value and the second average value are valid. data, and determine that log-derived data for which the difference between the first average value and the second average value is not significantly different is not the valid data,
The risk corresponding to the first average value and the second average value is calculated based on loan-related information of the first group of companies and the second group of companies,
A corporate credit evaluation information providing device in which the plurality of items correspond to information different from the loan-related information.
According to paragraph 1,
The processor,
Based on the distribution of the average risk occurrence probability according to the log-derived data, calculate the average risk occurrence probability corresponding to the log-derived data,
Calculating the potential default rate based on a weight for the average default rate,
A device for providing corporate credit rating information.
According to paragraph 1,
The processor,
Generate first log derived data indicating recency of the item by obtaining the last occurrence time of the item corresponding to the log data from the log data,
Generate second log derived data indicating the frequency of occurrence of the item by obtaining all dates on which the log data occurred from the log data,
Generating third log derived data indicating the magnitude of occurrence of the item by obtaining the number of occurrences of the log data from the log data,
A device for providing corporate credit rating information. According to paragraph 3,
The processor,
Based on the distribution of the average risk occurrence probability according to the log-derived data, the first average risk occurrence probability corresponding to the first log-derived data, the second average risk occurrence probability corresponding to the second log-derived data, and the third Calculate the third average risk occurrence probability corresponding to the log-derived data,
The first average probability of risk occurrence, the second average probability of risk occurrence, the third average probability of risk occurrence, a first weight corresponding to the first average probability of risk occurrence, and a second weight corresponding to the second average probability of risk occurrence Calculating the probability of occurrence of the potential risk based on the third weight corresponding to the weight and the third average probability of risk occurrence,
A device for providing corporate credit rating information. According to paragraph 4,
The processor,
Determining the first weight, the second weight, and the third weight by inputting the first log-derived data, the second log-derived data, and the third log-derived data into a machine learning model,
A device for providing corporate credit rating information. In a method of providing corporate credit evaluation information performed by a corporate credit evaluation information providing device,
collecting public data including intellectual property information, authentication information, and organizational information related to the enterprise;
generating log derived data corresponding to each of the plurality of items by extracting log data over time for each of the plurality of items included in the public data;
extracting valid data by verifying the significance of the log-derived data;
calculating a probability of occurrence of a potential risk of the company based on the log-derived data; and
Generating a credit evaluation result by determining the credit rating of the company based on the probability of occurrence of the potential risk
Including,
The step of extracting the valid data is,
For each of the log-derived data, calculating a first average value of the log-derived data for a first group of companies in which a risk occurred and a second average value of the log-derived data for a second group of companies in which a risk did not occur; and
A T-test is applied to the difference between the first average value and the second average value to determine whether the data is valid, and the log-derived data determined to have a significant difference between the first average value and the second average value are valid. determining the log-derived data as data, and determining that the difference between the first average value and the second average value is not significant, is not the valid data.
Including,
The risk corresponding to the first average value and the second average value is calculated based on loan-related information of the first group of companies and the second group of companies,
A method of providing corporate credit evaluation information, wherein the plurality of items correspond to information different from the information related to the loan.