Apparatus and method for reconciling multiple data sources based on industry characteristics

Abstract

The present invention relates to an industry-specific multi-data source automatic reconciliation device and method capable of performing automatic reconciliation of transaction history by differentially managing reliability and time consistency for multiple financial data sources by industry. The processor calculates weights by industry based on the industry code and transaction information of a target company, monitors the last update time for each data source used for automatic reconciliation, calculates the sensitivity of each data source to accounting processing by industry, determines whether to perform reconciliation for each data source based on the last update time and sensitivity, performs automatic reconciliation for each data source based on the determination result, filters candidate account items according to industry-specific accounting standards based on transaction information and calculates confidence levels, selects the account item with the highest score by calculating a score by summing the product of the industry-specific weight and confidence level for each candidate account item, and applies it as an industry-specific reconciliation rule. When an accounting expert modifies the selected account item, the reconciliation rule can be optimized through industry-specific misclassification pattern analysis and confidence level adjustment based on the modification details.

Description

Technology Field

[0001] The present invention relates to an industry characteristic-based multi-data source automatic reconciliation apparatus and method capable of performing automatic transaction reconciliation by multi-layeredly analyzing the industry characteristics of a company and differentially managing the reliability and time consistency of multiple financial data sources by industry. Background Technology

[0002] Generally, existing automated accounting systems can provide the function of collecting various financial data, such as card transaction history, account deposits and withdrawals, tax invoices, and PG (Payment Gateway) settlement data, and automatically journalizing them into accounting entries.

[0003] However, this existing technological approach has the following problems.

[0004] First, since a company's actual business model is determined solely by the industry code on the business registration certificate, misclassification may occur even among companies with the same industry code if their actual revenue or cost structures differ.

[0005] Second, all data sources are treated with equal weight, and uniform metabolic standards are applied even though important data sources differ by industry, which may reduce the accuracy of metabolic analysis by industry.

[0006] For example, in online commerce, PG settlement data is key to revenue recognition, but in professional services, the timing of tax invoice issuance is more important, so applying uniform reconciliation standards may reduce reconciliation accuracy.

[0007] Third, since this is not taken into account despite the fact that the update cycles of each data source differ, incomplete judgments may occur during real-time dialogue due to delays in some sources.

[0008] Fourth, there is a problem in that the same transaction item (for example, server hosting costs) should be treated differently as cost of goods sold, selling and administrative expenses, or assets depending on the industry, but this is not reflected.

[0009] Fifth, in the case of hybrid companies operating multiple business models (for example, selling their own products + a marketplace + providing SaaS), accurate analysis is impossible using single-industry rules.

[0010] Therefore, there is a need for technology that can improve the accuracy of automated transaction reconciliation by differentially managing the reliability and time-span consistency of multiple financial data sources according to industry characteristics. Prior art literature

[0011] Korean Registered Patent 10-2752894 (2025.01.07) The problem to be solved

[0012] One objective of the present invention, aimed at solving the problems described above, is to provide an industry characteristic-based multi-data source automatic reconciliation apparatus and method capable of performing automatic transaction reconciliation by multi-layeredly analyzing the industry characteristics of a company and differentially managing the reliability and time consistency of multiple financial data sources by industry.

[0013] The problems that the present invention aims to solve are not limited to those mentioned above, and other unmentioned problems will be clearly understood by a person skilled in the art from the description below. means of solving the problem

[0014] An industry-specific multi-data source automatic reconciliation device according to an embodiment of the present invention for solving the aforementioned problem comprises a communication module that acquires industry codes and transaction information of a target company, a database that stores industry codes and transaction information of a target company, and a processor that controls data source reconciliation processing according to industry-specific characteristics based on industry codes and transaction information of a target company. The processor calculates weights for each industry based on industry codes and transaction information of a target company, monitors the final update time for each data source used for automatic reconciliation, calculates the sensitivity of each data source to accounting processing for each industry, determines whether to perform reconciliation for each data source based on the final update time and sensitivity, performs automatic reconciliation for each data source based on the determination result, filters candidate account items and calculates confidence levels according to industry-specific accounting standards based on transaction information, selects the account item with the highest score by calculating a score by summing the product of the industry-specific weight and confidence level for each candidate account item, and applies it as an industry-specific reconciliation rule. When an accounting expert modifies the selected account item, the reconciliation rule can be optimized through industry-specific misclassification pattern analysis and confidence level adjustment based on the modification details.

[0015] In an embodiment, the processor is characterized by, when calculating weights by industry, obtaining the business registration industry code of a target company through a communication module, crawling the target company's website and mobile app to extract service types, payment methods, and provision methods, and analyzing transaction patterns including the temporal distribution of sales and purchases, the distribution of business partners' industries, and the proportion of payment methods from transaction data of a predetermined past period, and calculating weights by industry based on the extraction and analysis results.

[0016] In an embodiment, the processor is characterized by extracting the service description of the main page, the fee structure of the pricing page, and the customer type of the customer case page when crawling the website; extracting the app category, in-app payment items, and keywords of user reviews when analyzing the mobile app; calculating the temporal concentration of sales occurrence, the industry classification of major business partners, the ratio by payment method, the average transaction amount and frequency, and the ratio of regular recurring transactions when analyzing transaction patterns, deriving a similarity score by industry from each analysis result; synthesizing the business registration industry code, web / app analysis results, and transaction pattern analysis results, and assigning a high weight to recent data to calculate a weight for each of multiple industries.

[0017] In an embodiment, the processor is characterized by analyzing the frequency with which each data source contributed to the final judgment in the past reconciliation history for each industry when calculating sensitivity, measuring the increase in the misclassification rate when each data source is omitted, and normalizing the sensitivity to a value between 0 and 1 by reflecting the importance of each data source in the industry-specific standard accounting practices.

[0018] In an embodiment, the processor is characterized by, when determining whether to perform metabolism, if there is a data source that is delayed at the time of metabolism execution based on the last update time and sensitivity, if the sensitivity for said source is greater than or equal to a preset threshold, the determination of metabolism is withheld, and if the sensitivity is less than the threshold, the processor determines to perform metabolism by generating a prediction value based on past patterns.

[0019] In an embodiment, the processor is characterized by identifying industry information regarding the counterparty of a transaction history based on transaction information when filtering candidate account items, and filtering candidate account items based on a combination of the industry of the target company and the industry of the counterparty.

[0020] In an exemplary embodiment, the processor is characterized by displaying a flag indicating that review is needed if the difference in scores between the account with the highest score and the account with the second-highest score is less than a predetermined threshold when applying industry-specific reconciliation rules.

[0021] In an embodiment, the processor is characterized by, when optimizing a reconciliation rule, generating a new rule if the correction frequency for the same transaction pattern is greater than or equal to a preset threshold value, operating the new rule as a candidate for a predetermined period to verify accuracy, and promoting the new rule to a formal rule if, after verification, the error rate is less than a preset threshold value.

[0022] In an embodiment, the processor is characterized by analyzing whether a rule learned in a specific industry appears similarly in other industries when optimizing a metabolic rule, and if it is confirmed as a common pattern, expanding it into a rule common to all industries.

[0023] A method for automatically reconciling multiple data sources based on industry characteristics according to an embodiment of the present invention comprises: a step of obtaining industry codes and transaction information of a target company; a step of calculating weights by industry based on the industry codes and transaction information of the target company; a step of monitoring the final update time for each data source used for automatic reconciliation, calculating the sensitivity of each data source to accounting processing by industry, determining whether to perform reconciliation by each data source based on the final update time and sensitivity, and performing automatic reconciliation by data source based on the determination result; a step of filtering candidate account items and calculating confidence levels according to industry-specific accounting standards based on transaction information, and selecting an account item with the highest score through a score calculation in which the product of the industry-specific weight and confidence level for each candidate account item is summed, and applying it as an industry-specific reconciliation rule; and a step of optimizing the reconciliation rule by analyzing industry-specific misclassification patterns and adjusting confidence levels based on the modification details when an accounting expert modifies the selected account item.

[0024] In addition to this, other methods for implementing the present invention, other systems, and computer-readable recording media for recording a computer program for executing said method may be further provided. Effects of the invention

[0025] As described above, according to the present invention, the industry characteristics of a company are analyzed in a multi-layered manner, and the reliability and time consistency of multiple financial data sources are managed differentially by industry to perform automatic reconciliation of transaction history.

[0026] In addition, the present invention can reduce the misclassification rate of automated reconciliation by improving the accuracy of industry identification through multi-layered analysis of not only business registration industry codes but also actual business models and transaction patterns.

[0027] In addition, the present invention enables accurate metabolization even for companies operating hybrid business models through a multi-industry weighted blending method.

[0028] In addition, the present invention can simultaneously ensure the accuracy and completeness of real-time metabolism through delay correction that considers the update cycle by data source and sensitivity by industry.

[0029] In addition, the present invention can accurately reflect industry-specific accounting practices and tax standards by applying different accounting standards depending on the industry to the same transaction.

[0030] In addition, the present invention can implement a self-generating system in which metabolic accuracy improves as the usage period increases by learning the modification history of accounting experts and continuously optimizing the rules.

[0031] In addition, the present invention can detect accounting fraud or errors at an early stage by automatically detecting abnormal transactions relative to normal patterns by industry.

[0032] In addition, the present invention can significantly improve the efficiency of group accounting operations by automatically identifying internal transactions and automating the journal entries for adjusting consolidated financial statements when operating multiple entities.

[0033] The effects of the present invention are not limited to those mentioned above, and other unmentioned effects will be clearly understood by a person skilled in the art from the description below. Brief explanation of the drawing

[0034] FIG. 1 is a diagram illustrating the overall operation of a multi-data source automatic metabolizer based on industry characteristics according to one embodiment of the present invention. FIG. 2 is a diagram illustrating the configuration of an industry-specific multi-data source automatic metabolization device according to one embodiment of the present invention. FIGS. 3 to 5 are drawings for illustrating the processor configuration of an industry-specific multi-data source automatic metabolization device according to an embodiment of the present invention. FIGS. 6 to 9 are flowcharts illustrating a multi-data source automatic reconciliation method based on industry characteristics according to an embodiment of the present invention. Specific details for implementing the invention

[0035] The advantages and features of the present invention and the methods for achieving them will become clear by referring to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below but may be implemented in various different forms. These embodiments are provided merely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the present invention, and the present invention is defined only by the scope of the claims.

[0036] The terms used in this specification are for describing embodiments and are not intended to limit the invention. In this specification, the singular form includes the plural form unless specifically stated otherwise in the text. The terms "comprises" and / or "comprising" used in this specification do not exclude the presence or addition of one or more other components in addition to the components mentioned. Throughout the specification, the same reference numerals refer to the same components, and "and / or" includes each of the mentioned components and all combinations of one or more. Although terms such as "first," "second," etc., are used to describe various components, these components are not limited by these terms. These terms are used merely to distinguish one component from another. Therefore, the first component mentioned below may be the second component within the technical scope of the invention.

[0037] Unless otherwise defined, all terms used herein (including technical and scientific terms) may be used in a meaning commonly understood by those skilled in the art to which the present invention pertains. Additionally, terms defined in commonly used dictionaries are not to be interpreted ideally or excessively unless explicitly and specifically defined otherwise.

[0038] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

[0039] Prior to the explanation, the meanings of the terms used in this specification are briefly explained. However, since the explanation of terms is intended to aid in understanding this specification, it should be noted that they are not used to limit the technical scope of the invention unless explicitly stated to be a limiting factor.

[0040] FIG. 1 is a diagram illustrating the overall operation of a multi-data source automatic metabolizer based on industry characteristics according to one embodiment of the present invention.

[0041] As illustrated in FIG. 1, the industry characteristic-based multi-data source automatic reconciliation device (100) of the present invention can perform automatic reconciliation of transaction history by receiving an automatic reconciliation request for transaction history from a user terminal (200) or an external server (300), analyzing the industry characteristics of the company in a multi-layered manner, and managing the reliability and time consistency of multiple financial data sources differently by industry.

[0042] Here, the industry-specific multi-data source automatic transcription device (100) of the present invention may include both a standing device such as a PC (Personal Computer), Network TV, HBBTV (Hybrid Broadcast Broadband TV), Smart TV, and IPTV (Internet Protocol TV), and a mobile device or handheld device such as a smartphone, Tablet PC, Notebook, and PDA (Personal Digital Assistant).

[0043] Additionally, the industry-specific multi-data source automatic reconciliation device (100) of the present invention may be connected to an external server (300) and a user terminal (200) for communication and transmit the results of the automatic reconciliation of transaction history to the external server (300) and the user terminal (200).

[0044] Here, the network connecting the industry-specific multi-data source automatic reconciliation device (100) and the external server (300), or the industry-specific multi-data source automatic reconciliation device (100) and the user terminal (200) includes both wired and wireless networks, and refers collectively to a communication network that supports various communication standards or protocols for pairing or / and data transmission and reception between the industry-specific multi-data source automatic reconciliation device (100) and the external server (300), or the industry-specific multi-data source automatic reconciliation device (100) and the user terminal (200).

[0045] These wired and wireless networks include all communication networks currently or to be supported in the future by standards, and can support all one or more communication protocols for them.

[0046] These wired / wireless networks may be formed by networks for wired connections such as USB (Universal Serial Bus), CVBS (Composite Video Banking Sync), Component, S-Video (analog), DVI (Digital Visual Interface), HDMI (High Definition Multimedia Interface), RGB, and D-SUB, and communication standards or protocols for them, and networks for wireless connections such as Bluetooth, RFID (Radio Frequency Identification), infrared communication (IrDA: infrared Data Association), UWB (Ultra Wideband), ZigBee, DLNA (Digital Living Network Alliance), WLAN (Wireless LAN) (Wi-Fi), Wibro (Wireless broadband), Wimax (World Interoperability for Microwave Access), HSDPA (High Speed ​​Downlink Packet Access), LTE / LTE-A (Long Term Evolution / LTE-Advanced), and Wi-Fi Direct, and communication standards or protocols for them.

[0047] The industry characteristic-based multi-data source automatic reconciliation device (100) of the present invention calculates weights for each industry based on the industry code and transaction information of the target company, monitors the final update time for each data source used for automatic reconciliation, calculates the sensitivity of each data source to accounting processing for each industry, determines whether to perform reconciliation for each data source based on the final update time and sensitivity, performs automatic reconciliation for each data source based on the determination result, filters candidate account items according to industry accounting standards based on transaction information and calculates confidence level, selects the account item with the highest score by calculating a score by summing the product of the industry weight and confidence level for each candidate account item, and applies it as an industry reconciliation rule, and if an accounting expert modifies the selected account item, the reconciliation rule can be optimized by analyzing industry misclassification patterns and adjusting confidence level based on the modification details.

[0048] Here, the industry-specific multi-data source automatic reconciliation device (100) of the present invention, when calculating weights by industry, obtains the business registration industry code of the target company through a communication module, crawls the target company's website and mobile app to extract service type, payment method, and provision method, and analyzes transaction patterns including the temporal distribution of sales and purchases, the distribution of business partners' industries, and the proportion of payment methods from transaction data of a predetermined past period, and can calculate weights by industry based on the extraction and analysis results.

[0049] For example, the industry characteristic-based multi-data source automatic reconciliation device (100) of the present invention can extract service descriptions on main pages, fee structures on pricing pages, and customer types on customer case pages when crawling websites, extract app categories, in-app payment items, and keywords of user reviews when analyzing mobile apps, calculate the temporal concentration of sales generation, industry classification of major trading partners, ratios by payment method, average transaction amount and frequency, and regular recurring transaction ratios when analyzing transaction patterns, derive similarity scores by industry from each analysis result, combine business registration industry codes, web / app analysis results, and transaction pattern analysis results, and calculate weights for each of multiple industries by giving high weights to recent data.

[0050] The industry-specific multi-data source automatic reconciliation device (100) of the present invention can calculate sensitivity by analyzing the frequency with which each data source contributed to the final judgment in the past reconciliation history for each industry, measuring the increase in the misclassification rate when each data source is omitted, and normalizing the sensitivity to a value between 0 and 1 by reflecting the importance of each data source in the industry-specific standard accounting practices.

[0051] The industry-specific multi-data source automatic metabolism device (100) of the present invention, when determining whether to perform metabolism, if there is a data source that is delayed at the time of performing metabolism based on the last update time and sensitivity, if the sensitivity for the source is greater than or equal to a preset threshold, the metabolism determination is withheld, and if the sensitivity is less than the threshold, a prediction value based on past patterns is generated to determine whether to perform metabolism.

[0052] The industry characteristic-based multi-data source automatic reconciliation device (100) of the present invention can identify industry information regarding the business partner of the transaction history based on transaction information when filtering account candidate, and can filter account candidate based on the combination of the industry of the target company and the industry of the business partner.

[0053] The industry-specific multi-data source automatic reconciliation device (100) of the present invention can display a flag indicating that review is needed if the difference in scores between an account with the highest score and an account with the second-highest score is less than a predetermined threshold when applying industry-specific reconciliation rules.

[0054] The industry-specific multi-data source automatic reconciliation device (100) of the present invention can, when optimizing a reconciliation rule, generate a new rule if the correction frequency for the same transaction pattern is greater than or equal to a preset threshold value, verify the accuracy by operating the new rule as a candidate for a predetermined period, and if the error rate is less than a preset threshold value after verification, promote the new rule to an official rule.

[0055] The industry characteristic-based multi-data source automatic metabolization device (100) of the present invention can analyze whether a rule learned in a specific industry appears similarly in other industries when optimizing metabolization rules, and if it is confirmed as a common pattern, it can be expanded into a common rule for all industries.

[0056] In this way, the present invention can perform automatic reconciliation of transaction records by multi-layeredly analyzing the industry characteristics of a company and differentially managing the reliability and time consistency of multiple financial data sources by industry.

[0057] In addition, the present invention can reduce the misclassification rate of automated reconciliation by improving the accuracy of industry identification through multi-layered analysis of not only business registration industry codes but also actual business models and transaction patterns.

[0058] In addition, the present invention enables accurate metabolization even for companies operating hybrid business models through a multi-industry weighted blending method.

[0059] In addition, the present invention can simultaneously ensure the accuracy and completeness of real-time metabolism through delay correction that considers the update cycle by data source and sensitivity by industry.

[0060] In addition, the present invention can accurately reflect industry-specific accounting practices and tax standards by applying different accounting standards depending on the industry to the same transaction.

[0061] In addition, the present invention can implement a self-generating system in which metabolic accuracy improves as the usage period increases by learning the modification history of accounting experts and continuously optimizing the rules.

[0062] In addition, the present invention can detect accounting fraud or errors at an early stage by automatically detecting abnormal transactions relative to normal patterns by industry.

[0063] In addition, the present invention can significantly improve the efficiency of group accounting operations by automatically identifying internal transactions and automating the journal entries for adjusting consolidated financial statements when operating multiple entities.

[0064] FIG. 2 is a diagram illustrating the configuration of an industry-specific multi-data source automatic metabolization device according to one embodiment of the present invention.

[0065] As illustrated in FIG. 2, the industry characteristic-based multi-data source automatic reconciliation device (100) of the present invention may include a communication module (110) for acquiring industry codes and transaction information of a target company, a database (120) for storing industry codes and transaction information of a target company, and a processor (130) for controlling data source reconciliation processing according to industry characteristics based on industry codes and transaction information of a target company.

[0066] The processor (130) calculates weights by industry based on the industry code and transaction information of the target company, monitors the final update time for each data source used for automatic reconciliation, calculates the sensitivity of each data source to accounting processing by industry, determines whether to perform reconciliation by data source based on the final update time and sensitivity, performs automatic reconciliation by data source based on the determination result, filters candidate account items according to industry accounting standards based on transaction information and calculates confidence level, selects the account item with the highest score by calculating a score by summing the product of the industry weight and confidence level for each candidate account item, and applies it as an industry reconciliation rule, and if an accounting expert modifies the selected account item, the reconciliation rule can be optimized by analyzing industry misclassification patterns and adjusting confidence level based on the modification details.

[0067] Here, when calculating weights by industry, the processor (130) obtains the business registration industry code of the target company through the communication module (110), crawls the target company's website and mobile app to extract service type, payment method, and provision method, and analyzes transaction patterns including the temporal distribution of sales and purchases, the distribution of business partners' industries, and the proportion of payment methods from transaction data of a predetermined past period, and can calculate weights by industry based on the extraction and analysis results.

[0068] For example, the processor (130) can extract the service description of the main page, the fee structure of the pricing page, and the customer type of the customer case page when crawling the website, extract the app category, in-app payment items, and keywords of user reviews when analyzing the mobile app, calculate the temporal concentration of sales generation, the industry classification of major business partners, the ratio by payment method, the average transaction amount and frequency, and the ratio of regular recurring transactions when analyzing transaction patterns, derive a similarity score by industry from each analysis result, combine the business registration industry code, web / app analysis results, and transaction pattern analysis results, and calculate a weight for each of the multiple industries by giving a high weight to recent data.

[0069] Additionally, when the processor (130) analyzes transaction patterns, if a new transaction pattern that does not belong to an existing industry classification is detected, it registers it as a new industry candidate, and when sufficient transaction history is accumulated, it can generate an independent rule set for the pattern.

[0070] Next, the processor (130) can normalize the sensitivity to a value between 0 and 1 by analyzing the frequency with which each data source contributed to the final judgment in the past reconciliation history for each industry when calculating the sensitivity, measuring the increase in the misclassification rate when each data source is omitted, and reflecting the importance of each data source in the standard accounting practices for each industry.

[0071] Here, when a plurality of data sources include at least two of card transaction history, account deposit and withdrawal history, electronic tax invoice, PG settlement history, and cash receipt, the processor (130) may set the sensitivity of the PG settlement history to the highest level for the commerce industry, set the sensitivity of the electronic tax invoice and card transaction history to the highest level for the SaaS industry, and set the sensitivity of the electronic tax invoice to the highest level for the manufacturing industry.

[0072] Next, when determining whether to perform metabolism, the processor (130) may, based on the last update time and sensitivity, if there is a data source that is delayed at the time of performing metabolism, withhold the determination of metabolism if the sensitivity for said source is greater than or equal to a preset threshold, and if the sensitivity is less than the threshold, generate a prediction value based on past patterns and determine to perform metabolism.

[0073] And, when filtering account candidate, the processor (130) can identify industry information for the business partner of the transaction history based on transaction information and filter account candidate based on the combination of the business of the target company and the business partner's business.

[0074] In addition, the processor (130) may apply different account codes depending on the industry when applying industry-specific reconciliation rules.

[0075] For example, the processor (130) classifies the transaction item as cost of goods sold if it is a SaaS industry and as communication expenses if it is a manufacturing industry when the transaction item is a delivery fee if it is a commerce industry and as selling expenses if it is a manufacturing industry, and classifies the transaction item as one of welfare expenses, office supplies expenses, or current development expenses depending on the industry and use if the transaction item is a software subscription fee.

[0076] In addition, the processor (130) may apply different standards for capitalization and expense treatment by industry when applying industry-specific reconciliation rules.

[0077] For example, the processor (130) can treat the productization stage costs of technology development costs as intangible assets in the SaaS industry, treat molds and equipment as tangible assets in the manufacturing industry, treat consumable tools as expenses, and generate depreciation entries by applying the industry-specific standard useful life.

[0078] In addition, the processor (130) can apply differential timing for the recognition of sales and expenses by industry when applying industry-specific reconciliation rules.

[0079] For example, the processor (130) may recognize revenue based on the time of PG settlement confirmation in the case of a commerce and marketplace industry, recognize revenue after a predetermined period has elapsed after processing advance payments by allocating them over the service provision period in the case of a SaaS and subscription service industry, recognize revenue based on the time of tax invoice issuance in the case of a manufacturing and wholesale / retail industry, and recognize revenue based on the project progress rate or the time of achieving contract milestones in the case of a professional service industry.

[0080] Next, the processor (130) may display a flag indicating that review is needed if, when applying industry-specific reconciliation rules, the difference in scores between the account with the highest score and the account with the second-highest score is less than a predetermined threshold.

[0081] And, the processor (130) can assign a reliability score to the final metabolic result when applying industry-specific metabolic rules.

[0082] Here, the processor (130) assigns high confidence if all essential data sources are up to date, industry weights are concentrated in a specific industry, and do not correspond to anomaly detection, automatically confirms the final metabolite result if the assigned confidence score is above a preset threshold, and requests a review of the final metabolite result if the assigned confidence score is below the threshold.

[0083] Next, when optimizing the exchange rule, the processor (130) can create a new rule if the frequency of modification for the same transaction pattern is greater than or equal to a preset threshold value, verify the accuracy of the new rule by operating it as a candidate for a predetermined period, and if the error rate is less than a preset threshold value after verification, the new rule can be promoted to an official rule.

[0084] Additionally, when optimizing the metabolic rules, the processor (130) can analyze whether the rules learned in a specific industry appear similarly in other industries, and if confirmed as a common pattern, can expand them into common rules for all industries.

[0085] Next, the processor (130) can build a monthly or quarterly normal transaction pattern profile for each industry, and during the reconciliation process, if the amount differs by more than a predetermined multiple from the normal transaction pattern, or if an unexpected account item appears during the period, or if a high-value transaction with a new business partner occurs, it can classify it as an anomaly candidate, and perform time-series anomaly detection to lower the reconciliation reliability of the anomaly candidate transaction classified as an anomaly candidate and mark it as a subject for review.

[0086] Here, when the processor (130) performs time series anomaly detection, it can process the increase in transactions during the year-end promotion period of the commerce industry and the quarter-end contract renewal period of the SaaS industry as a normal pattern by reflecting industry-specific seasonality and recurring event information.

[0087] Next, the processor (130) can perform multi-corporate integrated reconciliation by determining whether a counterparty is an affiliate when reconciling transaction details for multiple corporations, identifying internal transactions, matching transaction details of both sides of the internal transaction, and if the business sectors of both corporations are different, cross-applying each business sector rule to verify the consistency of account items and amounts, and automatically generating a removal entry when preparing consolidated financial statements for internal transactions for which consistency has been confirmed.

[0088] Here, when the processor (130) performs multi-corporate consolidation reconciliation, if the amounts of both sides of an internal transaction are inconsistent, it may display a review flag on both sides, suggest possible causes such as separate handling of transportation costs, discounts, or input errors, and request the accounting officer to adjust the difference.

[0089] Meanwhile, according to one embodiment of the present invention, the processor (130) of the industry-specific multi-data source automatic metabolization device (100) may be composed of one or more cores and may include a processor for data analysis and deep learning, such as a central processing unit (CPU) of a computing device, a general purpose graphics processing unit (GPGPU), and a tensor processing unit (TPU).

[0090] The processor (130) can read a computer program stored in memory and perform data processing for machine learning according to an embodiment of the present invention. The processor (130) can perform operations for learning a neural network. The processor (130) can perform calculations for learning a neural network, such as processing input data for learning in deep learning (DL), extracting features from input data, calculating errors, and updating the weights of the neural network using backpropagation. At least one of the CPU, GPGPU, and TPU of the processor (130) can process the learning of a network function. For example, the CPU and GPGPU can together process the learning of a network function and data classification using the network function. In addition, in an embodiment of the present invention, the processors (130) of a plurality of computing devices can be used together to process the learning of a network function and data classification using the network function. In addition, the computer program executed in the computing device according to an embodiment of the present invention may be a CPU, GPGPU, or TPU executable program.

[0091] In addition, the communication module (110) of the industry-specific multi-data source automatic reconciliation device (100) may include a wireless internet module, a short-range communication module, a location information module, etc.

[0092] A wireless internet module refers to a module for wireless internet access, configured to transmit and receive wireless signals in a communication network based on wireless internet technologies.

[0093] Wireless internet technologies include, for example, WLAN (Wireless LAN), Wi-Fi (Wireless-Fidelity), Wi-Fi (Wireless Fidelity) Direct, DLNA (Digital Living Network Alliance), WiBro (Wireless Broadband), WiMAX (World Interoperability for Microwave Access), HSDPA (High Speed ​​Downlink Packet Access), HSUPA (High Speed ​​Uplink Packet Access), LTE (Long Term Evolution), LTE-A (Long Term Evolution-Advanced), etc., and the above wireless internet module transmits and receives data according to at least one wireless internet technology within a range that includes internet technologies not listed above.

[0094] From the perspective that wireless internet access via WiBro, HSDPA, HSUPA, GSM, CDMA, WCDMA, LTE, LTE-A, etc. is achieved through a mobile communication network, a wireless internet module that performs wireless internet access through a mobile communication network can be understood as a type of mobile communication module.

[0095] A short-range communication module is for short-range communication and can support short-range communication by using at least one of Bluetooth™, RFID (Radio Frequency Identification), Infrared Data Association (IrDA), UWB (Ultra Wideband), ZigBee, NFC (Near Field Communication), Wi-Fi (Wireless-Fidelity), Wi-Fi Direct, and Wireless USB (Wireless Universal Serial Bus) technologies.

[0096] A location information module is a module for obtaining the server's location (or current location), and representative examples include the GPS (Global Positioning System) module or the WiFi (Wireless Fidelity) module.

[0097] And, the database (120) of the industry-specific multi-data source automatic reconciliation device (100) may include at least one type of storage medium among flash memory type, hard disk type, multimedia card micro type, card type memory (e.g., SD or XD memory, etc.), RAM (Random Access Memory, RAM), SRAM (Static Random Access Memory), ROM (Read-Only Memory, ROM), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory), magnetic memory, magnetic disk, and optical disk.

[0098] The industry-specific multi-data source automatic reconciliation device (100) of the present invention may operate in connection with web storage that performs memory storage functions on the internet. The description of the database described above is merely an example and is not limited thereto.

[0099] Additionally, the industry-specific multi-data source automatic metabolization device (100) may further include an output unit and an input unit.

[0100] Here, the output unit can output information of any form generated or determined by the processor and information of any form received by the communication module.

[0101] For example, the output section may include at least one of a liquid crystal display (LCD), a thin film transistor-liquid crystal display (TFT LCD), an organic light-emitting diode (OLED), a flexible display, and a 3D display. Some of these display modules may be configured to be transparent or light-transmitting so that the outside can be seen through them. This may be referred to as a transparent display module, and representative examples of transparent display modules include TOLED (Transparent OLED).

[0102] Additionally, the input unit may receive user input. The input unit may include keys and / or buttons on a user interface or physical keys and / or buttons for receiving user input. A computer program according to embodiments of the present disclosure may be executed in accordance with user input through the input unit.

[0103] In addition, the input unit may receive a signal by detecting the user's button operation or touch input, or receive the voice or movements of the user, etc., through a camera or microphone and convert them into an input signal. For this purpose, speech recognition technology or motion recognition technology may be used.

[0104] Additionally, the input unit may be implemented as an external input device connected to an industry-specific multi-data source automatic transcription device (100). For example, the input device may be at least one of a touch pad, a touch pen, a keyboard, or a mouse for receiving user input, but this is merely an example and is not limited thereto.

[0105] For example, the input unit may recognize user touch input. In some cases, the input unit may have the same configuration as the output unit. The input unit may be composed of a touch screen implemented to receive user selection input. The touch screen may use any one of the following methods: contact capacitance method, infrared light detection method, surface ultrasonic (SAW) method, piezoelectric method, or resistive method. The detailed description of the touch screen described above is merely an example according to one embodiment of the present disclosure, and various touch screen panels may be employed in the platform server of the industry characteristic-based multi-data source automatic transcription device (100). The input unit composed of a touch screen may include a touch sensor. The touch sensor may be configured to convert changes such as pressure applied to a specific part of the input unit or capacitance occurring in a specific part of the input unit into an electrical input signal. The touch sensor may be configured to detect not only the location and area of ​​the touch but also the pressure at the time of the touch. When there is a touch input to the touch sensor, the corresponding signal(s) are sent to the touch controller. The touch controller can process the signal(s) and then transmit the corresponding data to the processor. This enables the processor to recognize whether or not a specific area of ​​the input part has been touched.

[0106] The industry-specific multi-data source automatic reconciliation device (100) of the present invention may include other components for performing a server environment of a server. The industry-specific multi-data source automatic reconciliation device (100) may include any type of device. The industry-specific multi-data source automatic reconciliation device (100) may be a digital device equipped with a processor and memory and having computational capabilities, such as a laptop computer, notebook computer, desktop computer, web pad, or mobile phone.

[0107] The industry-specific multi-data source automatic reconciliation device (100) of the present invention may be a computing system capable of generating a user interface according to embodiments of the present invention and providing information to a user terminal via a network. The industry-specific multi-data source automatic reconciliation device (100) of the present invention may be a cloud server. The industry-specific multi-data source automatic reconciliation device (100) may be a web server that processes services.

[0108] FIGS. 3 to 5 are drawings for illustrating the processor configuration of an industry-specific multi-data source automatic metabolization device according to an embodiment of the present invention.

[0109] As illustrated in FIG. 3, the processor (130) of the industry characteristic-based multi-data source automatic reconciliation device according to the present invention may include: a data collection module (131) that collects the business registration industry code of a target company, website and mobile app information, and past transaction data; an industry identification module (132) that analyzes the collected information and calculates weights for each of the multiple industries; a data source management module (133) that monitors the final update time for each data source, calculates the source sensitivity for each industry, and determines a correction policy for delayed sources; a reconciliation engine module (134) that applies multiple industry rules to transaction history and determines account items using a weighted blending method; a learning optimization module (135) that learns the correction history of accounting experts and automatically adjusts the rule weights for each industry; and an output module (136) that outputs reconciliation results, reliability scores, and items requiring review.

[0110] Here, the processor (130) may further include an anomaly detection module (137) that builds a normal transaction pattern profile by industry and detects abnormal transactions during the reconciliation process.

[0111] Additionally, the processor (130) may further include a multi-legal entity management module (138) that identifies internal transactions between multiple legal entities, verifies consistency, and generates a consolidated financial statement adjustment entry.

[0112] As illustrated in FIG. 4, the industry identification module (132) may include a web crawler module, an app analysis module, a transaction pattern analysis module, and a weight calculation module that calculates an industry weight by combining the analysis results of each module.

[0113] Additionally, as illustrated in FIG. 5, the metabolite engine module (134) may include a rule database that stores an industry-specific account mapping table, an industry-specific assetization standard table, and an industry-specific revenue recognition time table, and a blending engine module that receives transaction history and industry weights and queries the rule database to determine the optimal account.

[0114] Meanwhile, as shown in FIG. 3, the data collection module (131) can collect transaction details, tax invoices, settlement details, etc. from external data providers such as credit card companies, banks, the National Tax Service, and PG companies.

[0115] In addition, the data collection module (131) can crawl the target company's website and mobile app to collect service / product information, payment methods, delivery policies, etc.

[0116] Next, the industry identification module (132) can calculate the industry of the target company and multiple industry weights by comprehensively analyzing business registration information, web / app crawling data, and past transaction patterns.

[0117] And, the data source management module (133) can monitor the last update time of each data source, calculate the source sensitivity by industry, and determine a correction policy for delayed sources.

[0118] Next, the reconciliation engine module (134) automatically converts transaction details into accounting journals by applying different account mappings by industry, capitalization / expense processing standards, and sales / expense recognition timing.

[0119] The dialogue engine module (134) determines the final account item using a multiple industry weighted blending method, and the learning optimization module (135) learns the modification history of accounting experts to automatically adjust the rule weights for each industry and detect new industry patterns.

[0120] The output module (136) can provide the user with automatic metabolic results, reliability scores, items requiring review, etc.

[0121] FIGS. 6 to 9 are flowcharts illustrating a multi-data source automatic reconciliation method based on industry characteristics according to an embodiment of the present invention.

[0122] As illustrated in FIG. 6, the present invention can obtain the industry code and transaction information of the target company (S10).

[0123] Next, the present invention can calculate weights for each industry based on the industry code and transaction information of the target company (S20).

[0124] Herein, the present invention obtains the business registration industry code of a target company, crawls the target company's website and mobile app to extract service types, payment methods, and provision methods, and analyzes transaction patterns including the temporal distribution of sales and purchases, the distribution of business partners' industries, and the proportion of payment methods from transaction data of a predetermined past period, and can calculate industry-specific weights based on the extraction and analysis results.

[0125] In addition, the present invention monitors the final update time for each data source used for automatic reconciliation, calculates the sensitivity of each data source to accounting processing for each industry, determines whether to perform reconciliation for each data source based on the final update time and sensitivity, and performs automatic reconciliation for each data source based on the determination result (S30).

[0126] Here, the present invention can normalize the sensitivity to a value between 0 and 1 by analyzing the frequency with which each data source contributed to the final judgment in the past reconciliation history for each industry when calculating sensitivity, measuring the increase in the misclassification rate when each data source is omitted, and reflecting the importance of each data source in the standard accounting practices for each industry.

[0127] In addition, when determining whether to perform metabolism, the present invention may, based on the last update time and sensitivity, if there is a delayed data source at the time of metabolism execution, withhold the metabolism determination if the sensitivity for said source is greater than or equal to a preset threshold, and if the sensitivity is less than said threshold, generate a prediction value based on past patterns to determine whether to perform metabolism.

[0128] Next, the present invention can filter candidate account items and calculate confidence levels based on industry-specific accounting standards based on transaction information, and select the account item with the highest score by calculating a score by summing the product of the industry-specific weight and confidence level for each candidate account item, and apply it as an industry-specific reconciliation rule (S40).

[0129] Herein, when filtering candidate account items, the present invention identifies industry information regarding the counterparty of the transaction history based on transaction information, and can filter candidate account items based on a combination of the industry of the target company and the industry of the counterparty.

[0130] In addition, when applying industry-specific reconciliation rules, the present invention can display a flag indicating that review is needed if the score difference between the account with the highest score and the account with the second-highest score is less than a predetermined threshold.

[0131] Next, the present invention can optimize reconciliation rules by analyzing industry-specific misclassification patterns and adjusting confidence levels based on the modification details when an accounting expert modifies a selected account item (S50).

[0132] Here, the present invention, when optimizing a metabolic rule, generates a new rule if the correction frequency for the same transaction pattern is greater than or equal to a preset threshold value, operates the new rule as a candidate for a predetermined period to verify accuracy, and if the error rate is less than a preset threshold value after verification, promotes the new rule to an official rule.

[0133] In addition, when optimizing metabolic rules, the present invention analyzes whether rules learned in a specific industry appear similarly in other industries, and if confirmed as a common pattern, can be extended to common rules for all industries.

[0134] Figure 7 is a diagram illustrating the process of determining whether to perform metabolism for each data source based on the update time and sensitivity of the data source.

[0135] As illustrated in FIG. 7, the present invention can monitor the final update time for each data source used in automatic metabolism (S32).

[0136] Next, the present invention can calculate the sensitivity of each data source to accounting processing according to industry (S33).

[0137] In addition, the present invention can determine whether to perform metabolism for each data source based on the last update time and sensitivity (S34).

[0138] Next, the present invention can determine whether the sensitivity of a delayed data source at the time of metabolic execution is greater than or equal to a preset threshold based on the last update time and sensitivity (S35).

[0139] Next, the present invention may defer metabolic determination if the sensitivity of the delayed data source is greater than or equal to a preset threshold (S36), and determine to perform metabolic determination by generating a past pattern-based prediction value if the sensitivity of the delayed data source is less than the threshold (S37).

[0140] Figure 8 is a diagram illustrating the application of metabolite rules by industry.

[0141] As illustrated in FIG. 8, the present invention can calculate the filtering of candidate accounts and the degree of certainty according to industry-specific accounting standards based on transaction information (S42).

[0142] Here, the present invention identifies industry information regarding a business partner of a transaction history based on transaction information, and can filter candidate account items based on a combination of the industry of the target company and the industry of the business partner.

[0143] Next, the present invention can calculate an account candidate score by summing the product of the industry-specific weight and the confidence level for each account candidate (S44).

[0144] Next, the present invention can select an account with the highest score and apply it as an industry-specific reconciliation rule (S46).

[0145] Here, the present invention allows different account titles to be applied to the same transaction item depending on the industry when applying industry-specific reconciliation rules.

[0146] For example, in the present invention, if the transaction item is server hosting costs, it is classified as cost of goods sold if in the SaaS industry and as communication expenses if in the manufacturing industry; if the transaction item is shipping costs, it is classified as cost of goods sold if in the commerce industry and as selling expenses if in the manufacturing industry; and if the transaction item is software subscription fees, it can be classified as any one of employee welfare expenses, office supplies expenses, or current development expenses depending on the industry and use.

[0147] In addition, the present invention allows for differential application of capitalization and expense treatment standards by industry when applying industry-specific reconciliation rules.

[0148] For example, in the present invention, if it is a SaaS industry, the costs of the productization stage among technology development expenses are treated as intangible assets, and if it is a manufacturing industry, molds and equipment are treated as tangible assets and consumable tools are treated as expenses, and depreciation entries can be generated by applying the industry-specific standard useful life.

[0149] In addition, the present invention allows for differential application of the timing of sales and expenses by industry when applying industry-specific reconciliation rules.

[0150] For example, the present invention recognizes revenue based on the time of PG settlement confirmation in the case of commerce and marketplace industries, recognizes revenue after a predetermined period has elapsed after processing advance payments by allocating them over the service provision period in the case of SaaS and subscription service industries, recognizes revenue based on the time of tax invoice issuance in the case of manufacturing and wholesale / retail industries, and recognizes revenue based on the project progress rate or the time of achieving contract milestones in the case of professional services industries.

[0151] Next, the present invention can display a flag indicating that review is needed when applying industry-specific reconciliation rules if the difference in scores between an account with the highest score and an account with the second-highest score is less than a predetermined threshold.

[0152] In addition, the present invention can assign a reliability score to the final metabolic result when applying industry-specific metabolic rules.

[0153] Here, the present invention assigns high confidence if all essential data sources are up to date, industry weights are concentrated in a specific industry, and do not correspond to anomaly detection, automatically confirms the final metabolite result if the assigned confidence score is above a preset threshold, and requests a review of the final metabolite result if the assigned confidence score is below the threshold.

[0154] Furthermore, when optimizing a metabolic rule, the present invention generates a new rule if the correction frequency for the same transaction pattern is greater than or equal to a preset threshold value, operates the new rule as a candidate for a predetermined period to verify accuracy, and if the error rate is less than a preset threshold value after verification, the new rule can be promoted to an official rule.

[0155] Figure 9 is a diagram illustrating the metabolic rule optimization process through verification.

[0156] As illustrated in FIG. 9, the present invention can optimize the reconciliation rule through verification by checking the modification details of an accounting expert for an account item (S52), analyzing the misclassification pattern by industry based on the modification details if there is a modification to the account item (S54), and performing a confidence level adjustment of the account item (S56) (S58).

[0157] Here, the present invention collects details modified by an accountant or tax accountant after automatic reconciliation, classifies modification types such as account item changes, recognition timing adjustments, amount modifications, and business partner changes, extracts misclassification patterns that frequently occur in each industry to analyze misclassification patterns by industry, and can adjust the weights of industry-specific rules based on the misclassification patterns.

[0158] Meanwhile, the present invention can perform the industry identification process as follows.

[0159] First, the present invention can collect basic industry information, for example, by collecting the industry code of a business registration certificate to generate a primary industry signal.

[0160] Furthermore, the present invention can identify actual business models based on web / apps, for example, by crawling the website and mobile app of a target company to extract information such as type information of services / products provided (physical goods, digital content, subscription services, marketplaces, etc.), payment method information (one-time purchase, subscription / recurring payment, pay-as-you-go, project-based, etc.), and delivery / provision method information (direct delivery, third-party seller intermediation, online delivery, etc.), and can classify detailed business models based on the extracted information (for example, commerce, company-owned online store, open market, SaaS, manufacturing, professional services, bio, fintech, etc.).

[0161] Next, the present invention can extract industry signals based on transaction patterns by analyzing transaction data from the past N months (e.g., 6 months) to extract temporal distribution patterns of sales / purchases (monthly even distribution vs. concentration at specific points in time), distribution patterns of business partners (proportion of advertising agencies, cloud services, logistics companies, manufacturers, etc.), proportion patterns of payment methods (card, bank transfer, PG, cash receipt, etc.), average transaction amount and frequency patterns, and regular / recurring transaction ratio patterns, and can calculate similarity scores by industry based on each pattern.

[0162] In addition, the present invention can calculate weights for multiple industries, and can calculate a final weight for each industry by combining the derived industry signals.

[0163] For example, the present invention can assign a final weight to each industry, such as Commerce 0.6 and SaaS 0.4.

[0164] Here, the present invention can reflect changes in the business model by giving high weight to recent data.

[0165] In addition, the present invention can perform the data source management process as follows.

[0166] First, the present invention can define a set of data sources, and defines a set of data sources S = {card transaction, account deposit / withdrawal, tax invoice, PG settlement, cash receipt, electronic tax invoice, etc.} used in automatic reconciliation.

[0167] Furthermore, the present invention records the final update time for each data source by monitoring the update cycle for each source.

[0168] Here, the present invention learns the normal update cycle for each source (for example, card transactions once a day, PG settlement once a week) and can calculate the delay time of each source relative to the current time.

[0169] Next, the present invention can calculate source sensitivity by industry, and can calculate the importance (sensitivity) of each data source on accounting treatment for each industry.

[0170] The present invention, as a method for calculating sensitivity, analyzes the frequency in which each source contributed to the final judgment in the past history of the relevant industry, measures the increase in the misclassification rate when the source is omitted, reflects standard accounting practices by industry (for example, commerce is PG settlement, manufacturing is tax invoice-centered), and can normalize the sensitivity to a value between 0 and 1.

[0171] Furthermore, when performing a correction policy decision for a delay source, the present invention compares the delay time of each source and the industry-specific sensitivity at the time of metabolism.

[0172] Here, if a source with a sensitivity of 0.7 or higher is delayed, the source data can be classified as a required waiting item to withhold metabolic judgment and displayed in a 'waiting' state until the source is updated.

[0173] In addition, if a source with a sensitivity below the threshold is delayed, a predicted value is generated based on the pattern of past same period / similar transactions, and a provisional reconciliation is performed using the predicted value, but a 'correction applied' flag is displayed, and a reconciliation can be performed when the actual data is updated.

[0174] In addition, the present invention can perform the process of applying metabolic rules by industry as follows.

[0175] First, the present invention can receive transaction details.

[0176] Here, specific transaction details (transaction date and time, business partner, amount, description, data source, etc.) are received.

[0177] Next, the present invention can analyze the business sector of a client, identify the client's business sector based on the client's business registration number or trade name, and narrow down candidate account items by considering the combination of the company's business sector and the client's business sector.

[0178] For example, you can narrow down the candidate account categories to advertising expenses for Company (SaaS) + Client (Advertising), or transportation costs or cost of goods sold for Company (Commerce) + Client (Logistics).

[0179] Furthermore, the present invention can map account items by industry, and different account items can be applied to the same transaction item depending on the industry.

[0180] For example, server hosting costs can be applied as cost of goods sold in the SaaS industry and as communication expenses (administrative expenses) in general manufacturing, shipping costs can be applied as cost of goods sold in the commerce industry and as selling expenses in manufacturing (B2B), and software subscription fees can be applied as employee welfare expenses or office supplies in the SaaS industry, or as current development expenses in manufacturing (if applicable).

[0181] Next, the present invention can perform weighted blending of multiple industries by applying industry-specific weights and calculating account candidate and confidence scores for each industry, and the final account can be determined by the following formula.

[0182] For each account item, the score of the account item is calculated using the formula Score(account) = Σ(industry weight × account confidence in the corresponding industry), and the account item with the highest score is selected, and if the difference from the second-highest score is less than a threshold (e.g., 0.2), a 'review needed' flag can be displayed.

[0183] In addition, the present invention may apply capitalization / expense treatment standards, and first, industry-specific capitalization standards are applied.

[0184] Here, in SaaS / software development, the productization stage of technology development costs can be treated as an intangible asset; in manufacturing, molds and equipment can be treated as tangible assets and consumable tools as expenses; and in commerce, website construction costs can be treated as intangible assets or expenses according to industry-specific standards.

[0185] In addition, the present invention allows for the application of industry-specific standard useful lives and the selection of the straight-line method or the declining-balance method for each asset type as a depreciation policy.

[0186] Furthermore, the present invention can determine the timing of revenue / cost recognition, and different recognition timings can be applied depending on the industry.

[0187] For example, in commerce / marketplaces, the point of PG settlement confirmation (purchase confirmation after delivery or automatic confirmation) is applied; in SaaS / subscription services, the point of prorated recognition over the service provision period (revenue recognition after the period elapses following advance payment processing) is applied; in manufacturing / wholesale and retail, the point of tax invoice issuance (point of goods delivery) is applied; in professional services (consulting, development), the point based on project progress or the point of achieving contract milestones is applied; and in bio / research and development, the point of revenue recognition at each stage, such as down payments, milestone payments, and royalties, can be applied.

[0188] In addition, the present invention can automatically make journal entries when accounting period adjustment is required by comparing the date of transaction data with industry-specific recognition standards.

[0189] In addition, the present invention can perform a learning-based rule optimization process as follows.

[0190] First, the present invention collects details modified by an accountant or tax accountant after automatic reconciliation and can classify modification types such as changes in account names, adjustments in recognition timing, modifications in amounts, and changes in business partners.

[0191] In addition, the present invention can analyze misclassification patterns by industry by extracting misclassification patterns that frequently occur in each industry.

[0192] For example, in the SaaS industry, there is a case where 'cloud service costs' were classified as 'communication expenses' but were later corrected to 'cost of goods sold,' and in the commerce industry, there is confusion between 'marketing expenses' and 'advertising expenses.'

[0193] Next, the present invention can adjust the weights of industry-specific rules based on misclassification patterns.

[0194] Here, as an adjustment method, the certainty of a specific account is increased / decreased under specific conditions (customer, amount range, description keywords, etc.), and a new condition-account mapping rule is automatically generated. The generated rule can be verified as a 'candidate' for a certain period, and if the error rate is low, it can be promoted to a 'formal' rule.

[0195] Furthermore, the present invention allows a new pattern that does not belong to an existing industry classification to be detected and registered as a new industry candidate.

[0196] Here, the present invention can generate an independent rule set for the corresponding pattern when sufficient transaction history is accumulated (e.g., 100 or more).

[0197] For example, if an 'online education platform' is perceived as an intermediate form between existing 'SaaS' and 'professional services,' it can be classified as a separate industry.

[0198] In addition, the present invention can perform the time series anomaly detection linkage process as follows.

[0199] First, the present invention can learn monthly / quarterly normal trading patterns for each industry.

[0200] In other words, it is possible to learn monthly / quarterly normal transaction patterns for each industry, such as the temporal distribution of sales / expenses, amount ranges by major account category, seasonality, and recurring events (for example, year-end promotions in commerce or quarterly contract renewals in SaaS).

[0201] Next, the present invention can classify transactions satisfying the following conditions during the metabolic process as abnormal candidates.

[0202] Here, the abnormal candidate classification conditions may include a first condition in which the amount is at least three times the average (or at least two times the standard deviation) compared to the normal pattern by industry, a second condition in which an unexpected account item appears in the relevant month, and a third condition in which high-value transactions are made with new business partners.

[0203] In the case of an abnormal candidate transaction, the present invention can lower the metabolic reliability (e.g., 0.5 or less) and flag it as a target for automatic review.

[0204] Next, the present invention assigns a reliability score (0 to 1) to the final metabolic result.

[0205] Here, the present invention can assign a high score when all essential source data is up to date, when multiple industry weights are clearly one-sided, and when no anomaly is detected.

[0206] In addition, the present invention allows for automatic confirmation when the reliability is greater than or equal to a threshold value (e.g., 0.8), and may request a review when the reliability is less than the threshold value.

[0207] In addition, the present invention can perform a multi-corporate / business unit integration process as follows.

[0208] First, the present invention can identify transactions between corporations subject to metabolism.

[0209] Here, the present invention can identify transactions between corporations subject to reconciliation, such as fund transfers and service provision between corporations within the same group, when the counterparty is an affiliate.

[0210] The present invention can match the transaction details of both corporations when determined to be an internal transaction.

[0211] Next, the present invention can verify consistency by cross-applying each industry rule when the industries of both parties to a transaction are different.

[0212] For example, the present invention applies service provision → revenue in the case of Company A (manufacturing) and service receipt → expense in the case of Company B (commerce), and verifies whether the account items and amounts match symmetrically; in the event of a discrepancy, the judgment of the side with higher reliability may be prioritized, or a review flag may be displayed for both sides.

[0213] Furthermore, the present invention can automatically generate removal entries when preparing consolidated financial statements for cases confirmed as internal transactions.

[0214] The present invention can automatically generate consolidation adjustment journal entries, such as the removal of unrealized gains and losses and the offsetting of receivables and payables.

[0215] In this way, the present invention can perform automatic reconciliation of transaction records by multi-layeredly analyzing the industry characteristics of a company and differentially managing the reliability and time consistency of multiple financial data sources by industry.

[0216] In addition, the present invention can reduce the misclassification rate of automated reconciliation by improving the accuracy of industry identification through multi-layered analysis of not only business registration industry codes but also actual business models and transaction patterns.

[0217] In addition, the present invention enables accurate metabolization even for companies operating hybrid business models through a multi-industry weighted blending method.

[0218] In addition, the present invention can simultaneously ensure the accuracy and completeness of real-time metabolism through delay correction that considers the update cycle by data source and sensitivity by industry.

[0219] In addition, the present invention can accurately reflect industry-specific accounting practices and tax standards by applying different accounting standards depending on the industry to the same transaction.

[0220] In addition, the present invention can implement a self-generating system in which metabolic accuracy improves as the usage period increases by learning the modification history of accounting experts and continuously optimizing the rules.

[0221] In addition, the present invention can detect accounting fraud or errors at an early stage by automatically detecting abnormal transactions relative to normal patterns by industry.

[0222] In addition, the present invention can significantly improve the efficiency of group accounting operations by automatically identifying internal transactions and automating the journal entries for adjusting consolidated financial statements when operating multiple entities.

[0223] The method according to one embodiment of the present invention described above may be implemented as a program (or application) and stored on a medium to be executed in combination with a server, which is hardware.

[0224] The aforementioned program may include code encoded in a computer language such as C, C++, JAVA, or machine language, which can be read by the computer's processor (CPU) through the computer's device interface, in order for the computer to read the program and execute the methods implemented in the program. Such code may include functional code related to functions that define the necessary functions for executing the methods, and may include control code related to execution procedures necessary for the computer's processor to execute the functions according to a predetermined procedure. Additionally, such code may further include memory reference code regarding where (address) additional information or media necessary for the computer's processor to execute the functions should be referenced in the computer's internal or external memory. In addition, if the processor of the computer needs to communicate with any other computer or server located remotely in order to execute the above functions, the code may further include communication-related code regarding how to communicate with any other computer or server located remotely using the communication module of the computer, and what information or media to transmit or receive during communication.

[0225] The above-mentioned storage medium refers to a medium that stores data semi-permanently and is readable by a device, rather than a medium that stores data for a short period of time, such as a register, cache, or memory. Specifically, examples of the above-mentioned storage medium include, but are not limited to, ROM, RAM, CD-ROM, magnetic tape, floppy disk, and optical data storage device. That is, the above-mentioned program may be stored on various recording media on various servers that the computer can access, or on various recording media on the user's computer. Additionally, the above-mentioned medium may be distributed across networked computer systems, and computer-readable code may be stored in a distributed manner.

[0226] The steps of the method or algorithm described in connection with embodiments of the present invention may be implemented directly in hardware, implemented as a software module executed by hardware, or implemented by a combination thereof. The software module may reside in RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), Flash Memory, a hard disk, a removable disk, a CD-ROM, or any form of computer-readable recording medium well known in the art to which the present invention belongs.

[0227] Although embodiments of the present invention have been described above with reference to the attached drawings, those skilled in the art will understand that the present invention may be implemented in other specific forms without altering its technical concept or essential features. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive.

Claims

Claim 1 A communication module for obtaining the industry code and transaction information of a target company; a database for storing the industry code and transaction information of the target company; The system includes a processor that controls data source reconciliation processing according to industry-specific characteristics based on the industry code and transaction information of the target company. The processor calculates weights for each of multiple industries by synthesizing transaction patterns, including the business registration industry code of the target company obtained through the communication module, information on service types, payment methods, and provision methods extracted by crawling the target company's website and mobile app, and the temporal distribution of sales and purchases, the distribution of trading partners' industries, and the proportion of payment methods analyzed from transaction data of a predetermined past period. It monitors the final update time for each data source used for automatic reconciliation, calculates the sensitivity of each data source to accounting treatment as a value between 0 and 1 by analyzing the frequency with which each data source contributed to the final judgment in the past reconciliation history for each industry and measuring the increase in the misclassification rate when the corresponding data source is omitted, determines whether to perform reconciliation for each data source based on the final update time and the sensitivity, and performs automatic reconciliation for each data source based on the determination result. If there is a data source that is delayed at the time of reconciliation execution based on the final update time, and the sensitivity of the corresponding data source is above a preset threshold, it suspends the reconciliation judgment and the corresponding If the sensitivity of the data source is below the threshold, it is determined to perform provisional reconciliation by generating a prediction value based on past patterns, and according to the determined result, automatic reconciliation is performed for each data source, filtered into a pre-mapped group of account candidates based on the combination of the industry code of the counterparty included in the transaction information and the industry code of the target company, and for each account candidate, a score is calculated using the formula Score(Account) = Σ(Industry Weight × Account Confidence in the Corresponding Industry), the account with the highest score is selected and applied as an industry-specific reconciliation rule.An industry characteristic-based multi-data source automatic reconciliation device characterized by classifying modification types, including account change, recognition point adjustment, and amount modification, when an accounting expert modifies the selected account, and optimizing the reconciliation rule by extracting misclassification patterns occurring in each industry and adjusting the account certainty of the corresponding industry. Claim 2 delete Claim 3 delete Claim 4 delete Claim 5 An industry characteristic-based multi-data source automatic metabolism device according to claim 1, wherein, when determining whether to perform the metabolism, the processor, based on the last update time and the sensitivity, when there is a data source that is delayed at the time of performing the metabolism, if the sensitivity for the source is greater than or equal to a preset threshold, the metabolism determination is withheld, and if the sensitivity is less than the threshold, the processor determines to perform the metabolism by generating a past pattern-based prediction value. Claim 6 An industry characteristic-based multi-data source automatic reconciliation device according to claim 1, wherein, when filtering the account candidate, the processor identifies industry information regarding the business partner of the transaction history based on the transaction information, and filters into a group of account candidate corresponding to the combination by referring to a predefined mapping table according to the combination of the industry code of the target company and the industry code of the business partner. Claim 7 An industry-specific multi-data source automatic reconciliation device according to claim 1, characterized in that, when applying the industry-specific reconciliation rule, the processor displays a flag indicating that review is needed if the score difference between the account with the highest score and the account with the second-highest score is less than a predetermined threshold. Claim 8 An industry characteristic-based multi-data source automatic reconciliation device according to claim 1, wherein the processor, when optimizing the reconciliation rule, generates a new rule if the correction frequency for the same transaction pattern is greater than or equal to a preset threshold value, verifies the accuracy by operating the new rule as a candidate state for a predetermined period, and after the verification, if the error rate calculated as the ratio of the number of cases corrected by an accounting expert among the reconciliation results to which the rule was applied during the candidate state operation period is less than a preset threshold value, the new rule is promoted to an official rule. Claim 9 An industry-specific multi-data source automatic metabolization device according to claim 1, characterized in that, when optimizing the metabolization rule, the processor analyzes whether a rule learned in a specific industry appears similarly in other industries, and if confirmed as a common pattern, expands it into a rule common to all industries. Claim 10 A method for automatically reconciling multiple data sources based on industry characteristics, performed by a device, comprising: a step of obtaining the industry code and transaction information of a target company; a step of calculating weights for each of a plurality of industries by synthesizing transaction patterns including the business registration industry code of the target company, information on service types, payment methods, and provision methods extracted by crawling the target company's website and mobile app, and the temporal distribution of sales and purchases, the distribution of business partners' industries, and the proportion of payment methods analyzed from transaction data of a predetermined past period; a step of monitoring the final update time for each data source used for automatic reconciliation, and calculating the sensitivity of each data source to accounting treatment as a value between 0 and 1 by analyzing the frequency with which each data source contributed to the final judgment in the past reconciliation history for each industry and measuring the increase in the misclassification rate when the corresponding data source is omitted; a step of determining whether to perform reconciliation for each data source based on the final update time and the sensitivity, and performing automatic reconciliation for each data source based on the determination result; if there is a data source that is delayed at the time of reconciliation execution based on the final update time, if the sensitivity of the corresponding data source is above a preset threshold, the reconciliation judgment is withheld, and the sensitivity of the corresponding data source is above A step of determining to perform provisional reconciliation by generating a past pattern-based prediction value if below a threshold, and performing automatic reconciliation by data source according to the determined result; a step of filtering into a pre-mapped group of account candidates based on the combination of the industry code of the counterparty included in the transaction information and the industry code of the target company, and for each account candidate, calculating a score using the formula Score(Account) = Σ(Industry Weight × Account Confidence in the corresponding industry), selecting the account with the highest score, and applying it as an industry-specific reconciliation rule;An industry characteristic-based multi-data source automatic reconciliation method characterized by including the step of classifying modification types, including account change, recognition time adjustment, and amount modification, when an accounting expert modifies the selected account, and optimizing the reconciliation rule by extracting misclassification patterns occurring in each industry and adjusting the account certainty of the corresponding industry.

Images