A business partner recommendation device and method for recommending B2B business partners based on reverse engineering of a supply chain using artificial intelligence

Abstract

The present invention relates to a business partner recommendation system, device, and method for recommending B2B business partners based on supply chain reverse engineering using artificial intelligence, wherein the business partner recommendation system may include a user terminal for inputting a business registration number; and a business partner recommendation device that receives a business registration number from the user terminal and recommends a business partner through artificial intelligence-based matching based on the business registration number.

Description

Technology Field

[0001] The present invention relates to a business partner recommendation device and method for recommending B2B (Business to Business) business partners based on supply chain reverse engineering using artificial intelligence (AI). Background Technology

[0002] One of the most fundamental problems in the sales activities of domestic small and medium-sized enterprises (SMEs) is the lack of valid client information. Sales representatives lack a database of potential customers and rely on personal networks or vague guesswork to conduct sales; consequently, the success rate of cold calling without data targeting is less than 1%. Furthermore, the average annual fixed cost per salesperson is approximately 60 million won, resulting in a significant financial burden due to inefficient sales practices.

[0003] Furthermore, it is difficult for companies to identify who their competitors do business with or what supply chain structures top firms in the same industry have established. Since simple financial information alone cannot grasp actual supply relationships or transaction contexts, a problem of supply chain invisibility arises, leading to missed market opportunities.

[0004] Furthermore, due to the lack of a system to verify the financial soundness or credit status of trading partners in advance, companies are constantly exposed to verification risks, such as the occurrence of bad debts and repeated delivery delays.

[0005] Conventional methods for identifying B2B clients have included online searches, participation in trade shows, and referrals from acquaintances; however, these methods have limitations as they make systematic, data-driven targeting difficult and rely on unverified information.

[0006] Consequently, there is a growing need for a system that combines highly reliable, financial-grade corporate data with artificial intelligence technology to instantly reverse-engineer a supply chain using only a business registration number and recommend optimal B2B partners in real time.

[0007] The technology forming the background of the present invention is disclosed in Korean Registered Patent No. 10-1671890. The problem to be solved

[0008] The present invention aims to solve the problems of the aforementioned prior art by providing a supplier recommendation system, device, and method that recommends B2B suppliers based on supply chain reverse engineering using artificial intelligence. This is achieved by allowing a user to input only a single business registration number without a separate, complex registration process, and by linking in real-time with an external corporate database possessing financial sector-level reliability to reverse engineer the supply chain of the relevant company, and by analyzing the successful transaction patterns of leading companies in the same industry using artificial intelligence to automatically recommend optimal suppliers and sellers.

[0009] In addition, the present invention aims to provide intuitive recommendation results that enable sales representatives to make decisions within a short period of time by calculating scores and automatically ranking candidates for business partners based on multiple evaluation items such as industry suitability, financial soundness, transaction frequency, accessibility, and market timing.

[0010] In addition, the present invention aims to provide a revenue model that distinguishes and charges information disclosure and transaction progress in stages as interaction with business partners progresses, and to provide differentiated services according to multiple subscription grades.

[0011] However, the technical problems that the embodiments of the present invention aim to solve are not limited to the technical problems described above, and other technical problems may exist. means of solving the problem

[0012] As a technical means for achieving the above-mentioned technical task, a business partner recommendation system that recommends B2B business partners based on supply chain reverse engineering using artificial intelligence according to an embodiment of the present invention may include: a user terminal for inputting a business registration number; and a business partner recommendation device that receives a business registration number from the user terminal and recommends business partners through artificial intelligence-based matching based on the business registration number.

[0013] Additionally, the business partner recommendation device may include: an input receiving unit that receives a business registration number from the user terminal; a business information retrieval unit that queries business information data of a target company corresponding to the business registration number by linking with an API of an external corporate database; a supply chain reverse engineering unit that reverse engineers a supply chain network centered on the target company by analyzing actual transaction data included in the business information data; and a recommendation providing unit that generates recommended business partner information based on a supply chain graph corresponding to the supply chain network and provides it to the user terminal.

[0014] Additionally, the recommendation providing unit may include: a filtering unit that generates a candidate group of business partners by considering the supply chain graph and refines the candidate group by applying a filter to the candidate group; a score calculation unit that calculates scores for each candidate company in the refined candidate group based on a plurality of evaluation items; and a ranking calculation unit that ranks the top N business partners in advance based on the scores for the plurality of evaluation items.

[0015] Additionally, the recommendation providing unit may further include a crawling unit that collects unstructured data related to companies included in the supply chain graph, extracts valid information using a pre-set industry-specific keyword dictionary, and transmits it to the filtering unit, and a patterning unit that analyzes the supply chain structure patterns of top companies within the same industry as the target company using transaction history data of blue-chip companies collected from the supply chain graph and the external company database.

[0016] The above-described means for solving the problem are merely exemplary and should not be interpreted as intended to limit the invention. In addition to the exemplary embodiments described above, additional embodiments may exist in the drawings and the detailed description of the invention. Effects of the invention

[0017] According to the means for solving the problem of the present invention described above, the present invention provides a business partner recommendation system, device, and method that recommends B2B business partners based on supply chain reverse engineering using artificial intelligence, thereby enabling users to immediately use corporate analysis and business partner recommendation services by entering only a business registration number without a separate membership registration procedure or the submission of complex documents.

[0018] In addition, by providing the above-described business partner recommendation system, device, and method, the present invention can significantly improve the accuracy and integrity of data compared to existing methods relying on web crawling through real-time linkage with an external corporate database having financial sector-level reliability.

[0019] In addition, the present invention provides the above-described business partner recommendation system, device, and method, wherein the ranking algorithm comprehensively considers multiple evaluation items such as KSIC industry mapping rules, financial soundness, transaction frequency, accessibility, and timing, thereby significantly improving the matching hit rate in which actual meetings or transaction requests occur within the top candidate group of recommended business partners, and can provide results that are significantly improved compared to the success rate of the existing cold call method.

[0020] In addition, the present invention, in providing the above-mentioned business partner recommendation system, device, and method, provides a billing model for each transaction stage and a differential service based on subscription grade, thereby providing an environment in which users of various sizes and needs can flexibly utilize the service.

[0021] However, the effects obtainable from the present invention are not limited to those described above, and other effects may exist. Brief explanation of the drawing

[0022] FIG. 1 is a diagram showing the schematic configuration of a supplier recommendation system that recommends B2B suppliers based on supply chain reverse engineering using artificial intelligence according to one embodiment of the present invention. FIGS. 2 to 4 are drawings for explaining a business partner recommendation device that recommends B2B business partners based on supply chain reverse engineering using artificial intelligence according to an embodiment of the present invention. FIG. 5 is a flowchart of the operation of a business partner recommendation method that recommends B2B business partners based on supply chain reverse engineering using artificial intelligence according to one embodiment of the present invention. Specific details for implementing the invention

[0023] Embodiments of the present invention are described below with reference to the attached drawings so that those skilled in the art can easily implement the invention. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein. Furthermore, in order to clearly explain the present invention in the drawings, parts unrelated to the explanation have been omitted, and similar parts throughout the specification are denoted by similar reference numerals.

[0024] Throughout the specification of the present invention, when a part is described as being "connected" to another part, this includes not only cases where they are "directly connected," but also cases where they are "electrically connected" or "indirectly connected" with other elements interposed between them.

[0025] Throughout the specification of the present invention, when a member is described as being located "on," "on the upper," "on the top," "under," "on the lower," or "on the bottom" of another member, this includes not only cases where the member is in contact with the other member, but also cases where another member exists between the two members.

[0026] Throughout the specification of the present invention, when a part is described as "comprising" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components.

[0027] Throughout the specification of the present invention, some of the operations or functions described as being performed by a terminal, device, or device may instead be performed by a server connected to said terminal, device, or device. Likewise, some of the operations or functions described as being performed by a server may also be performed by a terminal, device, or device connected to said server.

[0028] Throughout the specification of the present invention, the term "at least one" may be defined as a term including both singular and plural forms, and it will be obvious that even if the term "at least one" is not present, each component may exist in a singular or plural form and may mean singular or plural. Furthermore, whether each component is provided in a singular or plural form may be changed according to the embodiment.

[0029] FIG. 1 is a diagram showing the schematic configuration of a business partner recommendation system (1000) that recommends B2B business partners based on supply chain reverse engineering using artificial intelligence according to an embodiment of the present invention. FIG. 2 to 4 are diagrams for explaining a business partner recommendation device that recommends B2B business partners based on supply chain reverse engineering using artificial intelligence according to an embodiment of the present invention.

[0030] For convenience of explanation, the customer recommendation system (1000) that recommends B2B business partners based on supply chain reverse engineering using artificial intelligence according to one embodiment of the present invention will be referred to as the present system (1000). In addition, the details shown (described) in the drawings of the present invention, including FIG. 1, may be applied equally to the description of the present system (1000) or the customer recommendation device (10), even if such details are omitted below.

[0031] Referring to FIGS. 1 to 4, the system (1000) may include a business partner recommendation device (100), an administrator terminal (200), a user terminal (300), and an external corporate database (400).

[0032] The business partner recommendation device (100) is a business partner recommendation device that recommends B2B business partners based on supply chain reverse engineering using artificial intelligence according to one embodiment of the present invention, and for convenience of explanation, it will be referred to as the device (100) below. The device (100) can provide screens of the service as shown in FIGS. 2 to 4 to a user terminal (300).

[0033] The device (100) may be a device or server that provides at least one of a webpage (homepage), program, application (app), service, and platform related to the above-mentioned business partner recommendation (in particular, a business partner recommendation that recommends B2B business partners based on supply chain reverse engineering using artificial intelligence). At this time, the webpage (homepage), program, application, service, and platform related to the above-mentioned business partner recommendation provided by the device (10) may each be referred to as the webpage (this homepage), the program, the app, the service, and the platform, respectively, for convenience of explanation below. In the example illustrated in FIG. 1, the device (100) is described as being provided in the form of a server capable of transmitting and receiving data through a network (5) with each of the components within the system (1000).

[0034] The device (100) receives a business registration number from a user terminal (300) and can recommend a business partner through artificial intelligence-based matching based on the business registration number. In particular, the device (100) receives a business registration number from a user terminal (300), reverse-engineers the supply chain structure by linking with an external corporate database (400), and performs artificial intelligence-based matching to recommend an optimal business partner to the user terminal (300).

[0035] Users of the device (100) may include users and administrators. Users may use the service by accessing the program, the app, the website, etc., using a terminal they possess (especially accessing after installation or without installation), and may also use the service without registering (logging in).

[0036] The administrator terminal (200) may refer to a terminal possessed by the administrator. The administrator may be a person who develops and distributes the device (100) and operates and manages the device (100). The administrator terminal (200) can manage the operation settings of the device (100) and can monitor and manage the operation status of the device (100).

[0037] The administrator terminal (200) can perform functions such as registering, modifying, and deleting key setting values ​​of the system (1000) (or the device (100)), such as, for example, a KSIC mapping rule table, a keyword dictionary by industry, and weights by evaluation item, and managing user accounts and subscription grades.

[0038] The user terminal (300) may refer to a terminal possessed by a user who wishes to use the service (B2B business partner discovery service) provided by the device (100). The user may be a person who wishes to receive recommendations for business partners by entering a business registration number. The user may include, for example, a sales representative, a purchasing manager, or a company representative, but is not limited thereto and may apply to anyone who wishes to receive recommendations for business partners.

[0039] The user terminal (300) can access the app, etc., input a business registration number, receive recommended business partner information from the device (100), and display it on the screen. In the present invention, the business registration number may be referred to by other terms such as business registration number.

[0040] In FIG. 1, the case in which one user terminal (300) is included within the system (1000) is described as an example, but this is merely an example to aid in understanding the invention and is not limited thereto. Preferably, the system (1000) may include multiple user terminals possessed by each of the multiple users, and in describing the invention below, the description of the user terminal (300) may be applied equally to the description of each of the multiple user terminals, even if the description is omitted below.

[0041] The external corporate database (400) is a database operated by a credible corporate credit rating agency (e.g., Korea Evaluation Data (KODATA)) that has reliability at the level of the financial sector, and may include basic corporate information, financial information, actual transaction data of sales outlets and purchase outlets (actual transaction data of sales outlets / purchases), credit information, default history, and transaction history data of sound companies for multiple companies. The external corporate database (400) may hold (store) data of more than 12 million companies, for example, and may be configured to enable real-time API calls to this data.

[0042] In the present invention, the terminal (i.e., the administrator terminal (200) and the user terminal (300) each) may include, for example, all types of wired and wireless communication devices such as PCS (Personal Communication System), GSM (Global System for Mobile communication), PDC (Personal Digital Cellular), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (WCode Division Multiple Access), Wibro (Wireless Broadband Internet) terminal, smartphone, smartpad, tablet PC, laptop, wearable device, desktop PC, etc., but is not limited thereto.

[0043] In addition, the device (10) can transmit and receive data by being linked with each of the administrator terminal (200), user terminal (300), and external corporate database (400) via a network (5), and can control the operation (e.g., screen display operation) of each of the administrator terminal (200) and user terminal (300).

[0044] Here, the network (5) may include, for example, a 3GPP (3rd Generation Partnership Project) network, an LTE (Long Term Evolution) network, a WIMAX (World Interoperability for Microwave Access) network, the Internet, a LAN (Local Area Network), a Wireless LAN (Wireless Local Area Network), a WAN (Wide Area Network), a PAN (Personal Area Network), a Bluetooth network, a NFC (Near Field Communication) network, a satellite broadcasting network, an analog broadcasting network, a DMB (Digital Multimedia Broadcasting) network, but is not limited thereto and may include various wired / wireless communication networks. A more detailed description of the device (100) is as follows.

[0045] Referring again to FIG. 1, the device (100) may include an input receiving unit (11), a corporate information inquiry unit (12), a supply chain reverse engineering unit (13), a recommendation providing unit (14), a risk monitoring unit (15), a step-by-step billing unit (16), and a membership management unit (17).

[0046] <Input receiver (11)>

[0047] The input receiving unit (11) can receive a Business Registration Number from the user terminal (300).

[0048] The input receiving unit (11) can provide a so-called 'Zero-Barrier' type input interface so that, for example, the corporate analysis process can be initiated immediately by simply entering a business registration number without a separate membership registration procedure or login process. At this time, the input receiving unit (11) provides a business registration number input field on the screen of the user terminal (300), and when the user enters a 10-digit business registration number, the number can be immediately transmitted to the corporate information lookup unit (12).

[0049] The input receiving unit (11) can perform a validation function to verify the validity of the input business registration number (e.g., number of digits, whether it is a number, National Tax Service checksum, etc.) and can transmit the business registration number whose validity has been confirmed (verified) to the corporate information inquiry unit (12).

[0050] In addition, the input receiving unit (11) can selectively receive additional filtering conditions from the user terminal (300), such as the business registration number, the industry category desired by the user, the region desired for transaction, and the company size condition.

[0051] <Company Information Inquiry (12)>

[0052] The corporate information lookup unit (12) can query corporate information data of a target company (user's company) corresponding to the business registration number received from the input receiving unit (11) in real time by linking with the API of the external corporate database (400). The corporate information lookup unit (12) can query and obtain the corporate information data of the target company by making an API call to the external corporate database (400). The company corresponding to the business registration number may be referred to by other terms such as target company, inquiry target company, etc.

[0053] Here, the aforementioned corporate information data may include the target company's basic corporate information, financial information, credit information, and transaction history data. The aforementioned transaction history data may include actual transaction data related to sellers and buyers (i.e., actual transaction data from sellers / buyers). In this case, the actual transaction data refers to data regarding actual transactions with sellers / buyers, and may include, for example, transaction volume, frequency, etc.

[0054] For example, the above-mentioned basic corporate information may include information such as the company name, representative, date of establishment, industry code, and location. The above-mentioned financial information may include information on financial statement items such as sales revenue, operating profit, assets, liabilities, and capital. The above-mentioned credit information may include information such as credit rating, credit score, delinquency history, and default history. The above-mentioned actual transaction data may include information such as a list of sellers, a list of buyers, transaction volume, and transaction frequency.

[0055] The corporate information lookup unit (12) can parse the data (corporate information data) obtained (received) from an external corporate database (400) to fit an internal data format and transmit it to the supply chain reverse engineering unit (13). At this time, the corporate information lookup unit (12) can improve the response speed when repeatedly looking up the same business number by caching the lookup results for a certain period.

[0056] That is, the corporate information lookup unit (12) can call the API of an external corporate database (400) in real time using the business registration number received from the input receiving unit (11), receive corporate information data corresponding to the business registration number, and transmit it to the supply chain reverse engineering unit (13).

[0057] Supply Chain Reverse Engineering Department (13)

[0058] The supply chain reverse engineering unit (13) can reverse engineer a supply chain network centered on a target company by analyzing the actual transaction data included in the company information data retrieved from the company information inquiry unit (12).

[0059] At this time, reverse engineering performed by the supply chain reverse engineering unit (13) may refer to a series of data processing processes in which, in the absence of a company's entire supply chain blueprint, fragmentary actual transaction data and financial data are used as input values ​​to extract identification information and transaction attributes of trading partners, and based on this, multi-layered connection relationships between companies are traced backward from the lower level to the upper level to restore (reconstruct) the entire network structure.

[0060] The above supply chain network may refer to a set of multi-layered transaction relationships formed around a target company, and may be referred to by other terms such as a transaction relationship network, the target company's supply chain structure, or a multi-layered supply chain network.

[0061] The supply chain reverse engineering unit (13) can design (configure, create) the supply chain network by analyzing the actual transaction data to identify primary trading companies that have a direct trading relationship with the target company, and further by tracking and identifying secondary trading companies that have a trading relationship with the primary trading companies.

[0062] Afterward, the supply chain reverse engineering unit (13) can extract transaction relationship attribute information such as transaction direction (purchase direction or sales direction), transaction volume, transaction frequency, and transaction duration within the designed supply chain network.

[0063] Subsequently, the supply chain reverse engineering unit (13) can model the extracted transaction relationship attribute information into a graph structure to generate a supply chain graph in which each company is a node and transaction relationships are edges. That is, by performing the above modeling, the supply chain reverse engineering unit (13) can generate a supply chain graph consisting of each company node and transaction relationship edge.

[0064] That is, the supply chain reverse engineering unit (13) can analyze corporate information data (especially actual transaction data) received from the corporate information inquiry unit (12), reverse engineer the supply chain network, and generate a supply chain graph composed of multi-layered transaction relationships. The supply chain reverse engineering unit (13) can transmit the supply chain graph generated based on the designed supply chain network to the patterning unit (143) and the filtering unit (142).

[0065] The device (100) can provide a supply chain reverse engineering unit (13) to identify hidden transaction relationships between companies that are not publicly disclosed, based on actual transactions, thereby providing the user with deep supply chain visibility that existing simple directory-type databases cannot provide.

[0066] <Recommendation Provider (14)>

[0067] The recommendation providing unit (14) can generate recommended business partner information based on a supply chain graph corresponding to the supply chain network provided by the supply chain reverse engineering unit (13) (i.e., a supply chain graph generated based on the designed supply chain network) and provide it to the user terminal (300). At this time, the recommended business partner information may refer to information regarding a customized recommended business partner that is customized and recommended to the user based on the business registration number entered by the user.

[0068] Specifically, the recommendation providing unit (14) may include, for example, a crawling unit (141), a filtering unit (142), a patterning unit (143), a score calculation unit (144), and a ranking calculation unit (145) to generate the recommended business partner information.

[0069] <Crawling section (141)>

[0070] The crawling unit (141) can automatically collect (web crawl) unstructured data related to the companies (hereinafter referred to as 'related companies') from an external network (Internet) based on information about the companies included in the supply chain graph generated by the supply chain reverse engineering unit (13). Subsequently, the crawling unit (141) can extract valid information from the collected data (unstructured data) using an industry-specific keyword dictionary pre-configured (defined) by an administrator and transmit it to the filtering unit (142) and the score calculation unit (144).

[0071] Here, the valid information may include specific item names, industry-related keywords, corresponding industry classification codes (KSIC codes), keyword occurrence frequencies, expressions related to a company's business expansion or the discovery of new business partners, etc. That is, the valid information may be structured data (structured data, formalized data) including industry classification code mapping information, item identification information, transaction possibility signal information, and supply chain change signal information.

[0072] At this time, when the crawling unit (141) collects unstructured data related to the above-mentioned related companies, it may consider (use) the official website of the company (related company), social network service (SNS), news articles, industry specialized media, public disclosure data, patent information, etc., as sources to be collected.

[0073] The crawling unit (141) can apply natural language processing (NLP) technology to the collected unstructured data (unstructured text data of the unstructured data) to extract information such as the main products handled by the company (related company), recent trends, the need to discover new business partners, and signals of supply chain changes.

[0074] Specifically, the crawling unit (141) can refer to the above-mentioned industry-specific keyword dictionary and, if a specific item name or industry-related keyword (e.g., "steel plate / coil / heavy plate") is included in the collected unstructured data (unstructured text data), map a corresponding industry classification code (KSIC code) based on the keyword and reflect this in the calculation of a score for a candidate company (a candidate for a recommended business partner).

[0075] The above-mentioned industry-specific keyword dictionary is a text analysis asset that defines the correspondence between a specific item name or related keyword and a KSIC industry code, and can be used to analyze unstructured data (unstructured text data) in the crawling unit (141) and the score calculation unit (144) to map industry codes and add scores. For example, if keywords such as "steel plate," "coil," or "heavy plate" are included in the text, the company may be given a score as a candidate for a C24 (primary metal manufacturing) related purchaser by the score calculation unit (144).

[0076] The crawling unit (141) can standardize the collected unstructured data and transmit it to the filtering unit (142) and the score calculation unit (144).

[0077] <Filtering section (142)>

[0078] The filtering unit (142) can generate a candidate group of business partners (hereinafter referred to as 'candidate group') by considering the supply chain graph generated by the supply chain reverse engineering unit (13). In particular, the filtering unit (142) can generate the candidate group based on data received from the supply chain reverse engineering unit (13) (i.e., the supply chain graph) and data received from the crawling unit (141) (structured valid information). The candidate group (candidate group of business partners) may refer to a set of candidates for recommended business partners (business partner candidates) recommended to the user.

[0079] Subsequently, the filtering unit (142) can refine the candidate group (i.e., remove unsuitable candidates, which are unsuitable candidates within the candidate group) by applying a filter to the generated candidate group. Here, the filter may include at least one of an industry code filter, a region filter, and a scale filter. For example, the filtering unit (142) can remove unsuitable candidates within the candidate group by sequentially applying an industry code filter, a region filter, and a scale filter to the candidate group, and can transmit the refined candidate group, from which unsuitable candidates have been removed, to the paneling unit (143) and the score calculation unit (144).

[0080] [Industry Code Filter]

[0081] The above industry code filter may be a filter designed to select a candidate group by reflecting transaction relationships between industry codes based on the Korean Standard Industrial Classification (KSIC).

[0082] Specifically, the above industry code filter can select only companies corresponding to industry codes that have a transactional relationship with the target company's industry code (KSIC industry code) as a candidate group by referring to the KSIC mapping rule table.

[0083] Here, the KSIC mapping rule table is a data structure that defines transaction relationships between industry codes, and may be a table that defines, in the form of rules, a list of industry codes that are preferentially traded as buyers and a list of industry codes that are preferentially traded as sellers for a specific industry code. That is, the KSIC mapping rule table is a unique data asset that defines transaction relationships between KSIC industry codes, and may store a list of industry codes that are preferentially classified as buyers and a list of industry codes that are preferentially classified as sellers for each industry code.

[0084] The above KSIC mapping rule table may include a table structure that stores a list of priority industry codes for purchasing locations and a list of priority industry codes for selling locations for each KSIC industry code, and the industry code filter may remove unsuitable candidates by determining whether the industry code of a company within the candidate group is included in the list (list of priority industry codes for purchasing locations and list of priority industry codes for selling locations) by referring to the above table.

[0085] For example, if the KSIC industry code of the target company is C25 (manufacturing of metal processing products), C24 (manufacturing of primary metals) and G46 (wholesale of industrial raw materials) may be mapped as priority industry codes for purchasing companies according to the KSIC mapping rule table, and C28 (manufacturing of machinery and equipment), C29 (manufacturing of automobiles and trailers), etc. may be mapped as priority industry codes for selling companies. In this case, when the filtering unit (142) applies the industry code filter as the filter to the candidate group, it selects (classifies) companies corresponding to the mapped industry code (i.e., priority industry codes for purchasing companies / selling companies mapped to the industry code of the target company) among the candidate group as priority selection (classification) targets to be maintained within the candidate group, and can determine companies corresponding to unmapped industry codes as unsuitable candidates and remove them.

[0086] [Region Filter]

[0087] The above-mentioned regional filter may be a filter designed to select (filter) a group of candidates based on a logistics radius or zone considering logistics costs and delivery efficiency. The above-mentioned regional filter may be designed to prioritize the selection of companies located within a certain radius or in the same / adjacent zone based on the location of the target company, and if the user separately specifies a desired region, filtering may be performed by applying the corresponding regional conditions.

[0088] [Scale Filter]

[0089] The above-mentioned size filter may be a filter designed to select (filter) a candidate group by applying conditions regarding sales revenue range and employee count range in order to select companies of a size that are highly likely to transact with the target company. By excluding (removing) from the candidate group companies that are excessively large beyond a predefined maximum allowable size or excessively small below a predefined minimum allowable size relative to the size of the target company, the above-mentioned size filter can ensure that only candidates with a high likelihood of actual transaction are selected and maintained within the candidate group.

[0090] Meanwhile, the filtering unit (142) can dynamically generate or expand the candidate group by identifying potential candidate companies that are not explicitly revealed on the supply chain graph based on the valid information collected, structured, and transmitted from the crawling unit (141).

[0091] Specifically, the filtering unit (142) analyzes the item identification information and industry-related keywords included in the valid information to identify the actual business activities that the target company is actually conducting in addition to the industry code (KSIC) officially registered by the target company, and searches for companies possessing key keywords in the upstream and downstream supply chains related to the actual business from an external network to additionally include them in the candidate group.

[0092] Additionally, the filtering unit (142) can analyze text data related to the discovery of new business partners, business expansion announcements, or new facility investments among the valid information, classify companies that have expressed an intention to trade externally at the current time as 'high-ranking active candidates,' and generate the candidate group based on this.

[0093] In addition, the filtering unit (142) can analyze supply chain change signals included in the valid information (e.g., news of supply suspension by a specific competitor, change in raw material supplier, etc.) to trace back companies with the technological capabilities and portfolios that can replace gaps in the target company's existing supply chain and include them in the candidate group.

[0094] Additionally, the filtering unit (142) can perform 'cross-validation filtering' using the valid information. That is, the filtering unit (142) can calculate the data matching rate between the structured corporate information obtained from the external corporate database (400) and the actual business activity data (website content, recent press releases, etc.) extracted from the crawling unit (141), and can improve the reliability of the candidate group by determining that a company with a matching rate below a preset threshold is an unsuitable candidate and removing it from the candidate group.

[0095] In this way, the filtering unit (142) can remove unsuitable candidates within the candidate group by applying an industry code filter, a region filter, and a size filter to the candidate group, and by applying cross-validation filtering using valid information, and can transmit the refined candidate group (refined business partner candidate group) from which unsuitable candidates have been removed to the filtering unit (143) and the score calculation unit (144).

[0096] <Patterning section (143)>

[0097] The patterning unit (143) can analyze the supply chain structure patterns of top companies within the same industry as the target company by using the supply chain graph received from the supply chain reverse engineering unit (13) and the company information inquiry unit (12) collected (obtained) from the external company database (400).

[0098] At this time, the above ‘good company’ may refer to a company among those stored in an external company database (400) for example, whose financial statements for a recently set period (for example, the last three business years) have been verified through disclosure or external audit, whose credit rating is above a certain standard, and which has no history of bankruptcy or closure.

[0099] In addition, the above 'top companies within the same industry' may refer to, for example, companies that fall within the top pre-set ratio (e.g., top 1% to 5%) based on sales revenue, operating profit margin, or market share among a set of companies belonging to the same KSIC industry code as the target company. The selection criteria, applicable indicators, and ratios of the top companies may be set through the administrator terminal (200).

[0100] The patterning unit (143) can extract transaction relationship attribute information, including the industry code, transaction volume, transaction frequency, and transaction duration of purchasing and selling locations newly secured by top companies in the same industry at the time of growth, by using transaction history data (e.g., actual transaction data and financial data) of a high-quality company received from an external corporate database (400). Here, the 'time of growth' may refer to a business year in which the sales revenue of the company has increased by more than a predetermined growth rate (e.g., 15% or more) compared to the previous year.

[0101] That is, the patterning unit (143) can identify a business year in which the sales growth rate compared to the previous year exceeds a pre-set growth rate (e.g., 15% or more) by analyzing the sales fluctuation data of each company for the business year among the top companies in the same industry among the above-mentioned excellent companies. Subsequently, the patterning unit (143) can identify purchasing and selling partners that are newly formed or have increased transaction volume at the growth time by comparing transaction history data before and after the growth time, and can extract transaction relationship attribute information including the industry code, transaction volume, transaction frequency, transaction duration, and location in the supply chain of the corresponding trading partner.

[0102] Subsequently, the patterning unit (143) can convert the extracted transaction relationship attribute information into the form of a supply chain graph composed of nodes and edges, wherein each company node is represented as a feature vector including attributes such as industry code, sales volume, and number of employees, and the transaction relationship between companies can be represented as an edge having attributes such as transaction volume, transaction frequency, and transaction period. After converting the extracted transaction relationship attribute information into the form of a supply chain graph, the patterning unit (143) can learn a common supply chain structure pattern using an artificial intelligence model. At this time, the artificial intelligence model may be a graph-based model that represents company nodes as feature vectors and represents the transaction relationship between companies as edges.

[0103] For example, the patterning unit (143) can use a Graph Neural Network (GNN) to perform learning using the supply chain graph of the above-mentioned leading companies as input data, and can derive supply chain expansion paths, types of new business partners, and changes in transaction density that commonly appear at the time of growth in the form of pattern vectors. That is, the patterning unit (143) can use a Graph Neural Network to input the above-mentioned supply chain graph as training data and derive pattern vectors that reflect supply chain expansion paths, industry distribution of new business partners, changes in transaction density, and characteristics of supply chain stage movement that commonly appear at the time of growth.

[0104] Subsequently, the patterning unit (143) can identify a group of similar companies that are similar to the target company in terms of size, industry, and growth stage by comparing the pattern vector with the corporate characteristic information, including the industry code, sales volume, number of employees, and recent sales growth rate of the target company received from the corporate information inquiry unit (12). The comparison can be performed by calculating the distance between feature vectors or by a similarity calculation method.

[0105] Subsequently, the patterning unit (143) can apply the derived pattern vector to the refined candidate group (candidate business partner group) received from the filtering unit (142) to evaluate how well the supply chain attributes of each candidate company within the refined candidate group match the growth timing characteristics included in the pattern vector (degree of correspondence). In other words, the patterning unit (143) can calculate the degree of correspondence by comparing the attribute values ​​of each candidate company based on the industry distribution, transaction type, and supply chain location characteristics included in the derived pattern vector for the refined candidate group received from the filtering unit (142).

[0106] Specifically, the patterning unit (143) can determine for each candidate company included in the refined candidate group i) whether the industry code of the candidate company is included in the industry code group that frequently appears at the growth time, ii) whether the transaction type of the candidate company matches the transaction type formed at the growth time or is greater than or equal to a preset similarity standard, and iii) whether the stage or position occupied by the candidate company on the supply chain graph corresponds to the same or adjacent stage as the supply chain position that characteristically appeared at the growth stage.

[0107] The patterning unit (143) can convert the industry matching, transaction type similarity, and supply chain location correspondence degree for each candidate company into numerical values ​​based on the results of the judgments for i) to iii) above, and calculate a weighted evaluation value by multiplying the numerical value by a pre-set item-specific weighting coefficient. The patterning unit (143) can calculate a correction score by adding the weighted evaluation value to the existing candidate score or by multiplying the existing candidate score by the weighted evaluation value. Subsequently, the patterning unit (143) can generate a reconstructed candidate group by rearranging the candidate group (i.e., refined candidate group) based on the correction score or by selecting only candidate companies that are equal to or greater than a pre-set standard score.

[0108] The above-mentioned reconstructed candidate group may be arranged to have a structure in which candidate companies with high responsiveness to the supply chain characteristics observed at the current growth stage of the target company are placed relatively higher compared to the above-mentioned candidate group (i.e., refined candidate group). This above-mentioned reconstructed candidate group may be a recommended trading partner candidate group optimized for the target company. The patterning unit (143) may transmit the above-mentioned reconstructed candidate group to the score calculation unit (144).

[0109] <Score Calculation Unit (144)>

[0110] The score calculation unit (144) can calculate scores for each candidate company in the refined candidate group received from the filtering unit (142) for multiple evaluation items. In particular, the score calculation unit (144) can calculate scores for each candidate company in the reconstructed candidate group, which is a reconstructed candidate group formed by reconstructing the refined candidate group through the patterning unit (143), for multiple evaluation items.

[0111] Here, the plurality of evaluation items may include, for example, at least one of structural compatibility, necessity, transaction frequency, accessibility, and timing (suitability for the timing of transaction formation). That is, the scores for each of the plurality of evaluation items may include a structural compatibility score, a necessity score, a transaction frequency score, an accessibility score, and a timing score.

[0112] The score for each evaluation item can be calculated in the range of 0 points or more and 100 points or less, and the score calculation unit (144) can calculate the final evaluation score by summing or weighting the scores for each evaluation item and then transmit it to the ranking calculation unit (145).

[0113] That is, the score calculation unit (144) can calculate scores for each of the five evaluation items—structural suitability, necessity, transaction frequency, accessibility, and timing—for each candidate company within the reconstructed candidate group received from the patterning unit (143) and transmit them to the ranking calculation unit (145).

[0114] Here, the structural compatibility score may refer to a score that evaluates how structurally compatible a candidate company within a reconstructed candidate group is with the target company's industry, items handled, and position in the supply chain structure. The score calculation unit (144) can calculate (derive) the structural compatibility score by calculating the degree of suitability between the target company and the candidate company's industry based on a KSIC mapping rule table, and by reflecting the degree of matching between the target company and the candidate company's items handled by referring to an industry-specific keyword dictionary.

[0115] The above need score may refer to a score that evaluates the extent to which a candidate company currently needs the product or service of a target company. The score calculation unit (144) can calculate the need score by analyzing financial data from an external company database (400) (e.g., raw material purchase ratio, inventory increase / decrease trend, etc.) and unstructured data collected from the crawling unit (141) (e.g., announcement for finding new suppliers, recruitment announcement for purchasing managers, etc.).

[0116] The above transaction frequency score may refer to a score that evaluates the frequency and continuity of transactions that a candidate company has previously conducted with companies in the same industry. The score calculation unit (144) can calculate a transaction frequency score that reflects the candidate company's transaction frequency and transaction duration by analyzing actual transaction history data from an external company database (400).

[0117] The above accessibility score may refer to a score that evaluates the possibility that a target company can actually approach a candidate company and complete a transaction. The score calculation unit (144) may calculate the accessibility score by considering the location of the candidate company, the size of the company, the status of existing transaction relationships, and the decision-maker access channels.

[0118] The above timing score may represent a score that evaluates the likelihood that a candidate company is actively seeking new clients at the current time, that is, the appropriateness of the timing for market entry. The score calculation unit (144) can calculate the timing score by analyzing information such as the candidate company's recent financial fluctuation trends, management changes, business expansion plans, and recent demand fluctuation signals.

[0119] That is, for example, the score calculation unit (144) can calculate a plurality of detailed indicator values ​​corresponding to each evaluation item, and then convert the detailed indicator values ​​into evaluation scores in the range of 0 points or more and 100 points or less according to a preset normalization rule or interval score conversion rule.

[0120] For example, the score calculation unit (144) calculates the structural compatibility score by: i) calculating the industry matching degree between the target company and the candidate company, the similarity of keywords for handled items, and the suitability of the supply chain position, respectively, and normalizing the indicator values ​​to calculate the structural compatibility score; ii) calculating the necessity score by calculating the raw material purchase ratio, the inventory increase / decrease rate, and whether a new supplier search signal has occurred as indicator values, and scoring the indicator values ​​according to a pre-set interval score table; iii) calculating the transaction activity indicator based on the number of past transactions, the duration of transactions, and the number of trading counterpart companies, and normalizing the indicators to convert them into a transaction frequency score; iv) calculating the accessibility score by calculating the distance between companies, the difference in company size, whether there is an existing network connection, and whether there is a decision-maker contact channel as indicator values, and calculating the accessibility score based on the indicator values; and v) analyzing event signals such as the recent financial fluctuation rate, whether there is a change in management, disclosure information regarding business expansion, and new recruitment announcements to calculate the transaction possibility indicator. The above indicators can be calculated, and the timing score can be calculated by converting them into a score.

[0121] In this way, the score calculation unit (144) can transmit the scores for multiple evaluation items calculated for each candidate company within the reconstructed candidate group to the ranking calculation unit (145).

[0122] <Ranking Calculation Unit (145)>

[0123] The ranking calculation unit (145) can rank the top N customers that are pre-set based on the scores for each of the plurality of evaluation items.

[0124] That is, the ranking calculation unit (145) calculates the final score (final evaluation score) of each candidate company by applying a pre-set weight to each evaluation item from the scores for each multiple evaluation item for each candidate company received from the score calculation unit (144), and can rank the top N business partners that are pre-set by sorting them in descending order based on the final score. Here, the number of top N partners may be a natural number and may be set and changed in various ways by the manager, for example, as 5.

[0125] That is, the score calculation unit (144) can calculate the final score for each candidate company within the candidate group reconstructed according to the following mathematical formula 1 by applying a pre-set weight to the scores for each of the plurality of evaluation items.

[0126] [Mathematical Formula 1]

[0127] Final Score = w1 × Structural Compatibility Score + w2 × Need Score + w3 × Transaction Frequency Score + w4 × Accessibility Score + w5 × Timing Score

[0128] Here, w1 to w5 may be weights reflecting the importance of each evaluation item, and said weights may be set based on system settings or training data and may be set and changed in various ways by an administrator.

[0129] For example, weights for each evaluation item may be applied as follows: 35% for structural suitability, 25% for necessity, 20% for transaction frequency, 15% for accessibility, and 5% for timing. However, these numerical examples are merely examples to aid in understanding the present invention and are not limited thereto, and the settings for the weights for each evaluation item may be set and adjusted (changed) in various ways through the administrator terminal (200) depending on industry characteristics, market conditions, user settings, etc.

[0130] The ranking calculation unit (145) can calculate ranking results based on the final scores of candidate companies within the reconstructed candidate group, dividing them into, for example, TOP N purchasing candidates and TOP N selling candidates. At this time, the value of N can be applied differently depending on the user's subscription level, for example (in particular, the higher the subscription level, the larger the value of N is set). As mentioned earlier, this value of N can be set and changed in various ways by the administrator. The ranking calculation unit (145) can transmit the calculated ranking results to the recommendation provision unit (14).

[0131] The recommendation providing unit (14) can provide the ranking result received from the ranking calculation unit (145) to the user terminal (300). In particular, the recommendation providing unit (14) can generate recommended business partner information including the ranking result and provide it to the user terminal (300).

[0132] In the above ranking results, information about the top N pre-set business partners may be included as information about recommended business partners (custom recommended business partner information), and in particular, the information about recommended business partners may include information about the top N candidates for purchasing partners and the top N candidates for selling partners.

[0133] The above recommended business partner information may further include, for example, information regarding the above ranking results, as well as i) detailed information about the recommended business partner (recommended business partner) including the company name, industry, location, main products handled, final score and score for each evaluation item, and summary information on financial soundness, and ii) information on the basis of recommendation summarizing the reasons why the company was recommended for each recommended business partner.

[0134] The recommendation provision unit (14) can distinguish between the purchase location candidate and the sales location candidate when providing the ranking results.

[0135] For example, when providing recommended business partner information, the recommendation provider (14) may limit the number of recommended business partners and the range of detailed information available for viewing according to the user's subscription level, and may process this based on the subscription level information received from the membership management unit (17).

[0136] The recommendation providing unit (14) can control the process so that when a user registers interest, requests a meeting, or expresses an intention to proceed with a transaction regarding a specific recommended business partner among the recommended business partner information provided by the user, the event corresponding to the registration of interest, the request for a meeting, or the intention to proceed with the transaction is transmitted to the step-by-step billing unit (16), thereby allowing billing corresponding to the step of the event to be processed.

[0137] <Risk Monitoring Department (15)>

[0138] The risk monitoring unit (15) can continuously monitor the credit status and financial risks of recommended business partners corresponding to the recommended business partner information based on the recommended business partner information provided to the user terminal (300), and provide a notification to the user terminal (300) when an abnormal sign is detected.

[0139] The risk monitoring unit (15) can periodically collect and analyze risk indicators such as changes in credit ratings, abnormal changes in financial statements, occurrence of delinquency history, and changes in default prediction scores of companies registered as recommended trading partners through linkage with an external corporate database (400). For example, the risk monitoring unit (15) can update and analyze the above risk indicators according to a preset monitoring cycle (e.g., daily, weekly, etc.), and can perform additional monitoring based on the event when a specific risk event (e.g., credit rating decline, occurrence of delinquency, etc.) occurs.

[0140] In addition, the risk monitoring unit (15) can predict the likelihood of a recommended business partner company going bankrupt using an artificial intelligence-based bankruptcy prediction model. The bankruptcy prediction model may be a model that has been trained in advance using financial data of the recommended business partner company (e.g., sales fluctuation rate, debt ratio, current ratio, operating profit margin, etc.), transaction history data, credit rating information, and delinquency history information as input data, and may calculate the likelihood of a bankruptcy occurring within a certain period (e.g., within 3 months) in the form of a probability value or a prediction score by analyzing the time series trend of the input data.

[0141] The risk monitoring unit (15) can calculate a composite risk score by combining unstructured data (e.g., negative news articles, articles related to legal disputes, information related to corporate reputation, etc.) collected from the crawling unit (141) with structured data such as the financial data and credit data. For example, the risk monitoring unit (15) can calculate a composite risk score by scoring the risk indicators based on structured data and the risk signals based on unstructured data, respectively, and then applying pre-set weights.

[0142] The risk monitoring unit (15) can generate risk alert information for the recommended business partner company and transmit it to the user terminal (300) when the calculated composite risk score exceeds a preset threshold.

[0143] <Stage-based billing section (16)>

[0144] The step-by-step billing unit (16) can differentiate the level of information disclosure step by step according to the progress of interaction between the user and the recommended business partner provided by the recommendation provider (14), and can perform step-by-step billing processing in response to the occurrence of an entry event into each step.

[0145] The stage-by-stage billing unit (16) can determine whether to enter a stage based on the user's behavior event log, and the behavior events may include requests to view recommended business partner information, requests to disclose corporate information, requests to confirm meeting schedules, requests to enter contract spaces, etc. The stages managed by the stage-by-stage billing unit (16) may include an anonymous matching stage (Stage 0), a real-name / corporate information disclosure stage (Stage 1), a meeting confirmation stage (Stage 2), and a contract space entry stage (Stage 3).

[0146] The above anonymous matching stage (Stage 0) may be a stage where an initial match with a recommended business partner is made, but the real name and specific corporate identification information of the business partner are anonymized. For example, in the above anonymous matching stage, information that can identify a company (e.g., company name, business registration number, contact information, etc.) is provided in a de-identified state, and the service may be provided without separate charges in that stage.

[0147] The above real name / company information disclosure stage (Stage 1) may be a stage entered when a user performs an information disclosure request event to verify the real name or detailed company information of a specific recommended business partner. When the above information disclosure request event occurs, the stage billing unit (16) may recognize the event as a stage entry event and charge a certain amount according to a pre-set fee table. For example, the information disclosed may include the company name, location, contact information, main business activities, company size, financial overview information, etc.

[0148] The above meeting confirmation stage (Stage 2) may be a stage entered when a meeting confirmation event occurs in which a user confirms a meeting schedule with a specific recommended business partner. The stage billing unit (16) may perform billing of a certain amount according to a pre-set billing policy when the above meeting confirmation event occurs, and the above meeting confirmation event may be determined based on system event logs such as schedule registration, schedule approval, or transmission of a meeting schedule confirmation message.

[0149] The above contract space entry stage (Stage 3) may be a stage entered when a user enters a dedicated contract space (Deal Room) within the system to proceed with the contract conclusion procedure with a specific recommended business partner. The stage billing unit (16) recognizes the contract space entry event as a stage entry event when the contract space entry event occurs, and may perform billing at a higher amount than Stage 1 and Stage 2 according to a pre-set billing policy. For example, the above contract space may be a collaborative space that performs functions such as sharing contract-related documents, managing contract negotiation records, drafting contracts, and electronic signatures.

[0150] The step-by-step billing unit (16) can automatically record each step entry event in the form of a system log (Log) within a database unit (not shown) in the device (100). The recorded log may include information such as a log identifier (Log ID), user identification information (User ID), step event type (Action Type), billing processing status (Billing Status), and processing time (Time Stamp). The step-by-step billing unit (16) can manage the billing occurrence history based on the system log and store the billing processing results linked to a user account.

[0151] The step-by-step billing unit (16) can be configured to perform contract-related functions within the device (100), such as drafting a contract, electronically signing, concluding a non-disclosure agreement (NDA), and storing contract negotiation records, within the contract space. Additionally, the step-by-step billing unit (16) can maintain a platform-based transaction management environment by restricting the external export of transaction history data and user behavior logs generated during the contract process and managing the data so that it is stored in the system's internal data storage unit.

[0152] < Membership Management Department (17)>

[0153] The membership management unit (17) can manage the service usage rights and service usage scope of the service according to the user's subscription grade. The membership management unit (17) stores and manages subscription grade information corresponding to the user account, and can control the recommendation service usage scope, information viewing scope, and the number of matching provided according to the said subscription grade. The subscription grades managed by the membership management unit (17) may include a free trial grade, a first paid grade, a second paid grade, and a third paid grade.

[0154] The aforementioned free trial tier may be a service tier provided without a separate subscription fee, and may offer basic services such as the ability to view a list of recommended business partners, view business partner details a limited number of times, and register business partners of interest. For example, restrictions on service usage for a certain period or a certain number of times may apply to the free trial tier.

[0155] The above first paid grade may be a basic paid grade provided to users who pay a monthly subscription fee. The membership management department (17) may provide the first paid grade user with the ability to view the recommended business partner list without limit, the ability to view detailed business partner information without limit, the ability to provide recommendation matching a set number of times per month, and the right to use the business partner recommendation service.

[0156] The above second paid tier may be a higher tier provided to users who pay a higher subscription fee than the first paid tier. The membership management department (17) may provide the second paid tier user with all the service functions provided in the first paid tier, along with additional services such as a function to increase the number of monthly recommendation matches, a function to check corporate stock value information, and a function to register and upload corporate promotional content.

[0157] The above third paid tier may be the highest subscription tier for corporate users or users utilizing premium services. The membership management department (17) may provide users of the third paid tier with all the service functions of the second paid tier, along with additional services such as an additional increase in the number of monthly recommendation matches, a provision of expert advisory services, and transaction negotiation support services. For example, the expert advisory services may include consultation services provided by professional personnel such as patent attorneys, lawyers, and accountants.

[0158] The membership management unit (17) can provide the user's current subscription level information to the recommendation provision unit (14), and the recommendation provision unit (14) can control the scope of recommended business partners provided, the number of recommendation matches, and information viewing rights based on the subscription level information. In addition, the membership management unit (17) performs billing and payment processing for each subscription level, processes subscription-related events such as subscription renewal, subscription level change, and subscription cancellation, and can store the processing results by reflecting them in the user account information.

[0159] The present device (100) can provide a business partner recommendation service based on KSIC mapping rules and industry-specific keyword dictionaries specialized for specific major industries, such as manufacturing (KSIC C24~C26) and wholesale (KSIC G46), for example. Additionally, the present device (100) can provide a supply chain recommendation service specialized for a specific industry domain (for example, the beauty / cosmetics industry, etc.), and in this case, it can provide recommendation results specialized for each type of supply chain participant in the industry, such as raw material / ingredient manufacturers, container / subsidiary material suppliers, OEM / ODM manufacturers, and distributors / brand companies. Furthermore, the present device (100) may provide various additional functions as follows, in addition to the functions described above.

[0160] <Supply chain shock transfer simulation section (18)>

[0161] For example, the device (100) may further include a supply chain shock transfer simulation unit (18).

[0162] The supply chain shock transfer simulation unit (18) receives a supply chain graph generated by the supply chain reverse engineering unit (13) as input data, injects a virtual risk event into a specific node (specific company) within the supply chain graph, and can simulate the process of the risk event transferring to another node along an edge on the graph. Here, the supply chain graph may be composed of nodes representing multiple companies and edges representing transaction relationships between companies, and attribute information such as transaction dependency, transaction item information, and transaction frequency may be assigned to each edge.

[0163] The above supply chain shock transfer simulation unit (18) may selectively inject virtual risk events stored in a risk event database (not shown) within the device (100) to perform a simulation, and the risk events may include events such as natural disasters, production facility failures, strikes, suspension of raw material supply, financial insolvency, and bankruptcy. When the risk event is injected into a specific node, the supply chain shock transfer simulation unit (18) calculates the risk transfer probability along the edge connected to the node, and if the transfer probability is greater than or equal to a preset threshold value, it may determine that the risk is transferred to the next node.

[0164] The above supply chain shock transfer simulation unit (18) may apply a transaction dependency index and a substitutability score as weights to each edge to calculate the probability of risk transfer. Here, the transaction dependency index may be calculated based on the proportion that a specific trading partner accounts for in total purchases or sales, and the substitutability score may be calculated by considering the existence of alternative suppliers handling the same item, the number of alternative suppliers, switching costs, and the time required for switching.

[0165] The supply chain shock transfer simulation unit (18) can calculate the probability that a risk originating from a specific node will reach a target enterprise node and the estimated time to reach it through risk transfer path analysis. The estimated time to reach (Time-to-Impact) can be calculated based on the average transaction lead time, logistics movement time, and inventory holding period of each edge, and the estimated damage scale (Financial Exposure) can be calculated in monetary units by considering transaction dependency, expected sales reduction rate, and alternative supply costs.

[0166] In addition, if the supply chain shock transfer simulation unit (18) determines, based on the simulation results, that the expected scale of damage to the target company exceeds a preset risk threshold, it may classify the risk event as a high-risk event and transmit a request to the recommendation provision unit (14) to search for an alternative supplier. Accordingly, the recommendation provision unit (14) may search for an alternative node capable of supplying the same item on the supply chain graph and provide a candidate for an alternative supply chain to the user terminal (300).

[0167] The device (100) can support predicting in advance the impact of risks that may occur at a specific company in the supply chain on a target company by providing such a supply chain shock transfer simulation unit (18), and establishing a response strategy for the risk of supply chain disruption in advance.

[0168] <ESG 공급망 투명성 분석부(19)>

[0169] Additionally, the device (100) may further include an ESG supply chain transparency analysis unit (19). The ESG supply chain transparency analysis unit (19) can evaluate supply chain-level ESG risks by collecting and analyzing Environment, Social, and Governance related data for multiple companies included in the supply chain graph generated by the supply chain reverse engineering unit (13).

[0170] The above ESG supply chain transparency analysis unit (19) can collect company-specific ESG-related information from external data sources through the crawling unit (141), and the above ESG-related information may include carbon emission information, history of environmental regulation violations, records of industrial accidents, records of labor human rights disputes, records of fair trade violations, and disclosure information related to corporate governance.

[0171] The above ESG supply chain transparency analysis unit (19) can calculate an ESG risk index for each company based on collected ESG data. For example, the above ESG risk index can be calculated by calculating an environmental risk score, a social risk score, and a governance risk score, respectively, and then weighting and summing them. Here, the environmental risk score can be calculated based on carbon emissions, the number of environmental regulation violations, and the size of environmental fines, and the social risk score can be calculated based on the occurrence of labor disputes, the frequency of industrial accidents, and human rights-related issues.

[0172] The above ESG supply chain transparency analysis unit (19) can calculate the overall ESG risk of the entire supply chain by considering not only the ESG risk index of the specific recommended trading partner company itself, but also the ESG risk indices of the upstream and downstream nodes that have a trading relationship with the company. The above supply chain ESG risk can be calculated by summing the ESG risk indices of the nodes connected on the supply chain graph according to the trading dependency weight.

[0173] In addition, the ESG supply chain transparency analysis unit (19) may set an ESG risk flag for the recommended trading partner when the calculated ESG overall risk exceeds a preset threshold value, and provide regulatory risk information that may occur when conducting a transaction with the recommended trading partner to the user terminal (300). For example, the regulatory risk information may include the possibility of export restrictions related to the Carbon Border Adjustment Mechanism (CBAM), the possibility of transaction restrictions due to violations of environmental regulations, etc.

[0174] The device (100) can support companies in identifying ESG risks throughout the supply chain in advance and establishing sustainable supply chain management strategies by providing an ESG supply chain transparency analysis unit (19).

[0175] <Supply chain similarity-based technology scouting department (20)>

[0176] Additionally, the device (100) may further include a supply chain similarity-based technology scouting unit (20). The technology scouting unit (20) can identify candidate companies capable of technology cooperation or strategic alliance by comparing and analyzing technology information held by a target company with technology information of companies included in a supply chain graph.

[0177] The technology scouting unit (20) can collect patent data, technology description documents, and technology keyword information of companies included in the target company and supply chain graph through the crawling unit (141). The technology information may include a patent classification code (IPC or CPC code), the title of the invention, a technology summary, and an applicable industry field.

[0178] The technology scouting unit (20) can generate technology vectors for each company based on the technology information and calculate technology similarity between companies. For example, the technology similarity can be calculated by comprehensively considering the degree of matching of patent classification codes, the similarity of technology keywords, and the similarity of application process steps.

[0179] In addition, the technology scouting unit (20) can calculate process structure similarity by analyzing process step information between companies on a supply chain graph. The process structure similarity can be calculated based on whether the same raw materials are used, similarity of production process steps, and similarity of manufacturing equipment types, and the process structure similarity and technology similarity can be combined to calculate a final technology cooperation possibility score.

[0180] The technology scouting unit (20) can classify companies whose technology cooperation potential score is higher than a preset threshold value as technology cooperation candidate companies and transmit them to the recommendation provision unit (14). In addition, if there is a high dependence on overseas parts or materials, the technology scouting unit (20) can prioritize searching for companies that possess the technological capability to replace the item and classify them as candidates for localization.

[0181] The device (100) can provide a technology-based business matching function that supports technology cooperation, joint research and development, and the discovery of strategic alliance partners beyond simple transaction relationship analysis by providing a supply chain similarity-based technology scouting unit (20).

[0182] <Survival prediction-based protection unit (21)>

[0183] Additionally, the device (100) may further include a survival prediction-based protection unit (21). The survival prediction-based protection unit (21) can predict the management stability and viability of a recommended business partner candidate company (i.e., a recommended business partner company corresponding to the recommended business partner information) among a plurality of companies included in the supply chain graph generated by the supply chain reverse engineering unit (13) within a certain period. The survival prediction-based protection unit (21) can analyze the viability of a company or the possibility of a change in management rights by collecting financial data, business activity data, and externally disclosed data for each company.

[0184] The above survival prediction-based protection unit (21) may apply a time series prediction model to predict future changes in the business status of a company based on time series data for each company. For example, the above time series prediction model may include one or more of a Recurrent Neural Network, a Long Short-Term Memory (LSTM), or a Transformer-based time series model, and may predict the possibility of mergers and acquisitions, changes in management rights, or closure of a company within a future period by using corporate activity data accumulated over a certain period as input data.

[0185] The above-mentioned survival prediction-based protection unit (21) can calculate multiple risk indicators to analyze the survival potential of a company. The risk indicators may include the company's core personnel attrition rate, patent maintenance status, whether a new legal entity has been established, information on changes in major business partners, and information on changes in financial indicators. Here, the core personnel attrition rate may be calculated based on recruitment announcement data, information on changes in the number of employees, and publicly available personnel movement data, and the patent maintenance status may be determined based on whether patent maintenance fees have been paid or the status of patent rights maintenance.

[0186] In addition, the above-mentioned survival prediction-based protection unit (21) can evaluate the possibility of a change in management rights by analyzing the history of new corporate establishment by the representative or key executive, information on the establishment of new companies within the same industry, and information on changes in governance structure between companies, and additionally evaluate the business sustainability of the company by analyzing whether transactions with major sales sources have ceased, trends in the decrease of transaction volume, and changes in transaction relationships within the supply chain.

[0187] The above survival prediction-based protection unit (21) can calculate a supply chain survival index based on the above risk indicators. The supply chain survival index is an indicator calculated by applying weights to a plurality of risk indicators, and can indicate the possibility that the company will stably survive within the current supply chain for a certain period.

[0188] The above survival prediction-based protection unit (21) may classify the company as having low management stability if the supply chain survival index is calculated to be below a preset threshold value, exclude the company from the recommended candidate list, or provide risk indication information to the user terminal (300). Additionally, the above survival prediction-based protection unit (21) may calculate the risk of potential technology transfer, potential changes in transaction conditions, or supply interruption that may occur when entering into a contract with the company, and provide this as warning information to the user terminal (300).

[0189] The device (100) can support the selection of stable business partners by providing a survival prediction-based protection unit (21), thereby not evaluating business partners based only on their current financial status, but by comprehensively considering future business stability and supply chain sustainability.

[0190] <Process compatibility verification department (22)>

[0191] Additionally, the device (100) may further include a process compatibility verification unit (22). The process compatibility verification unit (22) can verify the manufacturing process compatibility between the two companies by comparing and analyzing the product production requirements of the target company with the manufacturing facilities and process capabilities of the recommended trading partner company.

[0192] The process compatibility verification unit (22) can collect equipment information, production process information, and technical specification information of recommended business partners corresponding to the recommended business partner information provided by the recommendation provision unit (14) through the crawling unit (141). The equipment information may include the type of manufacturing equipment, equipment precision, maximum processing size, types of materials that can be produced, and equipment performance information. In addition, the process information may include processing steps, process processing precision, production line configuration, and quality control standards.

[0193] The process compatibility verification unit (22) can receive product design data or product specification data provided by the target company as input data, and the product design data may include product drawing information, required process steps, required precision tolerances, materials used, and quality standards.

[0194] The process compatibility verification unit (22) can determine whether manufacturing is possible by comparing the equipment information of the recommended business partner with the product specification information of the target company. For example, the process compatibility verification unit (22) can calculate a process compatibility score by comparing and analyzing the difference between equipment precision and required tolerance, the suitability between the specifications that can be processed by the equipment and the product size, and the types of materials that the equipment can process.

[0195] In addition, the process compatibility verification unit (22) can further extract equipment information and process information by analyzing image and text information included in the website, equipment catalog, or promotional materials of the recommended business partner company. For example, the process compatibility verification unit (22) can identify the type of equipment, equipment performance, and process capability range by using a natural language processing model for text information analysis and an image recognition model for image analysis.

[0196] The process compatibility verification unit (22) can evaluate the possibility of digital collaboration by comparing data exchange methods and equipment control interface information between the two companies. For example, the process compatibility verification unit (22) can calculate a digital collaboration readiness score by comparing and analyzing design data exchange formats, production control system interfaces, and data communication standards.

[0197] The process compatibility verification unit (22) may provide the calculated process compatibility score and digital collaboration readiness score to the score calculation unit (144) to be reflected in the process of calculating the recommended business partner evaluation score.

[0198] The device (100) can support more accurate business partner matching by providing a process compatibility verification unit (22), thereby verifying in advance the actual manufacturing process mismatch problem that may occur in a method of recommending business partners based only on industry information between companies.

[0199] <Real-time operation rate analysis unit (23)>

[0200] Additionally, the device (100) may further include a real-time operating rate analysis unit (23). The real-time operating rate analysis unit (23) can estimate the production facility operating status or factory operation status of a recommended client company to analyze the company's production capacity and delivery response capabilities.

[0201] The above real-time operating rate analysis unit (23) can estimate the level of production activity of a company using externally disclosed data and sensor-based data. For example, the above real-time operating rate analysis unit (23) can collect and analyze factory power usage data, industrial power consumption statistics, satellite or aerial image data, and logistics movement data.

[0202] The above real-time operating rate analysis unit (23) can estimate the operating rate of production facilities by analyzing the hourly fluctuation pattern of factory power consumption data. In addition, the above real-time operating rate analysis unit (23) can determine the level of factory operation activity by analyzing changes in night lighting brightness and vehicle movement patterns observed in satellite images or aerial images.

[0203] The above real-time operating rate analysis unit (23) can estimate the frequency of production volume shipments by analyzing the frequency of logistics vehicle entry and exit or logistics movement data observed in the area surrounding the company. By synthesizing the results of such multiple data analyses, the factory operating rate and production capacity index of the company can be calculated.

[0204] The above real-time operating rate analysis unit (23) can determine that the company is in a state of production capacity saturated and lower its recommendation priority if the calculated production capacity index is below a preset threshold value. Conversely, if a company with sufficient production capacity is identified, the company can be classified as a priority recommendation candidate.

[0205] In addition, the real-time operating rate analysis unit (23) can identify companies whose production operating rate has decreased rapidly over a certain period and classify them as potential negotiable companies, and can provide the company to the user terminal (300) by separately marking it as a candidate for transaction with the target company.

[0206] The device (100) can support a more realistic assessment of delivery response capabilities by providing a real-time operating rate analysis unit (23), which reflects not only the past financial information of recommended clients but also the production capacity and factory operation status at the current time.

[0207] Below, we will briefly examine the operation flow of the present invention based on the details described above.

[0208] FIG. 5 is a flowchart of the operation of a business partner recommendation method that recommends B2B business partners based on supply chain reverse engineering using artificial intelligence according to one embodiment of the present invention.

[0209] A business partner recommendation method (hereinafter referred to as the "business partner recommendation method") that recommends supply chain reverse engineering-based B2B business partners using artificial intelligence illustrated in FIG. 5 can be performed by the device (100) described above. Therefore, even if the content described below is omitted, the description of the device (100) can be applied equally to the description of the business partner recommendation method.

[0210] Referring to FIG. 5, in step S11, the input receiving unit can receive a business number from the user terminal.

[0211] Next, in step S12, the corporate information lookup unit can retrieve corporate information data of the target company (user's company) corresponding to the business registration number by linking with the API of an external corporate database.

[0212] Next, in step S13, the supply chain reverse engineering unit can reverse engineer a supply chain network centered on the target company by analyzing the actual transaction data included in the above company information data.

[0213] Next, in step S14, the recommendation provider can generate recommended business partner information based on a supply chain graph corresponding to the supply chain network and provide it to the user terminal.

[0214] In the description above, steps S11 to S14 may be further divided into additional steps or combined into fewer steps, depending on an embodiment of the present invention. Additionally, some steps may be omitted as necessary, and the order between steps may be changed.

[0215] The customer recommendation method according to one embodiment of the present invention may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either individually or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the present invention, or they may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware device may be configured to operate as one or more software modules to perform the operation of the present invention, and vice versa.

[0216] In addition, the aforementioned method for recommending business partners may also be implemented in the form of a computer program or application executed by a computer and stored on a recording medium.

[0217] The foregoing description of the present invention is for illustrative purposes only, and those skilled in the art will understand that other specific forms can be easily modified without altering the technical spirit or essential features of the present invention. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive. For example, each component described as a single unit may be implemented in a distributed manner, and components described as distributed may likewise be implemented in a combined form.

[0218] The scope of the present invention is defined by the claims set forth below rather than by the detailed description above, and all modifications or variations derived from the meaning and scope of the claims and equivalent concepts thereof should be interpreted as being included within the scope of the present invention.

Claims

Claim 1 A user terminal for inputting a business registration number; and a business partner recommendation device that receives a business registration number from the user terminal and recommends a business partner through AI-based matching based on the business registration number, wherein the business partner recommendation device comprises: an input receiving unit that receives a business registration number from the user terminal; a corporate information inquiry unit that queries corporate information data of a target company corresponding to the business registration number by linking with an API of an external corporate database; and a supply chain reverse engineering unit that analyzes actual transaction data included in the corporate information data to reverse engineer a supply chain network centered on the target company. The system includes a recommendation provision unit that generates recommended business partner information based on a supply chain graph corresponding to the supply chain network and provides it to a user terminal; the reverse engineering performed by the supply chain reverse engineering unit refers to a series of data processing processes that, in the absence of a company's overall supply chain blueprint, extract identification information and transaction attributes of trading partners using fragmentary actual transaction data and financial data as inputs, and reconstruct the entire network structure by tracing back the multi-layered connection relationships between companies from the lower level to the upper level based on the extracted information; the supply chain reverse engineering unit designs the supply chain network, which refers to a set of multi-layered transaction relationships formed around the target company, by analyzing the actual transaction data to identify primary trading companies that have a direct transaction relationship with the target company, and additionally tracing and identifying secondary trading companies that have transaction relationships with the primary trading companies; subsequently, within the designed supply chain network, it extracts transaction relationship attribute information including the transaction direction (purchase or sale direction), transaction volume, transaction frequency, and transaction duration; and subsequently, by modeling the extracted transaction relationship attribute information into a graph structure, each company is designated as a node and the transaction relationship Create a supply chain graph with edges, andBy analyzing the above actual transaction data, a supply chain graph generated based on a reverse-engineered supply chain network is transmitted to a patterning unit and a filtering unit. The above-mentioned business partner recommendation device further includes a survival prediction-based protection unit that predicts the management stability and viability of recommended business partner candidate companies for a predetermined period among a plurality of companies included in the supply chain graph generated by the supply chain reverse engineering unit. The survival prediction-based protection unit collects financial data, corporate activity data, and externally disclosed data for each company to analyze the viability of the company or the possibility of a change in management control. It applies a time series prediction model to predict future changes in the company's management status based on time series data for each company, wherein one or more of a Recurrent Neural Network, a Long Short-Term Memory (LSTM), or a Transformer-based time series model are used as the time series prediction model. It uses corporate activity data accumulated during a predetermined period as input data to predict the possibility of mergers and acquisitions, changes in management control, or closure of the company within a future predetermined period. Furthermore, it calculates a plurality of risk indicators to analyze the viability of the company, wherein the risk indicators include the company's core personnel attrition rate, patent maintenance status, and establishment of a new corporation. Considering the existence, information on changes in business partners, and information on changes in financial indicators; calculating the aforementioned core personnel attrition rate based on recruitment data, information on changes in the number of employees, and externally disclosed personnel movement data; determining the aforementioned patent maintenance status based on whether maintenance fees have been paid or the status of patent maintenance; additionally evaluating the potential for a change in management control by analyzing the history of new corporate establishment by the representative or key executives, information on the establishment of new companies within the same industry, and information on changes in corporate governance between companies; and additionally evaluating the business sustainability of the company by analyzing whether customers have discontinued transactions, trends in the decrease of transaction volume, and changes in transaction relationships within the supply chain.A business partner recommendation system that calculates a supply chain survival index based on risk indicators, which indicates the likelihood of a company remaining stable within the current supply chain for a predetermined period, wherein the supply chain survival index is calculated by applying weights to multiple risk indicators, and if the supply chain survival index is calculated to be below a predetermined threshold value, classifies the company as having low management stability, excludes the company from the recommendation candidate list, or provides risk indication information to the user terminal, and additionally derives the risk of potential changes in transaction conditions or supply interruption that may occur when entering into a contract with the company, and provides this as warning information to the user terminal. Claim 2 delete Claim 3 A business recommendation system according to claim 1, wherein the recommendation providing unit comprises: a filtering unit that generates a candidate group of business partners by considering the supply chain graph and refines the candidate group by applying a filter to the candidate group; a score calculation unit that calculates scores for each candidate company in the refined candidate group based on a plurality of evaluation items; and a ranking calculation unit that ranks the top N business partners in a predetermined order based on the scores for the plurality of evaluation items. Claim 4 In paragraph 3, the recommendation providing unit further comprises a crawling unit that collects unstructured data related to companies included in the supply chain graph, extracts valid information using a pre-set industry-specific keyword dictionary, and transmits it to the filtering unit, thereby forming a business partner recommendation system. Claim 5 In paragraph 3, the recommendation providing unit further comprises a patterning unit that analyzes the supply chain structure patterns of top companies within the same industry as the target company using the supply chain graph and transaction history data of blue-chip companies collected from the external company database, thereby forming a business partner recommendation system.

Images