Knowledge graph-based procurement risk detection method and device, equipment and medium

Abstract

The application provides a knowledge graph-based procurement risk detection method, device, equipment and medium. In the scheme, by acquiring procurement data such as supplier information and cargo information to be detected, a risk detection model constructed by a knowledge graph model based on historical procurement data and business data of a plurality of suppliers acquired from a public database is used to detect the procurement data to be detected, a risk detection result is generated, and a risk detection report is generated according to the risk detection result. Users perform abnormal processing on the nodes with risks according to the risk detection report until the nodes no longer have early warning information and can be transferred to subsequent links. Through the above method, the problem that the existing procurement risk control method is difficult to quickly locate and cannot comprehensively cover the entire procurement process is solved, comprehensive monitoring and rapid response of the procurement process are realized, and enterprises can better manage and respond to possible risks in procurement business.

Description

Technical Field

[0001] This application relates to the field of enterprise management technology, and in particular to a method, apparatus, equipment and medium for procurement risk detection based on knowledge graphs. Background Technology

[0002] Procurement is a crucial aspect of a company's production and operations, directly impacting production efficiency, material supply, and economic benefits. It is a significant driver of overall competitiveness. Effective procurement management can bring substantial economic benefits and market advantages to a company.

[0003] However, due to the complexity of the supply chain, which involves raw material procurement, supplier selection and management, logistics and transportation, inventory management, and production planning, the procurement process involves many potential risks. Traditional procurement risk control methods mainly rely on manual inspection and experience-based judgment, which makes it difficult to quickly locate risks and cannot fully cover the entire procurement process, lacking effective correlation verification and control mechanisms.

[0004] In conclusion, providing a technical solution that enables accurate risk assessment of procurement is an urgent problem to be solved. Summary of the Invention

[0005] This application provides a knowledge graph-based procurement risk detection method, device, equipment, and medium to address the problems of existing traditional procurement risk control methods being unable to quickly locate risks and comprehensively cover the entire procurement process, and lacking effective correlation verification and control mechanisms. It provides a technical solution that can achieve accurate risk assessment and early warning for each stage of procurement.

[0006] Firstly, this application provides a knowledge graph-based procurement risk detection method, including:

[0007] Obtain the procurement data to be tested, which includes supplier information and goods information;

[0008] The procurement data is subjected to risk detection using a pre-acquired risk detection model to obtain risk detection results, which include detection results for at least one dimension of the supplier and / or goods.

[0009] A risk detection report is generated based on the risk detection results;

[0010] The risk detection model is pre-constructed using a knowledge graph model based on historical procurement data and business data from multiple suppliers obtained from public databases.

[0011] In one specific implementation, the risk detection result includes at least one of the following:

[0012] Supplier's supply stability test results;

[0013] Supplier's delivery cycle test results;

[0014] Supplier qualification test results;

[0015] Supplier credit test results;

[0016] The supplier's social responsibility test results;

[0017] Price inspection results of the goods;

[0018] The quality inspection results of the goods;

[0019] Quantity inspection results of the goods;

[0020] Invoice information detection results.

[0021] In one specific implementation, before performing risk detection on the procurement data using a pre-acquired risk detection model and obtaining the risk detection result, the method further includes:

[0022] Obtain a set of data to be processed, which includes: historical procurement data stored in a database, and business data of multiple suppliers obtained from a public database; the multiple suppliers are suppliers that have cooperated with the company, as determined based on the historical procurement data.

[0023] The data in the dataset to be processed is preprocessed to obtain an intermediate dataset in JSON format with entity tags, relationship tags, and attribute tags added.

[0024] The data in the intermediate dataset is used to perform entity recognition and extract the relationships between entities to obtain a knowledge graph model.

[0025] The risk detection model is obtained based on the knowledge graph model.

[0026] In one specific implementation, obtaining the risk detection model based on the knowledge graph model includes:

[0027] In response to the user's first configuration operation, configure the calculation methods for supply stability indicators, delivery cycle indicators, qualification testing indicators, credit testing indicators, and social responsibility testing indicators associated with suppliers;

[0028] In response to the user's second configuration operation, configure the calculation methods for price detection indicators, quantity detection indicators, quality detection indicators, and invoice information detection indicators associated with goods;

[0029] In response to the user's third configuration action, configure the risk detection rules for each metric;

[0030] Based on the calculation methods of supply stability indicators, delivery cycle indicators, qualification testing indicators, credit testing indicators, and social responsibility testing indicators associated with suppliers, the calculation methods of price testing indicators, quantity testing indicators, quality testing indicators, and invoice information testing indicators associated with goods, the risk detection rules for each indicator, and the knowledge graph model, the risk detection model is constructed.

[0031] In one specific implementation, the preprocessing of the data in the dataset to be processed to obtain an intermediate dataset in JSON format with added entity tags, relationship tags, and attribute tags includes:

[0032] The data in the dataset to be processed is cleaned to remove duplicate data, missing values ​​and outliers, resulting in a cleaned dataset.

[0033] The data in the cleaned dataset is converted to JSON format to obtain the converted dataset.

[0034] Entity tags, relationship tags, and attribute tags are added to the data in the format-converted dataset to obtain the intermediate dataset.

[0035] In one specific implementation, the risk detection report includes the risk detection results, early warning details, and risk handling measures;

[0036] The early warning information includes: at least one dimension of the supplier and / or goods that are at risk, and the corresponding indicator information.

[0037] Secondly, this application provides a procurement risk detection device based on a knowledge graph, comprising:

[0038] The first processing module is used to acquire the procurement data to be tested, which includes supplier information and goods information;

[0039] The second processing module is used to perform risk detection on the procurement data using a pre-acquired risk detection model to obtain risk detection results, which include detection results for at least one dimension of the supplier and / or goods.

[0040] The third processing module is used to generate a risk detection report based on the risk detection results;

[0041] The risk detection model is pre-constructed using a knowledge graph model based on historical procurement data and business data from multiple suppliers obtained from public databases.

[0042] Thirdly, this application provides a computer device, comprising:

[0043] Processor, memory, and communication interface;

[0044] The memory stores computer-executed instructions;

[0045] The processor executes computer execution instructions stored in the memory, causing the processor to perform the knowledge graph-based procurement risk detection method as described in the first aspect.

[0046] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the knowledge graph-based procurement risk detection method as described in the first aspect.

[0047] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the knowledge graph-based procurement risk detection method as described in the first aspect.

[0048] This application provides a knowledge graph-based procurement risk detection method, apparatus, equipment, and medium. In this scheme, procurement data such as supplier information and goods information to be detected are first acquired. Then, using business data from multiple suppliers pre-collected based on historical procurement data and publicly available databases, a risk detection model constructed using a knowledge graph model is employed to perform risk detection on the procurement data. This generates at least one type of detection result for the supplier and / or goods, and a risk detection report is generated based on the results. The aim is to identify and assess potential risks in the procurement process, enabling enterprises to better manage and respond to them. This method solves the problems of existing procurement risk control methods, which struggle to quickly locate risks and comprehensively cover the entire procurement process, achieving comprehensive monitoring and rapid response throughout the procurement process. Attached Figure Description

[0049] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0050] Figure 1 A flowchart illustrating an embodiment of a knowledge graph-based procurement risk detection method provided in this application.

[0051] Figure 2 A flowchart illustrating a second embodiment of a knowledge graph-based procurement risk detection method provided in this application;

[0052] Figure 3A flowchart illustrating a third embodiment of a knowledge graph-based procurement risk detection method provided in this application;

[0053] Figure 4 A schematic diagram illustrating the principle of a knowledge graph-based risk detection method provided in this application embodiment;

[0054] Figure 5 A schematic diagram illustrating a specific implementation of a risk detection method based on a knowledge graph, provided in this application.

[0055] Figure 6 A schematic diagram of the structure of a knowledge graph-based procurement risk detection device embodiment provided in this application;

[0056] Figure 7 A schematic diagram of the structure of a computer device provided in an embodiment of this application.

[0057] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0058] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0059] Before introducing the embodiments of this application, the application background of the embodiments of this application will be explained first:

[0060] Procurement is the foundation of a company's production and operations, and a crucial link in reducing costs, increasing efficiency, and boosting profits. Due to the complexity of the supply chain, involving raw material procurement, supplier selection and management, logistics, inventory management, and production planning, and influenced by uncontrollable internal and external factors, varying degrees of risk exist at each stage of procurement. If these risks are not controlled and warned against in a timely manner, the procurement process can easily breed practices such as "behind-the-scenes operations," abuse of power for personal gain, fraud, choosing the expensive over the cheap, using inferior materials, and accepting kickbacks. It can also easily lead to problems such as stockpiling and waste, delayed delivery, and substandard quality.

[0061] Traditional procurement risk control methods primarily rely on manual inspection and experience-based judgment, typically requiring experienced professionals to meticulously examine and evaluate each stage of the procurement process based on their expertise and past experience. On the one hand, this method is time-consuming, inefficient, and susceptible to factors such as inspector fatigue, negligence, or misjudgment. Furthermore, its coverage is limited, usually only addressing specific risk points. Moreover, while manual inspection and experience-based judgment can identify some common risks, they are highly subjective and uncertain; different inspectors may have different assessments of the same risk, leading to inconsistencies in risk identification and management. Simultaneously, experience-based judgment struggles to address newly emerging risk types when the market environment changes rapidly or the supply chain undergoes significant adjustments.

[0062] On the other hand, traditional procurement risk control methods lack effective correlation verification and control mechanisms. In complex procurement processes, various links are often closely linked and mutually influential. For example, supplier selection, contract terms, and logistics arrangements can all affect the final procurement outcome. However, traditional risk control methods typically handle these links in isolation, lacking systematic process correlation verification, which can easily lead to information asymmetry and poor communication, thereby increasing the likelihood of risk occurrence. Furthermore, traditional procurement risk control methods struggle to quickly locate risk sources. When a problem occurs in a link of the supply chain, manual inspection and experience-based judgment require significant time and effort to trace and find the root cause. This delayed risk location approach may lead to the escalation of the problem, thus causing a greater impact on the company's production and operations.

[0063] In conclusion, traditional procurement risk control methods are proving inadequate in today's complex supply chain environment. Enterprises need to seek more systematic and intelligent risk control measures to achieve comprehensive monitoring and rapid response throughout the procurement process. Therefore, providing a technical solution that can accurately assess and warn of risks at each stage of procurement, quantifying and visualizing the types and relationships of nodes in the entire procurement process to provide a basis for enterprise management decisions, is an urgent problem to be solved.

[0064] Based on the aforementioned technical problems, the inventors, in the process of researching procurement risk control methods, discovered a knowledge graph-based procurement risk detection method. This method, through structured and semantic methods, can associate complex supplier information, goods information, and related business data, comprehensively covering the entire procurement process. It can analyze and detect risk factors in procurement data in real time and generate detailed risk detection reports, achieving accurate risk assessment and early warning for each stage of procurement. Based on this, this application provides a knowledge graph-based procurement risk detection method, apparatus, equipment, and medium. In this solution, data is obtained from Enterprise Resource Planning (ERP) systems and third-party public databases, and after preprocessing, forms an intermediate data set. Natural Language Processing (NLP) technology is used to identify entities, extract relationships, and populate attributes, loading the information into a graph database to construct a knowledge graph model. Then, quantitative indicators associated with suppliers and goods and their risk detection rules are configured, and thresholds are set to trigger a risk warning mechanism. This model is used for risk detection at each stage of procurement. When an indicator exceeds a threshold, the system outputs a risk detection report, prompting a risk warning and preventing business flow. Users must address any anomalies based on the risk detection report or by tracing back to previous stages, ensuring no further risk warnings are issued before proceeding with the process. This approach ensures transparency and reliability in the procurement process, enhancing the stability of the enterprise's supply chain.

[0065] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0066] Figure 1 A flowchart illustrating an embodiment of a knowledge graph-based procurement risk detection method provided in this application is shown below. Figure 1 As shown, the knowledge graph-based procurement risk detection method provided in this embodiment specifically includes:

[0067] S101: Obtain the procurement data to be tested, which includes supplier information and goods information.

[0068] In modern business operations, procurement, as a key link in supply chain management, has a significant impact on operational efficiency and cost control. Through effective procurement risk management, companies can ensure the smooth operation of production processes, promptly identify and assess potential risks in the procurement process, and thus take preventative measures to reduce the probability and impact of risks, maintaining the stability of the company's supply chain. Procurement data is a crucial basis for companies to conduct procurement risk monitoring, encompassing various information related to procurement activities.

[0069] Specifically, this knowledge graph-based procurement risk detection model acquires relevant data from the procurement personnel's input into the enterprise's ERP system for this procurement transaction. This model is pre-constructed using a knowledge graph approach, based on historical procurement data and business data from multiple suppliers obtained from publicly available databases. This data includes supplier information and goods information. Goods information includes the name, quantity, unit price, quality standards, and supplier of the procured items; supplier information includes the supplier's name, company registration information, business license, relevant qualification certificates, credit records, commercial reputation information, delivery cycle, and other information.

[0070] S102: Use a pre-acquired risk detection model to perform risk detection on the procurement data and obtain risk detection results, which include detection results for at least one dimension of the supplier and / or goods.

[0071] In this step, the knowledge graph-based procurement risk detection model performs risk detection on the relevant data of the aforementioned actual procurement business, obtaining risk detection results. The risk detection results cover at least one dimension of the following suppliers and / or goods: supplier supply stability detection results, supplier delivery cycle detection results, supplier qualification detection results, supplier credit detection results, supplier social responsibility detection results, goods price detection results, goods quality detection results, goods quantity detection results, and invoice information detection results.

[0072] Specifically, the supplier's supply stability test results reflect whether there are risks to the supplier's on-time delivery rate; the supplier's delivery cycle test results reflect whether there are risks to the supplier's ability to meet delivery time while maintaining production efficiency; the supplier's qualification test results reflect whether there are risks to the supplier in terms of production capacity, financial status, legal compliance, and credit record; the supplier's credit test results reflect whether there are risks to the supplier in fulfilling contracts and ensuring timely, high-quality, and sufficient delivery of products or services; the supplier's social responsibility risk test results reflect whether there are risks to the supplier in fulfilling social responsibilities, such as in terms of environment, labor rights, business ethics, product quality and safety, and sustainable development; the goods price test results reflect whether there are unreasonable fluctuations in the market price of the goods; the goods quality test results reflect whether the quality and performance of the goods meet the expected standards and requirements; the goods quantity test results reflect whether the actual quantity of goods delivered matches the contract or order requirements; and the invoice information test results reflect whether the information recorded on the invoice is accurate, complete, and consistent with the actual procurement business, and whether there is a risk of duplicate invoicing.

[0073] S103: Generate a risk detection report based on the risk detection results;

[0074] The risk detection model is pre-built using a knowledge graph model based on historical procurement data and business data from multiple suppliers obtained from public databases.

[0075] In this step, the knowledge graph-based procurement risk detection model performs risk detection based on relevant data from the aforementioned actual procurement operations. The resulting risk detection results, along with a pre-set risk detection report template, generate a risk detection report. This report covers risk assessment results, early warning details, and risk mitigation measures. Procurement personnel can use this report to develop solutions for areas with procurement risks, providing proactive decision support.

[0076] Specifically, this knowledge graph-based procurement risk detection model also updates the visualization layout of the knowledge graph related to this procurement transaction using the data visualization tool D3.js based on the risk detection results. This allows procurement personnel to combine the risk detection report with the information to understand the distribution and trends of procurement risks. The mathematical model formula for knowledge graph visualization is as follows:

[0077]

[0078] in, Representing entities The probability distribution, Representing relations The probability distribution, Visualization of entities and nodes in a knowledge graph. Visualization of relationships and edges in a knowledge graph. Representing entities and entity The weights between them Representing entities and entity The distance between them This indicates the visual layout.

[0079] Based on this mathematical model, interactive functions such as dragging, scaling, and panning of each node were implemented through code logic, and the knowledge graph was updated in real time according to the user's procurement data.

[0080] The knowledge graph-based procurement risk detection method provided in this application acquires procurement data input by procurement personnel into the enterprise's ERP system and performs a comprehensive risk assessment using a pre-built risk detection model. This procurement risk detection model, constructed using a knowledge graph approach based on historical procurement data and supplier business data from public databases, can perform risk detection across multiple dimensions of suppliers and goods. After the detection results are generated, the system automatically generates a risk detection report. Furthermore, the procurement risk detection model utilizes the D3.js data visualization tool to update the knowledge graph visualization layout related to procurement operations. Procurement personnel can explore risk data in depth through interactive functions and make more accurate risk management decisions based on the real-time updated knowledge graph. Through this method, the procurement risk detection model can more comprehensively and dynamically display procurement risk information, helping enterprises to identify and respond to potential risks in the procurement process in a timely manner, optimize supply chain management, improve operational efficiency, reduce costs, and maintain the stability of the enterprise's supply chain.

[0081] Figure 2 A flowchart illustrating a second embodiment of a knowledge graph-based procurement risk detection method provided in this application is shown below. Figure 2 As shown, based on the above embodiment, before S102, the knowledge graph-based procurement risk control method further includes constructing a knowledge graph model, specifically including:

[0082] S201: Obtain the data set to be processed, which includes: historical procurement data stored in the database, and business data of multiple suppliers obtained from the public database; the multiple suppliers are suppliers that have cooperated with the company, as determined by the historical procurement data.

[0083] In this step, data scraping tools or Application Programming Interfaces (APIs) are used to acquire the dataset to be processed from historical data in the enterprise's ERP system database and from third-party public databases. The ERP system, as the core data management platform within the enterprise, provides rich material data, supplier information, pricing information, product quality data, procurement receipt and invoice information, covering all aspects from procurement planning to order execution, goods receipt and acceptance, and invoice processing. Third-party public databases provide information based on historical procurement data, including registration information, qualification certificates, credit records, and business reputation of suppliers with whom the enterprise has established partnerships. This provides crucial reference for the enterprise's procurement decisions, helping to assess the reliability and risk level of suppliers. By comprehensively utilizing the internal ERP system and external public data sources, the acquired dataset provides a foundation for subsequent risk detection and analysis.

[0084] S202: Preprocess the data in the dataset to be processed to obtain an intermediate dataset in JSON format with entity tags, relationship tags, and attribute tags added.

[0085] In this step, the data in the dataset to be processed is first cleaned to remove duplicate data, missing values, and outliers, resulting in a cleaned dataset. In procurement data, duplicate data may arise due to multiple entries, system errors, or data merging; missing values ​​can appear in any field, such as supplier information, price, or quantity; outliers may manifest as abnormally high or low prices, quantities, or delivery cycles. By identifying and deleting duplicate, missing, and outlier records, cleaned data is obtained. This method reduces data redundancy and improves the accuracy and reliability of the dataset.

[0086] Next, the data in the cleaned dataset is converted to JavaScript Object Notation (JSON) format, resulting in the converted dataset. In JSON format, the data in the converted dataset is represented as objects, consisting of key-value pairs, where the key is the property name and the value is the corresponding property value. The JSON format dataset has a clear structure, facilitating efficient management and utilization later.

[0087] Finally, entity tags, relationship tags, and attribute tags are added to the data in the format-converted dataset to obtain an intermediate dataset. This aims to give the format-converted JSON dataset a clear semantic structure, facilitating the construction of a knowledge graph. Entity tags are used to identify objects with independent meaning in the dataset, enabling them to exist as nodes in the knowledge graph, such as suppliers, goods, and invoices. Relationship tags are used to describe the associations between different entities, such as purchasing relationships and supply relationships. Attribute tags are used to describe the characteristics or attributes of entities, such as the name, address, and credit rating of a supplier entity; the name, price, and quality standards of a goods entity; and the unit price, quantity, and amount of an invoice.

[0088] The intermediate data set obtained by preprocessing the dataset to be processed improves the understandability and operability of the data, laying the foundation for the construction of knowledge graphs and subsequent risk detection.

[0089] S203: Entity identification is performed on the data in the intermediate dataset using NLP, and the relationships between entities are extracted to obtain a knowledge graph model.

[0090] In this step, NLP technology is first used to identify entities in the preprocessed intermediate dataset. Through steps such as word segmentation, part-of-speech tagging, and named entity recognition, the names of each entity are obtained, such as the specific names of suppliers and goods, so that entities such as suppliers, goods, and invoices can exist as nodes in the knowledge graph later.

[0091] Next, using NLP technology, the relationships between entities in the intermediate dataset are extracted, such as the supply relationship between suppliers and goods, the procurement relationship between purchase orders and goods, and the association between multiple suppliers and procurement tenders, so that the relationships between entities can exist as edges in the knowledge graph.

[0092] Finally, specific attribute values ​​are filled into the entities and entity relationships, such as the supplier's name, address, credit rating, etc., the name, price, quantity, etc. of the goods, the signing date and signatory of the purchase order, and the unit price, quantity, and total amount of the goods on the invoice.

[0093] The dataset, after entity identification, relation extraction, and attribute filling using NLP technology, is loaded into a knowledge graph platform system with a graph database as its core storage in the form of graph data. Using knowledge graph technology, entities, entity relationships, and attribute values ​​are integrated and associated to construct a knowledge graph model that includes all aspects of the procurement process, all entity elements, and their interrelationships.

[0094] S204: Obtain the risk detection model based on the knowledge graph model.

[0095] This step requires comprehensive consideration of all aspects of the procurement process to achieve all-round risk control, including the upstream, downstream, and downstream stages of the procurement operation, ensuring comprehensive risk management before, during, and after the procurement process. Based on each stage of the procurement process, a unified description template and calculation method for risk assessment rules at each stage are defined, and a risk detection model is constructed.

[0096] Specifically, in the forward processing stage, the risk detection model should focus more on supplier selection. Based on a knowledge graph, the model integrates multi-dimensional information such as real-time unit price, customer feedback, relevant qualifications, social responsibility, and credit rating to comprehensively evaluate suppliers. In the node stage, the risk detection model should focus more on purchase order management. Based on the knowledge graph, it verifies the unit price and quantity in purchase orders. Unit price verification involves checking whether the order price is consistent with the market price to prevent price anomalies or fraudulent activities; quantity verification ensures that the order quantity matches the order contract or actual needs, avoiding over-purchasing or inventory backlog. In the backward processing stage, the risk detection model should focus more on goods acceptance and invoice verification. During goods acceptance, the system verifies the quantity to ensure that the actual received goods match the order, preventing shortages or excesses. Invoice verification involves a comprehensive check of unit price, quantity, and amount to ensure that the invoice information matches the purchase order and the actual received goods, preventing duplicate invoicing or incorrect amounts.

[0097] The knowledge graph-based procurement risk detection method provided in this application achieves comprehensive risk management of the procurement process by constructing and utilizing a knowledge graph model. First, a dataset to be processed is obtained from the enterprise ERP system and a third-party public database. Then, the dataset is preprocessed to obtain an intermediate dataset, ensuring the accuracy and reliability of the data. Next, NLP technology is used to identify entities from the intermediate dataset and extract the relationships between them, populate specific attribute values, and load this information into a graph database to construct a knowledge graph model. This model comprehensively displays each stage of the procurement process, entity elements, and their interrelationships, facilitating the subsequent construction of a risk detection model based on the knowledge graph. This method not only improves the understandability and operability of the data but also provides a powerful tool for enterprise decision support, ensuring the stability and reliability of the supply chain.

[0098] Figure 3 A flowchart illustrating a third embodiment of a knowledge graph-based procurement risk detection method provided in this application is shown below. Figure 3 As shown, based on the above embodiments, in step S204, the knowledge graph-based procurement risk control method specifically includes:

[0099] S301: In response to the user's first configuration operation, configure the calculation method of supply stability indicators, delivery cycle indicators, qualification testing indicators, credit testing indicators and social responsibility testing indicators associated with the supplier.

[0100] In this step, based on the supplier's supply transaction records in the enterprise's ERP system's historical data, the calculation methods for supply stability indicators, namely on-time delivery rate and delivery cycle indicators, are configured. The calculation formulas are as follows:

[0101]

[0102] Based on business data from multiple suppliers obtained from publicly available third-party databases, the calculation methods for qualification testing indicators, credit testing indicators, and social responsibility testing indicators are configured. Specifically, the qualification testing indicators are used to quantify whether the supplier has corporate registration information, business license, and relevant qualification certificates; the credit testing indicators are used to quantify whether the supplier has any negative records in terms of credit rating and business honor information; and the social responsibility testing indicators are used to quantify whether the supplier has any negative records in fulfilling its social responsibilities, such as safety and environmental protection.

[0103] It should be noted that the quantitative indicators associated with suppliers are used to detect whether there are risks in the early stages of the purchase order process.

[0104] S302: In response to the user's second configuration operation, configure the calculation method of the price detection index, quantity detection index, quality detection index and invoice information detection index associated with the goods.

[0105] In this step, based on historical data from the enterprise's ERP system and relevant prices of goods in a third-party public database, the calculation methods for price detection indicators, quantity detection indicators, quality detection indicators, and invoice information detection indicators associated with the goods are configured.

[0106] Specifically, price monitoring indicators related to goods include market price change rate, cost profit margin, and purchase order unit price, etc., and the calculation formula is shown below:

[0107]

[0108] Based on the price monitoring indicators associated with goods, we continuously monitor the market prices and fluctuations of raw materials, semi-finished products, and finished products. It should be noted that the market price change rate and cost-profit ratio of the price monitoring indicators associated with goods are used to detect whether there are risks in the upstream stages of the purchase order; the unit price of the purchase order associated with the price monitoring indicators is used to detect whether there are risks in the purchase order process.

[0109] Quantity monitoring indicators associated with goods are used to quantify quantity discrepancies during the procurement process, compare the quantity of purchase orders with the quantity of purchase contracts in real time, and compare the quantity of purchase orders with the actual quantity received, in order to detect whether there are risks in the purchase order and receiving stages.

[0110] Quality inspection indicators associated with goods are used to quantify whether the quality of purchased goods meets standards. Based on historical data from the enterprise's ERP system and customer complaint data, quality inspection data, and goods characteristics from publicly available third-party databases, quality inspection indicators associated with goods are configured, including the calculation methods for quality complaint rate and goods pass rate. The calculation formulas are as follows:

[0111]

[0112] It should be noted that the quality inspection indicators associated with the goods are used to detect whether there are risks in the pre-purchase and receiving stages of the purchase order.

[0113] Indicators for detecting invoice information related to goods include unit price volatility, quantity discrepancy rate, and total amount discrepancy rate. Additionally, key information such as invoice number and transaction date is compared to identify duplicate invoices. The relevant calculation formulas are shown below:

[0114]

[0115] The calculation methods for the aforementioned quantitative indicators related to suppliers and goods are designed to accurately measure and compare the differences between actual data and expected or standard data of goods, ensuring that procurement meets established standards or requirements in key aspects such as suppliers, prices, quantities, quality, and invoice information, thereby enhancing the transparency and traceability of risk detection at each stage of procurement.

[0116] S303: In response to the user's third configuration operation, configure the risk detection rules for each metric.

[0117] In this step, based on industry standards, company requirements, and historical data from the ERP system, thresholds are set for quantitative indicators associated with suppliers and goods, specifically ∓10%~20% as the fluctuation tolerance range for these quantitative indicators. If the risk detection model identifies a quantitative indicator deviating from the set threshold, a procurement risk warning mechanism is immediately triggered.

[0118] Specifically, by defining and configuring exception categories and message categories, message configurations are formed for each stage of the procurement process. The exception category is configured as a quantitative indicator deviating from a set threshold, and the message category is configured as "error". That is, when the quantitative indicator of a certain stage of the procurement business exceeds the risk threshold, that is, exceeds the fluctuation tolerance range, the risk detection system displays the message category as "error", the system will directly block the current business operation, issue a risk warning, and suggest countermeasures such as modification, deletion, or re-creation.

[0119] S304: Based on the calculation methods of supply stability indicators, delivery cycle indicators, qualification testing indicators, credit testing indicators, and social responsibility testing indicators associated with suppliers, the calculation methods of price testing indicators, quantity testing indicators, quality testing indicators, and invoice information testing indicators associated with goods, the risk detection rules for each indicator, and the knowledge graph model, construct a risk detection model.

[0120] At this point, the risk detection model has been completed. It is used to detect whether there are risks in each stage of procurement. It covers all aspects of risk control, including pre-procurement comprehensive market assessment of suppliers, in-process procurement risk node calculation and assessment, and post-procurement risk visualization data analysis and management. It facilitates the rapid identification of risk causes and prevents risk documents from flowing to the payment stage.

[0121] The knowledge graph-based procurement risk detection method provided in this application defines various quantitative indicators and their calculation methods related to suppliers and goods in response to user configuration operations. These indicators include supply stability, delivery cycle, qualification detection, credit detection, social responsibility detection, price detection, quantity detection, quality detection, and invoice information detection, aiming to accurately measure and compare the differences between actual data and expected or standard data. Thresholds for the quantitative indicators are set, with ∓10%~20% as the fluctuation tolerance range, and a risk detection model is built based on this. When the detection model identifies an indicator deviating from the set threshold, a risk warning mechanism is immediately triggered, providing error prompts and suggested countermeasures through message definition configuration of anomaly categories and message categories. This method covers all stages of the procurement process—pre-procurement, during-procurement, and post-procurement—facilitating enterprises to quickly locate the causes of risks, prevent risky documents from flowing to the payment stage, enhance the transparency and traceability of each stage of procurement, and improve the efficiency and accuracy of procurement risk management.

[0122] Figure 4 This is a schematic diagram illustrating the principle of a knowledge graph-based risk detection method provided in an embodiment of this application. Figure 5 This is a schematic diagram illustrating a specific implementation of a knowledge graph-based risk detection method provided in this application. Figure 4 As shown, the overall logic of this solution is:

[0123] First, the system obtains a dataset from the enterprise ERP system and third-party public databases. This dataset undergoes preprocessing, including data cleaning (removing duplicates, missing values, and outliers), conversion to JSON format, and adding entity, relationship, and attribute tags to form an intermediate dataset and improve accuracy and reliability. Then, using NLP technology, entity identification, relationship extraction, and attribute filling are performed on the intermediate dataset. This identifies entities such as suppliers, goods, and invoices, extracts relationships between them, and fills in attributes such as supplier name, address, and credit rating. Next, this information is loaded into a graph database to construct a knowledge graph model. This model comprehensively displays all stages of the procurement process, entity elements, and their interrelationships, providing a foundation for subsequent risk detection. Following user configuration, the system sets the calculation methods for quantitative indicators associated with suppliers and goods, and defines risk detection rules for each indicator. Thresholds for these quantitative indicators are set based on industry standards, enterprise requirements, and historical data. When an indicator deviates from the set threshold, a procurement risk warning mechanism is immediately triggered, halting the current business operation and displaying a risk warning. These steps enable the construction of a risk detection model covering all aspects of procurement, achieving comprehensive risk management before, during, and after the process, and ensuring the stability and reliability of the supply chain.

[0124] When the constructed knowledge graph-based procurement risk detection model is used to detect risk points at each stage of the procurement process, if a warning is issued at the current stage (i.e., the quantitative indicator of the current stage exceeds the risk threshold), a risk warning is generated, and a risk detection report for the current node is output. The risk detection report includes the risk detection results, warning details, and risk handling measures. The warning details include at least one dimension of the aforementioned risky supplier and / or goods, along with corresponding indicator information. Simultaneously, the knowledge graph is updated based on the information in the risk detection report. At this point, the user must handle the risk at the current procurement node based on the risk detection report or by tracing back to previous stages, identifying anomalies, and modifying the corresponding data. Otherwise, the business document cannot be transferred to subsequent stages (the system will prompt that the operation failed and that a risk warning exists in the current business process) until there are no more risk warnings at the current stage, only then can it be transferred to the next stage. For example, when performing risk detection on the issuance of purchase invoices, if a quantity risk is detected in the purchase invoice, a quantity risk warning will be issued when the purchase invoice is posted in the current stage. A risk detection report will be output and the invoice will be frozen. If the upstream stage is the purchase receipt / purchase order, the quantity factor will be directly checked in the upstream stage. If the frozen invoice node with the risk warning is not processed, the business documents cannot be transferred to subsequent stages. A diagram illustrating purchase risk assessment and positioning is shown below. Figure 5 As shown.

[0125] In one specific implementation, if a risk warning is detected at the current receiving node during risk detection of the goods, the specific handling method is as follows:

[0126] The purchase order has a quantity and unit of 200 BAG and a price of 10 CNY / KG. The conversion between the unit of order and the price is 1KG = 2 BAG. The system calculates the order quantity as 100KG based on the price unit. Upon receipt of goods, if the quantity per unit is less than 95KG or more than 110KG, a quantity risk warning is issued when saving the material document, and a risk detection report is output, preventing subsequent business operations. Based on the risk detection report or tracing back to previous stages, the purchase order quantity is re-verified for reasonableness, anomalies are investigated, and corresponding data is modified until the current stage no longer has a risk warning.

[0127] In another specific implementation, if a risk warning is detected when performing risk detection on the unit price of a purchase order, the specific handling method is as follows:

[0128] Purchase orders and material master data maintain consistent settings for local currency, basic unit of measurement, and price unit, with a standard price of 100 for material master data. When the unit price of a line item corresponding to a purchase order is lower than 90 or higher than 120, a risk warning is triggered when the purchase order is saved, and a risk detection report is output, preventing subsequent business operations. Based on the risk detection report or tracing back to previous stages, the reasonableness of the material master data prices in those stages is re-verified, anomalies are investigated, and corresponding data is modified until the current stage no longer exhibits risk warnings.

[0129] Figure 6 A schematic diagram of a knowledge graph-based procurement risk detection device embodiment provided in this application is shown below. Figure 6 As shown, the knowledge graph-based procurement risk detection device 600 includes:

[0130] The first processing module 601 is used to acquire the procurement data to be tested, which includes supplier information and goods information.

[0131] The second processing module 602 is used to perform risk detection on the procurement data using a pre-acquired risk detection model to obtain risk detection results, which include detection results for at least one dimension of the supplier and / or goods.

[0132] The third processing module 603 is used to generate a risk detection report based on the risk detection results;

[0133] The risk detection model is pre-built using a knowledge graph model based on historical procurement data and business data from multiple suppliers obtained from public databases.

[0134] The knowledge graph-based procurement risk detection device provided in this embodiment is used to execute the technical solutions in the aforementioned method embodiments. Its implementation principle and technical effects are similar, and will not be described again here.

[0135] In one specific embodiment, the second processing module 602 is specifically used to: include at least one of the following in the risk detection result:

[0136] Supplier's supply stability test results;

[0137] Supplier's delivery cycle test results;

[0138] Supplier qualification test results;

[0139] Supplier credit test results;

[0140] The supplier's social responsibility test results;

[0141] Price inspection results of the goods;

[0142] The quality inspection results of the goods;

[0143] Quantity inspection results of the goods;

[0144] Invoice information detection results.

[0145] In one possible implementation, the knowledge graph-based procurement risk detection device 600 further includes: a fourth processing module 604, used for:

[0146] Obtain the set of data to be processed, which includes: historical procurement data stored in the database, and business data of multiple suppliers obtained from public databases; the multiple suppliers are suppliers that have cooperated with the company, as determined by the historical procurement data.

[0147] The data in the dataset to be processed is preprocessed to obtain an intermediate dataset in JSON format with entity tags, relationship tags, and attribute tags added.

[0148] By using NLP to identify entities in the intermediate dataset and extracting the relationships between entities, a knowledge graph model is obtained.

[0149] Based on the knowledge graph model, obtain the risk detection model.

[0150] In one specific embodiment, the fourth processing module 604 is specifically used for:

[0151] In response to the user's first configuration operation, configure the calculation methods for supply stability indicators, delivery cycle indicators, qualification testing indicators, credit testing indicators, and social responsibility testing indicators associated with suppliers;

[0152] In response to the user's second configuration operation, configure the calculation methods for price detection indicators, quantity detection indicators, quality detection indicators, and invoice information detection indicators associated with goods;

[0153] In response to the user's third configuration action, configure the risk detection rules for each metric;

[0154] Based on the calculation methods of supply stability indicators, delivery cycle indicators, qualification testing indicators, credit testing indicators, and social responsibility testing indicators associated with suppliers, as well as the calculation methods of price testing indicators, quantity testing indicators, quality testing indicators, and invoice information testing indicators associated with goods, the risk detection rules for each indicator, and a knowledge graph model, a risk detection model is constructed.

[0155] In one specific embodiment, the fourth processing module 604 is specifically used for:

[0156] The data in the dataset to be processed is cleaned to remove duplicate data, missing values, and outliers, resulting in a cleaned dataset.

[0157] The data in the cleaned dataset is converted to JSON format to obtain the converted dataset.

[0158] Add entity labels, relationship labels, and attribute labels to the data in the format-converted dataset to obtain an intermediate dataset.

[0159] In one specific embodiment, the second processing module 602 is further configured to:

[0160] The risk detection report includes the risk detection results, warning details, and risk handling measures; the warning information includes: at least one dimension of the supplier and / or goods that are at risk and the corresponding indicator information.

[0161] The knowledge graph-based procurement risk detection device provided in any of the above embodiments is used to execute the technical solutions in the aforementioned method embodiments. Its implementation principle and technical effect are similar, and will not be repeated here.

[0162] Figure 7 A schematic diagram of the structure of the computer device provided in the embodiments of this application, such as... Figure 7 As shown, the computer device 700 includes:

[0163] Processor, memory, and communication interface;

[0164] The memory stores the instructions that the computer executes;

[0165] The processor executes computer execution instructions stored in the memory, causing the processor to execute the technical solutions provided in the aforementioned method embodiments. The implementation principle and technical effect are similar, and will not be repeated here.

[0166] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0167] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0168] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0169] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the technical solutions in the aforementioned method embodiments. The implementation principle and technical effects are similar and will not be repeated here.

[0170] This application provides a computer-readable storage medium storing computer-executable instructions. When executed by a processor, these instructions are used to implement the technical solutions described in the aforementioned method embodiments. The implementation principle and technical effects are similar and will not be repeated here.

[0171] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0172] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an application-specific integrated circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0173] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0174] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0175] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0176] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0177] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0178] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A procurement risk detection method based on knowledge graphs, characterized in that, include: Obtain the procurement data to be tested, which includes supplier information and goods information; The procurement data is subjected to risk detection using a pre-acquired risk detection model to obtain risk detection results, which include detection results for at least one dimension of the supplier and / or goods. A risk detection report is generated based on the risk detection results; The risk detection model is pre-constructed using a knowledge graph model based on historical procurement data and business data from multiple suppliers obtained from public databases.

2. The method according to claim 1, characterized in that, The risk detection results include at least one of the following: Supplier's supply stability test results; Supplier's delivery cycle test results; Supplier qualification test results; Supplier credit test results; The supplier's social responsibility test results; Price inspection results of the goods; The quality inspection results of the goods; Quantity inspection results of the goods; Invoice information detection results.

3. The method according to claim 1 or 2, characterized in that, Before obtaining the risk detection result by using a pre-acquired risk detection model to perform risk detection on the procurement data, the method further includes: Obtain a set of data to be processed, which includes: historical procurement data stored in a database, and business data of multiple suppliers obtained from a public database; the multiple suppliers are suppliers that have cooperated with the company, as determined based on the historical procurement data. The data in the dataset to be processed is preprocessed to obtain an intermediate dataset in JSON format with entity tags, relationship tags, and attribute tags added. The data in the intermediate dataset is used to perform entity recognition and extract the relationships between entities to obtain a knowledge graph model. The risk detection model is obtained based on the knowledge graph model.

4. The method according to claim 3, characterized in that, The step of obtaining the risk detection model based on the knowledge graph model includes: In response to the user's first configuration operation, configure the calculation methods for supply stability indicators, delivery cycle indicators, qualification testing indicators, credit testing indicators, and social responsibility testing indicators associated with suppliers; In response to the user's second configuration operation, configure the calculation methods for price detection indicators, quantity detection indicators, quality detection indicators, and invoice information detection indicators associated with goods; In response to the user's third configuration action, configure the risk detection rules for each metric; Based on the calculation methods of supply stability indicators, delivery cycle indicators, qualification testing indicators, credit testing indicators, and social responsibility testing indicators associated with suppliers, the calculation methods of price testing indicators, quantity testing indicators, quality testing indicators, and invoice information testing indicators associated with goods, the risk detection rules for each indicator, and the knowledge graph model, the risk detection model is constructed.

5. The method according to claim 3, characterized in that, The process of preprocessing the data in the dataset to be processed to obtain an intermediate dataset in JSON format with added entity tags, relationship tags, and attribute tags includes: The data in the dataset to be processed is cleaned to remove duplicate data, missing values ​​and outliers, resulting in a cleaned dataset. The data in the cleaned dataset is converted to JSON format to obtain the converted dataset. Entity tags, relationship tags, and attribute tags are added to the data in the format-converted dataset to obtain the intermediate dataset.

6. The method according to claim 1 or 2, characterized in that, The risk detection report includes the risk detection results, warning details, and risk handling measures; The early warning information includes: at least one dimension of the supplier and / or goods that are at risk, and the corresponding indicator information.

7. A procurement risk detection device based on knowledge graphs, characterized in that, include: The first processing module is used to acquire the procurement data to be tested, which includes supplier information and goods information; The second processing module is used to perform risk detection on the procurement data using a pre-acquired risk detection model to obtain risk detection results, which include detection results for at least one dimension of the supplier and / or goods. The third processing module is used to generate a risk detection report based on the risk detection results; The risk detection model is pre-constructed using a knowledge graph model based on historical procurement data and business data from multiple suppliers obtained from public databases.

8. A computer device, characterized in that, include: Processor, memory, and communication interface; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the knowledge graph-based procurement risk detection method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the knowledge graph-based procurement risk detection method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, It includes a computer program that, when executed by a processor, implements the knowledge graph-based procurement risk detection method as described in any one of claims 1 to 6.

Images