Purchase auditing process control method and system and electronic equipment

Abstract

The invention provides a procurement auditing process control method and system and electronic equipment, and relates to the field of procurement process auditing control, and the method comprises the steps: constructing a knowledge graph of a procurement process to determine a compliance verification strategy of the auditing process, and carrying out the automatic execution and reasoning of the procurement process through the processing of a structured rule of the knowledge graph; besides, according to the method, the approval strategy of the auditing process is determined by adopting the compliance verification result and the risk assessment result of the purchase data, so that the accuracy of the auditing process is improved, complex risk points in the purchase auditing process can be identified, and the compliance condition of suppliers can be cooperatively monitored in a full period; therefore, full-process coverage of the purchase auditing process is realized.

Description

Technical Field

[0001] This invention relates to the field of procurement process review and control, and in particular to a procurement review process control method, system, and electronic device. Background Technology

[0002] Corporate procurement is a major channel for corporate cash outflows, involving the highest amount of money in all aspects of corporate operations, thus requiring strict auditing and management. In specific scenarios, corporate procurement processes involve various complex and cumbersome procedures, primarily related to laws and regulations, company policies, financial standards, and audit requirements. However, in the process of managing the procurement auditing process, procurement departments, finance, and auditing personnel mainly rely on manual review of procurement documents. This approach suffers from low efficiency, inconsistent quality, high costs, and delays. Furthermore, existing auditing and verification rules are simple, with weak auditing capabilities, and cannot identify complex risk points (such as related-party transactions, bid rigging, and fraudulent procurement). Moreover, existing procurement auditing processes cannot provide full coverage and lack self-learning capabilities. Summary of the Invention

[0003] In view of this, the purpose of this invention is to provide a procurement review process control method, system, and electronic device. This method determines the compliance verification strategy of the review process by constructing a knowledge graph of the procurement process, thereby using the structured rule processing of the knowledge graph to automatically execute and reason about the procurement process. In addition, this method uses the compliance verification results and risk assessment results of the procurement data to determine the approval strategy of the review process, thereby improving the accuracy of the review process, identifying complex risk points in the procurement review process, and enabling full-cycle collaborative monitoring of the supplier's compliance status, thereby achieving full-process coverage of the procurement review process.

[0004] In a first aspect, embodiments of the present invention provide a method for controlling a procurement review process, the method comprising: A knowledge graph corresponding to the procurement process is constructed based on entity parameters, rule parameters, relationship parameters, and risk parameters corresponding to the procurement data. Knowledge graphs are used to determine the compliance verification strategies corresponding to the audit process, and the compliance verification results corresponding to the procurement data are determined through the compliance verification strategies. Based on the risk identification data corresponding to the procurement process, determine the risk assessment strategy corresponding to the audit process, and determine the risk assessment result corresponding to the procurement data through the risk assessment strategy. The approval strategy corresponding to the review process is determined by the compliance verification results and risk assessment results. Based on the approval strategy, determine the decision support data corresponding to the procurement data, and use the decision support data to determine the decision approval data corresponding to the procurement data.

[0005] Optionally, the steps for constructing a knowledge graph corresponding to the procurement process based on entity parameters, rule parameters, relationship parameters, and risk parameters corresponding to the procurement data include: Entity parameters are determined based on supplier data, product data, contract data, order data, invoice data, employee data, and department data contained in the procurement data; The rule parameters are determined based on the procurement authority data, supplier access data, price compliance data, contract terms data, invoice compliance data, and payment approval data contained in the procurement data; The relationship parameters are determined based on the supplier association data, market price association data, product category association data, employee permission association data, and blacklist association data contained in the procurement data; Risk parameters are determined based on the related-party transaction risk data, bid rigging risk data, fraudulent procurement risk data, and abnormal pricing risk data contained in the procurement data; A knowledge graph corresponding to the procurement process is constructed using entity attribute parameters corresponding to entity parameters, rule logic parameters corresponding to rule parameters, relation semantic parameters corresponding to relation parameters, and risk extraction parameters corresponding to risk parameters.

[0006] Optionally, the steps of using knowledge graphs to determine the compliance verification strategy corresponding to the audit process, and determining the compliance verification result corresponding to the procurement data based on the compliance verification strategy, include: Based on the applicant's procurement authority, approval process authority, and monetary authority determined by the knowledge graph, the corresponding authority verification strategy for the review process is determined. The supplier verification strategy corresponding to the review process is determined based on the supplier access data, supplier blacklist data, supplier qualification data, and related conflict data identified by the knowledge graph. Based on the market price comparison results, historical price comparison results, and price comparison record results determined by the knowledge graph, the price verification strategy corresponding to the audit process is determined. Based on the contract verification data, process verification data, and invoice verification data determined by the knowledge graph, the compliance verification strategy corresponding to the audit process is determined. The compliance verification strategy corresponding to the audit process is determined by utilizing the authorization verification strategy, supplier verification strategy, price verification strategy, and compliance verification strategy. The compliance verification results for the procurement data under the permission verification strategy, supplier verification strategy, price verification strategy, and compliance verification strategy are obtained through the compliance verification strategy.

[0007] Optionally, the steps of determining the risk assessment strategy corresponding to the audit process based on the risk identification data corresponding to the procurement process, and determining the risk assessment result corresponding to the procurement data through the risk assessment strategy, include: Based on the procurement process, a relationship graph between suppliers and employees is determined, and the related risk identification data corresponding to the relationship graph is used to determine the related transaction assessment strategy corresponding to the audit process. Obtain the similarity of quotation features among suppliers in the procurement process, and use the data on bid rigging risk identification corresponding to the quotation feature similarity to determine the bid rigging and collusion assessment strategy corresponding to the audit process; Based on the reasonableness data, authenticity data, logistics tracking data and acceptance record data of suppliers in the procurement process, determine the false procurement risk identification data corresponding to the procurement process, and use the false procurement risk identification data to determine the false procurement assessment strategy corresponding to the audit process; By using the corresponding product pricing data and historical product data under the procurement process, we can determine the pricing anomaly risk identification data for products, and use the pricing anomaly risk identification data to determine the pricing anomaly assessment strategy corresponding to the audit process. By using the procurement data and turnover data of suppliers in the procurement process, fraud risk identification data is determined, and fraud assessment strategies corresponding to the audit process are determined using the fraud risk identification data. The risk assessment strategy for the review process is determined based on the assessment strategies for related-party transactions, bid rigging, fraudulent procurement, abnormal pricing, and fraud. Risk assessment results were obtained for procurement data under the following risk assessment strategies: related party transaction assessment strategy, bid rigging assessment strategy, fraudulent procurement assessment strategy, abnormal pricing assessment strategy, and fraud assessment strategy.

[0008] Optionally, the steps for determining the approval strategy corresponding to the audit process based on compliance verification results and risk assessment results include: The compliance status assessment results for the procurement process are determined by using the compliance verification results, and the risk level assessment results for the procurement process are determined by using the risk assessment results. Decision support information for the procurement process is generated based on the compliance status assessment results and risk level assessment results. Based on decision support information, historical statistical data, supplier profile data, price comparison data, and risk analysis data are determined for the procurement process. The approval strategy for the review process is determined based on historical statistical data, supplier profile data, price comparison data, and risk analysis data.

[0009] Optionally, the steps of determining the decision support data corresponding to the procurement data based on the approval strategy, and using the decision support data to determine the decision approval data corresponding to the procurement data, include: Decision support data for procurement data is determined by using historical statistical data, supplier profile data, price comparison data, and risk analysis data corresponding to the approval strategy. Decision support data is used to determine the approval and decision results corresponding to the approval strategy, and the approval results, decision results, historical statistics, supplier profile data, price comparison data and risk analysis data are saved to a pre-set database; The decision-making and approval data corresponding to the procurement data are determined based on the approval and decision-making results.

[0010] Optionally, before the step of constructing a knowledge graph corresponding to the procurement process based on entity parameters, rule parameters, relationship parameters, and risk parameters corresponding to the procurement data, the method further includes: Obtain the document files corresponding to the procurement process and the approval process, and determine the corresponding parsing strategy for the document files; The target information contained in the document file is extracted using a parsing strategy; the target information includes at least: information about the procuring entity, information about the procuring object, time information, contract terms information, and invoice information. Procurement data is determined based on target information.

[0011] Optionally, after determining the decision support data corresponding to the procurement data based on the approval strategy, and using the decision support data to determine the decision approval data corresponding to the procurement data, the method further includes: Real-time acquisition of supplier-related information change data, litigation data, public opinion data, and financial monitoring data; The risk level of suppliers is determined by using information change data, litigation data, public opinion data, and financial monitoring data. The approval strategy is updated based on the risk level results.

[0012] Secondly, the present invention provides a procurement review process control system, the system comprising: The knowledge graph construction module is used to build a knowledge graph corresponding to the procurement process based on entity parameters, rule parameters, relationship parameters and risk parameters corresponding to procurement data. The compliance verification result determination module is used to determine the compliance verification strategy corresponding to the audit process using a knowledge graph, and then determine the compliance verification result corresponding to the procurement data based on the compliance verification strategy. The risk assessment result determination module is used to determine the risk assessment strategy corresponding to the audit process based on the risk identification data corresponding to the procurement process, and to determine the risk assessment result corresponding to the procurement data through the risk assessment strategy. The approval strategy determination module is used to determine the approval strategy corresponding to the review process based on compliance verification results and risk assessment results. The decision review and control module is used to determine the decision support data corresponding to the procurement data based on the approval strategy, and to determine the decision approval data corresponding to the procurement data using the decision support data.

[0013] Thirdly, embodiments of the present invention also provide an electronic device, which includes a processor and a memory, the memory storing computer-executable instructions that can be executed by the processor, and the processor executing the computer-executable instructions to implement the steps of the procurement review process control method provided in the first aspect.

[0014] Fourthly, embodiments of the present invention also provide a storage medium storing computer-executable instructions, which, when invoked and executed by a processor, cause the processor to implement the steps of the procurement review process control method provided in the first aspect.

[0015] This invention provides a procurement review process control method, system, and electronic device. In controlling an enterprise's procurement review process, the method first constructs a knowledge graph corresponding to the procurement process based on entity parameters, rule parameters, relationship parameters, and risk parameters corresponding to the procurement data. Then, it uses the knowledge graph to determine the compliance verification strategy corresponding to the review process, and uses this strategy to determine the compliance verification result corresponding to the procurement data. Next, it determines the risk assessment strategy based on the risk identification data corresponding to the procurement process, and uses this strategy to determine the risk assessment result. Then, it determines the approval strategy based on the compliance verification result and the risk assessment result. Finally, it determines the decision support data corresponding to the procurement data based on the approval strategy, and uses this decision support data to determine the decision approval data corresponding to the procurement data. This method determines the compliance verification strategy of the review process by constructing a knowledge graph of the procurement process, thereby automatically executing and reasoning about the procurement process using the structured rule processing of the knowledge graph. Furthermore, the method uses the compliance verification result and risk assessment result of the procurement data to determine the approval strategy of the review process, improving the accuracy of the review process, identifying complex risk points in the procurement review process, and enabling full-cycle collaborative monitoring of supplier compliance, thus achieving full-process coverage of the procurement review process.

[0016] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0018] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 A flowchart of a procurement review process control method provided in an embodiment of the present invention; Figure 2 A flowchart of step S101 of a procurement review process control method provided in an embodiment of the present invention; Figure 3 A flowchart of step S102 of a procurement review process control method provided in an embodiment of the present invention; Figure 4 A flowchart of step S103 of a procurement review process control method provided in an embodiment of the present invention; Figure 5 A flowchart of step S104 of a procurement review process control method provided in an embodiment of the present invention; Figure 6 A flowchart of step S105 of a procurement review process control method provided in an embodiment of the present invention; Figure 7 A flowchart preceding step S101 of a procurement review process control method provided in an embodiment of the present invention; Figure 8 A flowchart following step S105 of a procurement review process control method provided in an embodiment of the present invention; Figure 9 This is a schematic diagram of the structure of a procurement review process control system provided in an embodiment of the present invention; Figure 10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0020] icon: 910 - Knowledge Graph Construction Module; 920 - Compliance Verification Result Determination Module; 930 - Risk Assessment Result Determination Module; 940 - Approval Strategy Determination Module; 950 - Decision Review and Control Module; 101 - Processor; 102 - Memory; 103 - Bus; 104 - Communication interface. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] To facilitate understanding of this embodiment, a procurement review process control method disclosed in this embodiment of the invention will first be introduced, such as... Figure 1 As shown, the method includes: Step S101: Construct a knowledge graph corresponding to the procurement process based on the entity parameters, rule parameters, relationship parameters, and risk parameters corresponding to the procurement data.

[0023] Based on multi-dimensional core parameters involved in the entire procurement process, a structured procurement compliance knowledge graph is constructed. Entity parameters cover core business objects such as suppliers, products, contracts, orders, invoices, employees, and departments; rule parameters integrate various compliance standards, including laws and regulations (such as the Bidding Law), company policies (procurement authority, approval processes), financial regulations (budget control, invoice compliance), and audit requirements (matching of three documents, price reasonableness); relationship parameters clarify the logical connections between entities, including supplier-blacklist relationships, employee-authority relationships, product-category relationships, and price-market price relationships; risk parameters focus on hidden risk types such as related-party transaction risks, conflict of interest risks, bid rigging risks, and fraudulent procurement risks. During construction, ontology modeling technology can be used to define entity attributes, relational semantics, and rule logic. Multi-source data can be integrated through information extraction and knowledge fusion technologies, supporting complex reasoning and dynamic updates to ensure the knowledge graph comprehensively and accurately maps the core elements of procurement compliance management.

[0024] Step S102: Use knowledge graphs to determine the compliance verification strategy corresponding to the audit process, and use the compliance verification strategy to determine the compliance verification result corresponding to the procurement data.

[0025] Leveraging a pre-built procurement process knowledge graph, a multi-dimensional compliance verification strategy is automatically generated, covering the core compliance requirements of the entire procurement process. This strategy includes: authorization compliance verification (verifying the applicant's procurement authority, whether the approval process conforms to the authorization matrix, and whether the amount exceeds the authorized limit); supplier compliance verification (checking whether the supplier is on the access list / blacklist, whether the qualifications are valid, and whether there are related-party transactions or conflicts of interest); price compliance verification (comparing the procurement price with the market price / historical price, identifying abnormal premiums, and checking price comparison records); contract compliance verification (verifying whether the contract terms meet the template requirements, the reasonableness of payment terms, and whether there are legal risk clauses); invoice compliance verification (verifying the authenticity of invoices, matching the order, warehouse receipt, and invoice, and ensuring consistency between the tax rate and the header); and process compliance verification (investigating violations such as procurement before approval, splitting orders to circumvent approval, and failure to tender when required). The verification process is then automatically executed according to structured rules, with detailed records of the verification results for each rule. Non-compliant items are automatically marked and the cited compliance basis is clearly stated, forming clear and traceable procurement data compliance verification results.

[0026] Step S103: Determine the risk assessment strategy corresponding to the audit process based on the risk identification data corresponding to the procurement process, and determine the risk assessment result corresponding to the procurement data through the risk assessment strategy.

[0027] Based on risk identification data accumulated in the procurement process (including historical fraud cases, multi-dimensional supplier characteristic data, and procurement business operation data), a multi-dimensional AI risk assessment strategy is constructed. This strategy integrates specialized risk models for related-party transaction identification, bid rigging identification, fraudulent procurement identification, abnormal pricing identification, and fraud pattern identification. It identifies hidden related-party transactions by analyzing multi-dimensional features such as enterprise registration data, equity relationships, personnel appointments, addresses, and phone numbers; it judges bid rigging behavior based on features such as the similarity of multiple supplier quotes, IP addresses, bidding times, and contact information; it identifies cases of fictitious procurement to embezzle funds by combining data such as the rationality of procurement needs, supplier authenticity, logistics tracks, and acceptance records; it accurately identifies abnormally high or low-priced procurements based on market prices, historical prices, and prices of similar products; and it matches fraud patterns such as frequent small-amount procurement to circumvent approvals and monopolistic procurement by specific suppliers by learning from the characteristics of historical fraud cases. Each risk model outputs a specific risk score, which is then combined to form an overall risk assessment result corresponding to the procurement data, clarifying the risk level and risk type.

[0028] Step S104: Determine the approval strategy corresponding to the audit process based on the compliance verification results and risk assessment results.

[0029] By combining compliance verification results with risk assessment results, an intelligent approval strategy is constructed to achieve precise routing and risk control of the approval process. Specifically, for low-risk and fully compliant procurement transactions (such as routine office supplies purchases, small amounts, and high supplier credit ratings), automatic or rapid approval processes are implemented, significantly improving approval efficiency. For medium-risk or partially non-compliant procurement transactions (such as new supplier collaborations, prices slightly higher than market benchmarks, and contract terms requiring further confirmation), the approval process is intelligently routed to the appropriate approver (such as procurement manager, finance personnel, and legal personnel), and risk warnings and explanations of non-compliance items are simultaneously pushed. For high-risk or seriously non-compliant procurement transactions (such as suppliers on blacklists, clear related-party transactions, abnormally high prices, and contracts with significant legal risks), the approval process can be forcibly intercepted, and the matter can be pushed to the compliance or audit departments for special investigation. Subsequent processes can only proceed after the issues are rectified or compliance is confirmed, thus mitigating compliance risks from the source.

[0030] Step S105: Determine the decision support data corresponding to the procurement data based on the approval strategy, and use the decision support data to determine the decision approval data corresponding to the procurement data.

[0031] Based on established approval strategies, the system integrates relevant data from the entire procurement process to generate comprehensive decision support data, providing precise decision-making assistance to approvers. This decision support data includes historical procurement records (purchase prices of similar or identical goods, evaluations of partner suppliers), market price comparison analysis, detailed risk assessment reports (including risk points, causes, and scope of impact), compliance verification details, and other core information. Approvers, based on this decision support data and the actual business situation, make a comprehensive judgment to determine the corresponding decision approval data for the procurement data: for automatically approved transactions, an approval certificate is directly generated; for transactions requiring manual approval, the approver records their approval opinion (agree, reject, return for modification), forming formal decision approval data; for transactions forcibly blocked, an investigation initiation notice and a process suspension certificate are generated simultaneously, ensuring the approval process is traceable and verifiable. Simultaneously, the system records all process data and the review trajectory, constructing a complete chain of evidence to provide data support for subsequent traceability.

[0032] Optionally, step S101 involves constructing a knowledge graph corresponding to the procurement process based on entity parameters, rule parameters, relationship parameters, and risk parameters corresponding to the procurement data, such as... Figure 2 As shown, it includes: Step S201: Determine entity parameters based on the supplier data, product data, contract data, order data, invoice data, employee data, and department data contained in the procurement data.

[0033] The determination of entity parameters focuses on the core participants and key business entities throughout the entire procurement process. This allows for the precise extraction and analysis of specific data corresponding to seven core entity categories from the procurement data. Supplier data includes supplier name, qualification certificate information, access status, business registration information, and legal records; Product data includes product name, specifications, category, market reference price, historical purchase price, etc. Contract data includes contract number, signing date, payment method, warranty period, and liability clauses for breach of contract; Order data includes order number, quantity purchased, unit price, total price, delivery deadline, etc. The invoice data includes invoice number, tax number, invoice amount, tax amount, header information, and invoice authenticity verification results. Employee data includes the name of the person in charge, their department, procurement authority level, and historical transaction records; Departmental data includes department names, functional divisions, and scope of approval authority.

[0034] By structuring and organizing this data, the core components of the entity layer of the knowledge graph are identified.

[0035] Step S202: Determine the rule parameters based on the procurement authority data, supplier access data, price compliance data, contract terms data, invoice compliance data, and payment approval data contained in the procurement data.

[0036] The rule parameters are based on the full-dimensional requirements of procurement compliance management and can extract and integrate six categories of rule-related data from procurement data: The procurement authority data includes procurement authority matrices for different positions / departments, amount approval thresholds, multi-level approval process specifications, etc., which are in line with the authorization requirements in the company's system; Supplier access data covers supplier access qualification standards, blacklist identification rules, qualification validity verification specifications, etc., and is aligned with laws and regulations and supplier management requirements. Price compliance data includes the deviation threshold between the purchase price and the market price / historical price, the requirements for the price comparison process, and the criteria for judging abnormal premiums, in response to the price reasonableness requirements in the audit. The contract terms data includes templates for essential contract terms, rules for identifying unfavorable terms, and standards for determining legal risk terms, which comply with legal regulations and compliance review requirements. The invoice compliance data covers rules for verifying the authenticity of invoices, matching standards for the three documents (order, warehouse receipt, and invoice), applicable tax rates, and rules for verifying the consistency of the invoice header, which are in line with financial regulations. Payment approval data includes payment period requirements, payment prerequisites (such as acceptance upon completion), and rules for blocking abnormal payments, which connects the financial approval process with company policies.

[0037] Step S203: Determine the relationship parameters based on the supplier association data, market price association data, product category association data, employee permission association data, and blacklist association data contained in the procurement data.

[0038] Relationship parameters focus on the core relationship logic between entities, and can extract and define the relationship types corresponding to five major categories of key relationship data from procurement data: Supplier association data clearly defines the attribution relationship between suppliers and blacklists, the hidden relationships between suppliers and employees / other companies (such as equity relationships, personnel employment relationships), and the supply relationship between suppliers and goods; Market price correlation data defines the comparison relationship between the purchase price and the market benchmark price and historical purchase prices (e.g., higher than, lower than, or the same as). Product category association data determines the hierarchical relationship between a single product and its category, and between a category and its parent category; The data linking employee permissions clearly defines the affiliation between employees and their departments, and the matching relationship between employees and their corresponding purchasing permissions; The blacklist-related data covers the inclusion / removal relationship between suppliers and the blacklist, the correspondence between blacklist rules and violations, and comprehensively maps the business logic and compliance relationships between entities.

[0039] Step S204: Determine the risk parameters based on the related party transaction risk data, bid rigging risk data, fraudulent procurement risk data, and abnormal pricing risk data contained in the procurement data.

[0040] Risk parameters revolve around the core hidden risk types in procurement operations, and four main categories of risk-related characteristic data can be extracted from procurement data: Related party transaction risk data includes multi-dimensional correlation characteristics such as business registration and equity data of suppliers and employees / other companies, cross-employment data of personnel, and overlapping address and telephone data; The data on bid rigging and collusion risks covers abnormal characteristics such as the similarity of quotes from multiple suppliers, the correlation of bidding IP addresses, the synchronization of bidding times, and the overlap of contact information; Data related to the risk of fraudulent procurement includes data on the rationality assessment of procurement needs, data on the authenticity verification of suppliers, logistics tracking records, and data on the completeness of acceptance documents. Abnormal pricing risk data includes data on the deviation between purchase price and market price, historical price, and price of similar products, as well as data on the sufficiency of pricing basis.

[0041] By analyzing these risk characteristic data, the core identification dimensions of the knowledge graph risk layer are clarified.

[0042] Step S205: Construct a knowledge graph corresponding to the procurement process using entity attribute parameters corresponding to entity parameters, rule logic parameters corresponding to rule parameters, relation semantic parameters corresponding to relation parameters, and risk extraction parameters corresponding to risk parameters.

[0043] Using ontology modeling techniques, the aforementioned parameters are structurally integrated to construct a procurement process knowledge graph. First, core attributes and data types of each entity are defined based on entity attribute parameters (such as supplier qualification validity period, employee permission level, and product category attributes). Second, the execution logic and priority of the rule layer are clarified based on rule logic parameters (such as nested conditions like "purchase amount > 100,000 and the supplier is a new supplier requiring director approval" and multi-dimensional relational logic). Third, the types of relationships and constraints between entities are defined through relational semantic parameters (such as explicit relational semantics like "supplier - belongs to - blacklist" and "employee - has - procurement authority"). Simultaneously, core indicators for risk identification are extracted based on risk extraction parameters (such as equity relationship characteristics of related-party transactions and price similarity characteristics of bid rigging). Finally, combining information extraction and knowledge fusion technologies (such as entity alignment and merging of duplicate entities, and knowledge completion and prediction of missing relationships), the parameters of the four dimensions of entities, rules, relationships, and risks are organically integrated to construct a structured procurement process knowledge graph that supports complex reasoning and dynamic updates, providing a foundation for subsequent compliance verification and risk identification.

[0044] Specifically, the knowledge graph is constructed using information extraction and knowledge fusion. Entity identification: BERT-NER is used to identify entities such as regulatory clauses, compliance requirements, procurement categories, and supplier qualifications. Relationship extraction: BERT+RelationClassifier is used to identify relationships (e.g., "Medical device procurement requires a medical device business license"). Knowledge fusion: Entity alignment technology (based on name similarity and attribute similarity) is used to merge duplicate entities, and knowledge completion (TransE / RotatE) is used to predict missing relationships. Graph storage: Neo4j is used, with 100,000+ nodes and 500,000+ relationships. Inference engine: Cypher query + rule-based reasoning is used, such as "MATCH (p:procurement)-[:required]->(c:qualification)<-[:possess]-(s:supplier) WHERE p.id=$procurementID RETURN s" to query compliant suppliers. Dynamic updates: Regulatory websites are monitored, and Web Scraping+NLP is used to extract new regulations and automatically update the graph.

[0045] Optionally, step S102 involves using a knowledge graph to determine the compliance verification strategy corresponding to the audit process, and then determining the compliance verification result corresponding to the procurement data based on the compliance verification strategy. Figure 3 As shown, it includes: Step S301: Based on the applicant's procurement authority, approval process authority, and monetary authority determined by the knowledge graph, determine the corresponding authority verification strategy for the review process.

[0046] Based on the entity layer (employees, departments), rule layer (procurement authority rules), and relationship layer (employee-authority relationship, department-approval authority relationship) of the procurement process knowledge graph, the core dimensions and standards of authorization verification are clarified, and an authorization verification strategy is formed.

[0047] Specifically, this includes: First, verifying whether the applicant has the purchasing authority for the corresponding product category and amount range, and matching the relationship between employees and purchasing authority in the knowledge graph; second, verifying whether the approval process conforms to the company's authorization matrix, ensuring that the approval nodes, approver levels and process specifications defined in the rule layer are consistent; and third, setting a threshold for the amount of authority, and judging whether the current purchasing amount exceeds the authority boundaries of the applicant and the approval process based on the corresponding rules of "purchasing amount - approval level" in the knowledge graph, ensuring that the authority verification fully covers the matching requirements of "person-authority-matter".

[0048] Step S302: Determine the supplier verification strategy corresponding to the audit process based on the supplier access data, supplier blacklist data, supplier qualification data and related conflict data determined by the knowledge graph.

[0049] Based on the entity layer (suppliers), rule layer (supplier admission rules), relationship layer (supplier-blacklist relationship, supplier-related enterprise relationship), and risk layer (related transaction risk, conflict of interest risk) of the knowledge graph, a multi-dimensional supplier verification strategy is constructed.

[0050] Specific requirements include: verifying whether suppliers are included in the enterprise access list and excluding suppliers on the blacklist; verifying the authenticity, validity and expiration of supplier qualification documents (such as business licenses and industry qualification certificates); and identifying whether there are any hidden related-party transactions or conflicts of interest between suppliers and enterprise employees or other partners by analyzing multi-dimensional features such as industrial and commercial equity data, cross-employment information of personnel, and address and telephone correlation, so as to ensure that the supplier's cooperation qualifications are compliant.

[0051] Step S303: Based on the market price comparison results, historical price comparison results, and price comparison record results determined by the knowledge graph, determine the price verification strategy corresponding to the audit process.

[0052] Based on the knowledge graph, the entity layer (products, purchase orders, historical purchase records), relationship layer (price-market price relationship, product-category price relationship), and rule layer (price compliance rules) are used to clarify the core standards and processes for price verification and form a price verification strategy.

[0053] Specifically, this includes: comparing the purchase price with the market benchmark price and historical purchase prices of similar products stored in the knowledge graph, and setting a reasonable deviation threshold (such as ±10%); determining whether there are abnormal premiums or low-price purchases, and focusing on verifying price deviations without reasonable basis; at the same time, requiring verification of whether there are complete and compliant inquiry and comparison records in the procurement process, ensuring that the price formation process complies with the company's system requirements, and controlling price risks from the dual dimensions of "price reasonableness + process compliance".

[0054] Step S304: Determine the compliance verification strategy corresponding to the audit process based on the contract verification data, process verification data, and invoice verification data determined by the knowledge graph.

[0055] Based on a knowledge graph-based entity layer (contracts, orders, invoices), rule layer (contract terms rules, invoice compliance rules, process compliance rules), and relationship layer (order-contract-invoice association), this system integrates the verification requirements for three core data types—contracts, processes, and invoices—to form a specialized compliance verification strategy. Contract verification focuses on the completeness of clauses (whether they conform to the company template), the reasonableness of payment terms, the clarity of liability for breach of contract, and the absence of clauses with adverse legal risks. Process verification focuses on checking for violations such as "procurement before approval," "splitting orders to circumvent high-amount approvals," and "failure to tender when required." Invoice verification covers verifying the authenticity of invoices, matching the order, warehouse receipt, and invoice documents, ensuring the correctness of tax rate application, and ensuring the consistency between the invoice header and the purchasing entity, thus guaranteeing the compliance and relevance of these three types of business data.

[0056] Step S305: Determine the compliance verification strategy corresponding to the audit process using the authorization verification strategy, supplier verification strategy, price verification strategy, and compliance verification strategy.

[0057] The permission verification strategy, supplier verification strategy, price verification strategy, and contract-process-invoice specific compliance verification strategy formed in steps S301 to S304 are systematically integrated to construct a comprehensive compliance verification strategy system covering the entire procurement process. During the integration process, the execution order of each specific strategy is clearly defined (e.g., basic permission and supplier qualification verification first, followed by in-depth verification of price, contracts, etc.), the priority determination of verification results (e.g., serious violations are marked first), and the cross-strategy correlation verification logic (e.g., invoice verification requires correlation with order and contract data). This ensures that the verification strategies are comprehensive, conflict-free, and fully map the compliance rules and correlation logic in the knowledge graph.

[0058] Step S306: Obtain the compliance verification results corresponding to the procurement data under the permission verification strategy, supplier verification strategy, price verification strategy, and compliance verification strategy respectively through the compliance verification strategy.

[0059] Based on the integrated comprehensive compliance verification strategy, automated verification is performed on procurement data dimension by dimension, outputting the verification results under each specific strategy and summarizing them into a final compliance verification report. Specifically, this includes: permission verification results clearly indicating whether the applicant's permissions match, whether the approval process is compliant, and highlighting non-compliance items such as exceeding authorized amounts and missing approval nodes; supplier verification results reflecting the supplier's access status, qualification validity, and associated risks, listing issues such as blacklist matching and expired qualifications; price verification results presenting price comparison deviations, abnormal premium determination, and the completeness of price comparison records; and contract-process-invoice verification results explaining the compliance of clauses, process standardization, matching of the three documents (contract, process, and invoice), and the authenticity of invoices. All non-compliance items are linked to corresponding rules and bases in the knowledge graph, ensuring that the verification results are traceable and explainable, providing a clear basis for subsequent approval decisions.

[0060] The compliance risk score can be achieved using stacking, integrating multiple models. The first layer of base learners consists of: Logistic Regression (linear baseline); XGBoost (capturing non-linearity and interaction); LightGBM (efficiently handling large-scale data); CatBoost (automatically handling categorical features); and Deep Neural Network (DNN) (learning complex patterns). Features include: {purchase amount, supplier qualification completeness, historical compliance records, price reasonableness, number of risk points in contract terms}, totaling 128 dimensions. The second layer of meta-learners uses Logistic Regression to fuse the predicted probabilities of the five base learners, learning the optimal weights. Training uses 5-fold cross-validation; the base learners are trained on the training set, and the meta-learners are trained on the validation set to avoid overfitting. Calibration uses Platt Scaling to calibrate the probability output, ensuring that the predicted probability equals the true risk probability. Experiments show an AUC of 0.94, a 3.3% improvement compared to a single XGBoost (0.91).

[0061] Optionally, step S103 involves determining the risk assessment strategy corresponding to the audit process based on the risk identification data corresponding to the procurement process, and then determining the risk assessment result corresponding to the procurement data through the risk assessment strategy. Figure 4 As shown, it includes: Step S401: Based on the procurement process, determine the relationship graph between suppliers and employees, and use the associated risk identification data corresponding to the relationship graph to determine the related transaction assessment strategy corresponding to the audit process.

[0062] Based on full-scale data from the procurement process, a multi-dimensional relationship graph is constructed between suppliers, company employees, and other partners. The core data sources for this graph include company registration information, equity structure data, cross-employment records, and data on overlapping business addresses and contact information for identifying related risks. The strategy explicitly uses graph analysis to uncover hidden relationships: focusing on verifying whether suppliers and their agents have equity connections or kinship ties, whether suppliers and other partner companies share the same ultimate controlling shareholder, and whether employees hold concurrent positions at supplier companies. By quantifying the strength of these relationships (such as equity percentage, job level, and address overlap), a risk assessment standard for related-party transactions is established, resulting in a practical related-party transaction evaluation strategy to accurately identify hidden related-party transactions that circumvent compliance requirements.

[0063] Step S402: Obtain the similarity of quotation features among suppliers in the procurement process, and use the collusion risk identification data corresponding to the quotation feature similarity to determine the collusion assessment strategy corresponding to the audit process.

[0064] The system comprehensively collects full bidding / quotation data from multiple suppliers throughout the procurement process, extracting core quotation features including numerical similarity of quotation amounts, consistency of quotation composition (e.g., cost percentage of each item), and synchronization of quotation modification records. It also integrates key data for identifying bid-rigging risks: whether the IP addresses used by suppliers are in the same network segment or the same physical address, whether the submission time of bid documents is highly concentrated, whether the reserved contact information (phone number, email address) is duplicated or related, and whether there are records of unusually close historical cooperation between suppliers. The strategy clearly uses feature similarity algorithms (such as cosine similarity and edit distance) to quantify the correlation of quotations, combined with cross-validation of multi-dimensional non-quotation features, to establish a bid-rigging risk assessment model, forming a bid-rigging assessment strategy to accurately identify illegal bidding behaviors.

[0065] Step S403: Based on the reasonableness data, authenticity data, logistics trajectory data and acceptance record data of the supplier in the procurement process, determine the false procurement risk identification data corresponding to the procurement process, and use the false procurement risk identification data to determine the false procurement assessment strategy corresponding to the audit process.

[0066] By integrating four core types of data for identifying fraudulent procurement risks, a comprehensive assessment dimension is constructed: First, data on the rationality of procurement needs, including the matching degree between the procurement target and the enterprise's business scenario, the suitability of the procurement quantity / amount with the actual usage, and the rationality of the procurement timing; second, data on the authenticity of suppliers, covering the supplier's business registration status, whether the business scope matches the procurement target, the verifiability of the actual business address and contact information, and past business performance records; third, logistics trajectory data, focusing on verifying the authenticity of logistics tracking numbers, the consistency between the transportation route and the procurement delivery address, and the completeness of goods receipt records; and fourth, acceptance record data, including the standardization of acceptance vouchers, the authenticity of the acceptance personnel's signatures, and the consistency between the acceptance results and the description of the procurement target. Based on this data, multi-dimensional cross-validation rules are established to form a fraudulent procurement assessment strategy, accurately identifying behaviors of fictitious procurement to obtain funds.

[0067] Step S404: Use the corresponding product pricing data and historical product data under the procurement process to determine the pricing anomaly risk identification data of the product, and use the pricing anomaly risk identification data to determine the pricing anomaly assessment strategy corresponding to the audit process.

[0068] A pricing anomaly risk identification system is constructed by collecting three types of core data: first, commodity pricing data, including current purchase quotations, publicly available market benchmark prices for the same period, transaction prices of similar commodities in the same industry, and pricing basis explanations in supplier quotations; second, historical commodity data, covering the purchase unit price of similar commodities in the past, the correlation between purchase volume and price, and historical price fluctuation trends; and third, comparative data of similar commodities, including price differences between different suppliers for the same type of commodity and the matching degree between the commodity's intrinsic value and the quoted price. The strategy clearly defines a pricing anomaly assessment strategy by constructing a price prediction model (trained based on historical and market data), setting reasonable price deviation thresholds (such as 30% higher than the market average price or lower than the cost price), and combining this with verification of the sufficiency of pricing basis. This strategy accurately identifies illegal procurement behaviors such as high-price purchases without reasonable basis or dumping at low prices.

[0069] Step S405: Determine fraud risk identification data by using the procurement data and turnover data of suppliers under the procurement process, and use the fraud risk identification data to determine the fraud assessment strategy corresponding to the audit process.

[0070] In-depth analysis of data for identifying fraud risks in the procurement process: First, supplier-related procurement data, including supplier cooperation frequency (e.g., frequent acceptance of small-amount purchase orders), concentration of purchase amounts (e.g., a single supplier's long-term monopoly on specific product categories), and the uniqueness of procurement contract terms (e.g., clauses without substantive performance requirements); second, data on the departure of responsible personnel, focusing on tracking their departure within a short period after the completion of procurement transactions (e.g., within 3 months) and any abnormal characteristics of their procurement activities before departure (e.g., concentrated purchases from a single supplier); third, characteristic data of historical fraud cases, including operational patterns and data characteristics of investigated fraudulent activities. The strategy clearly establishes fraud risk assessment rules by matching historical fraud patterns with the correlation analysis of procurement data and departure data, forming a fraud assessment strategy to accurately identify fraudulent behaviors such as frequent small-amount purchases to circumvent approvals and monopolistic procurement by specific suppliers.

[0071] Step S406: Determine the risk assessment strategy corresponding to the audit process based on the related party transaction assessment strategy, bid rigging assessment strategy, fraudulent procurement assessment strategy, pricing anomaly assessment strategy, and fraud assessment strategy.

[0072] The five specific assessment strategies—related-party transactions, bid rigging, fictitious procurement, abnormal pricing, and fraud—developed in steps S401 to S405 are systematically integrated to construct a comprehensive risk assessment strategy system. The integration process clarifies: first, weight allocation rules, setting different strategy weights based on the degree of harm each risk poses to the enterprise (e.g., fictitious procurement and related-party transactions are more harmful); second, cross-validation logic, requiring the risk assessment of the same procurement transaction to meet the collaborative verification of multiple specific strategies (e.g., bid rigging may be accompanied by abnormal pricing); third, risk level classification standards, setting low, medium, and high risk thresholds based on the assessment results of each specific strategy; and fourth, priority execution order, first executing basic risk identification (e.g., related-party transactions and abnormal pricing), then executing complex fraud pattern matching, ensuring comprehensive, efficient, and thorough risk assessment that aligns with the actual risk scenarios of procurement transactions.

[0073] Step S407: Obtain the risk assessment results corresponding to the procurement data under the risk assessment strategies of related party transaction assessment strategy, bid rigging assessment strategy, fraudulent procurement assessment strategy, pricing anomaly assessment strategy and fraud assessment strategy respectively.

[0074] Based on the integrated, multi-dimensional risk assessment strategy, automated risk assessments are performed on procurement data category by category, outputting precise risk assessment results under each specific strategy: Related-party transaction assessment results clarify the existence of hidden relationships, the strength level of the relationships, and the core basis for such relationships; bid rigging and collusion assessment results explain the similarity score of pricing characteristics, anomalies in non-pricing characteristics, and the probability of bid rigging risk; fraudulent procurement assessment results report anomalies in various dimensions such as demand, suppliers, logistics, and acceptance, and the level of fraudulent risk; pricing anomaly assessment results present the price deviation range, the sufficiency of pricing basis, and the confidence level of anomaly risk; fraud assessment results list the matching degree of fraud patterns, characteristics of abnormal procurement / departure data, and the conclusion of fraud risk judgment. Finally, all results are summarized to generate a comprehensive risk assessment report including the overall risk level, core risk points, risk cause analysis, and confidence score, providing a clear basis for subsequent approval decisions and risk prevention and control.

[0075] In the specific implementation, SHAP (SHapley Additive exPlanations) can be used to interpret the risk score. The marginal contribution of each feature to the prediction is calculated based on game theory Shapley Values. The implementation process is as follows: TreeSHAP (optimized for tree models, complexity O(TLD^2), where T is the number of trees, L is the number of leaves, and D is the depth) is used for fast calculation. Visualization: Force Plot - shows the push / pull force of each feature in a single sample; Summary Plot - shows the global feature importance ranking; Dependence Plot - shows the relationship curve between features and predictions. Application: When a high risk is determined, an explanation report is generated: "Risk score 85 points, main reasons: supplier lacks relevant qualifications (contribution +30 points), abnormally high purchase amount (+20 points), 3 risk points in contract terms (+15 points)". Manual review: Reviewers can view the SHAP explanation to quickly locate problems, improving review efficiency by 50%.

[0076] Optionally, step S104, which determines the approval strategy corresponding to the audit process based on compliance verification results and risk assessment results, such as... Figure 5 As shown, it includes: Step S501: Use the compliance verification results to determine the compliance status assessment results corresponding to the procurement process, and use the risk assessment results to determine the risk level assessment results corresponding to the procurement process.

[0077] Based on comprehensive compliance verification results, the compliance status assessment of the procurement process is clarified from multiple dimensions, including compliance of permissions, suppliers, prices, and contracts, processes, and invoices. Specifically, it is divided into three categories: fully compliant (all dimensions pass verification, no non-compliance markers), partially non-compliant (minor violations in some non-core dimensions, not affecting overall business security), and severely non-compliant (violations in core dimensions, such as cooperation with blacklisted suppliers, unqualified procurement, and invoice forgery). Simultaneously, the assessment results of five risk categories—related-party transactions, bid rigging, and fraudulent procurement—output by the AI ​​risk model can be combined. Based on risk confidence, scope of impact, and characteristic contribution, three risk levels are defined: low risk (risk score below the threshold, no hidden risk points), medium risk (a single non-fatal risk exists, risk is controllable), and high risk (multiple or fatal risks exist, potentially causing significant losses). During the assessment process, compliance status and risk level need to be cross-matched (e.g., "fully compliant but medium risk," "partially non-compliant and high risk") to form a more accurate dual assessment conclusion, providing a core basis for subsequent approval strategy formulation.

[0078] Step S502: Generate decision support information corresponding to the procurement process based on the compliance status assessment results and risk level assessment results.

[0079] Based on a dual assessment of compliance status and risk level, decision support information is generated to suit different scenarios, ensuring that approvers can quickly grasp the key decision-making basis. The decision support information is centered on "compliance-risk," specifically integrating multi-dimensional data: for "fully compliant + low-risk" scenarios, it can focus on providing basic information such as process compliance confirmation and supplier credibility certification; for "partially non-compliant + medium-risk" scenarios, it can detail non-compliant items and their basis, risk point characteristics, and supplier historical compliance records; for "severely non-compliant + high-risk" scenarios, it can focus on in-depth information such as risk cause analysis (combined with SHAP value feature contribution explanation), prediction of the impact of violations, and the handling results of similar violation cases. The information content can cover core modules such as historical procurement records, supplier evaluations, market price comparisons, and risk analysis reports, ensuring that approvers can quickly grasp the overall business picture without additional data retrieval.

[0080] Step S503: Based on decision support information, determine the historical statistical data, supplier profile data, price comparison data, and risk analysis data corresponding to the procurement process.

[0081] Based on the generated decision support information, four types of key decision data are further broken down and extracted to provide quantitative and concrete support for the formulation of approval strategies, as follows: Historical statistics include compliance rate, approval cycle, success rate of exception handling, and historical risk occurrence rate for similar procurement business of enterprises, such as "automatic approval rate of low-risk procurement of the same category in the past year is 95%"; Supplier profile data: Integrates supplier full lifecycle monitoring data, covering the validity of access qualifications, business / judicial / public opinion risk records, historical cooperation performance, compliance evaluation scores, blacklist association status, etc., to comprehensively reflect the reliability of suppliers; Price comparison data: includes the current purchase price and the market benchmark price, historical similar purchase prices, deviation of quotations from the same industry, completeness of inquiry and comparison records, explanation of the causes of abnormal premiums / low prices, and direct access to price compliance verification results; Risk analysis data: clearly define specific risk types (such as related-party transactions, bid rigging), risk characteristics (such as equity linkage, excessive bid similarity), risk confidence scores, scope of impact (such as estimated financial losses, compliance penalty risks), and the contribution of core risk characteristics derived from SHAP value analysis (such as "unqualified suppliers contribute 30% of the risk score").

[0082] Step S504: Determine the approval strategy corresponding to the audit process based on historical statistical data, supplier profile data, price comparison data, and risk analysis data.

[0083] Based on four types of core decision-making data and combined with the dual assessment conclusions of compliance status and risk level, targeted approval strategies are formulated, forming a three-tiered approval system of automatic approval, hierarchical routing, and mandatory interception, as detailed below: Automatic approval / fast approval strategy: When the procurement business is "fully compliant + low risk", and historical statistics show that the compliance rate of similar businesses is high, the supplier profile data has no bad records, and the price comparison is not abnormal, the automatic approval mechanism is triggered, or only one-level fast approval is required (such as confirmation by the procurement specialist), which greatly reduces the approval cycle. Tiered routing approval strategy: If the supplier is "partially non-compliant + medium risk" or "fully compliant + medium risk", the supplier will be routed precisely based on decision data: suppliers with minor negative records will be routed to the purchasing manager, suppliers with abnormal prices but controllable risks will be routed to the finance department, and suppliers with contract terms that need to be confirmed will be routed to the legal department. Risk warnings and decision data will be pushed at the same time to help approvers make quick judgments. Forced interception and special handling strategy: When a procurement business is "seriously non-compliant" or "high-risk" (such as cooperating with blacklisted suppliers, clear characteristics of fraudulent procurement, or abnormal price fluctuations exceeding the threshold), the approval process will be forcibly intercepted, and the case will be pushed to the compliance department / audit department for special investigation, along with complete risk analysis data and evidence chain. The approval process can only be restarted after the non-compliance issues are rectified or compliance is confirmed.

[0084] The formulation of approval strategies also takes into account the optimization experience accumulated through continuous learning mechanisms to ensure that strategies are dynamically adapted to business scenarios and risk changes, balancing approval efficiency with the accuracy of risk control.

[0085] Optionally, step S105 involves determining the decision support data corresponding to the procurement data based on the approval strategy, and then using the decision support data to determine the decision approval data corresponding to the procurement data. Figure 6 As shown, it includes: Step S601: Determine the decision support data corresponding to the procurement data by using historical statistical data, supplier profile data, price comparison data, and risk analysis data corresponding to the approval strategy.

[0086] Based on the established approval strategy, decision support data corresponding to procurement data is extracted and integrated from multi-dimensional core data to ensure that the data fully aligns with approval decision-making needs. Specifically, this includes: Historical statistics: Integrate quantitative data such as compliance rate, approval cycle, success rate of exception handling, and historical risk occurrence rate of similar procurement business of enterprises. At the same time, it can be linked to the optimization experience accumulated by the continuous learning mechanism (such as "the automatic pass rate of low-risk procurement of the same category in the past year is 95%, with no subsequent violation records"), to provide a reference for predicting approval results. Supplier profile data: Covers the entire lifecycle monitoring data of suppliers, including the validity of access qualifications, business registration change records, legal risks (litigation, enforcement), negative public opinion information, financial and operating status, historical cooperation performance evaluation, compliance score and blacklist association status, etc., comprehensively reflecting the reliability of suppliers; Price comparison data: Detailed presentation of the deviation range and causes between the current purchase price and the market benchmark price, historical similar purchase prices, and industry quotations. It includes complete inquiry and comparison records, verification basis for abnormal premiums / low prices, and integrates the threshold standards in price compliance verification to ensure that the reasonableness of prices is traceable. Risk analysis data: Identify the risk type (related party transactions, bid rigging, etc.), risk confidence score, core risk characteristics, and the contribution of features derived from SHAP value analysis (e.g., "the supplier's lack of relevant qualifications contributes 30% to the risk score"). Combine this with the handling results of similar risk cases to form a complete analysis of the causes and scope of impact of the risk.

[0087] By structurally integrating the four types of data, the decision support data, based on the principles of "intuitive, accurate, and practical," provides approvers with a one-stop decision-making basis without the need to access scattered data.

[0088] Step S602: Use decision support data to determine the approval results and decision results corresponding to the approval strategy, and save the approval results, decision results, historical statistical data, supplier profile data, price comparison data and risk analysis data to the preset database.

[0089] By utilizing integrated decision support data and combining it with the rule logic of the approval strategy to determine the approval and decision results, all relevant data is persistently stored in a pre-set database to support audit traceability and continuous learning. Specifically: Result determination: The approval results are clearly categorized into three types: "automatic approval", "tiered routing approval", and "mandatory interception". Low-risk and fully compliant procurement transactions are determined as "automatic approval", medium-risk or partially non-compliant transactions are determined as "tiered routing approval" (clearly routed to the corresponding nodes such as procurement manager / finance / legal department), and high-risk or seriously non-compliant transactions are determined as "mandatory interception". The decision results include specific approval opinions, risk rectification requirements (such as "supplementing supplier qualification documents" and "re-verifying price basis"), and instructions to initiate interception investigations, ensuring that the results are executable and traceable.

[0090] Data storage: Approval results and decision-making results, along with historical statistical data, supplier profile data, price comparison data, and risk analysis data, will be completely saved to the database. Stored data must meet the requirements of "compliance audit traceability and evidence chain management," including original procurement application documents, approval workflow records, compliance verification details, risk assessment reports, and other full-process trace data. Simultaneously, it will provide training data for continuous learning mechanisms (such as false alarm / missed alarm case data and approval feedback data), achieving a closed loop of data storage, reuse, and optimization.

[0091] Step S603: Determine the decision approval data corresponding to the procurement data based on the approval results and decision results.

[0092] Based on the established approval and decision-making results, corresponding decision-making and approval data for procurement are generated, clarifying the subsequent execution path and core requirements of the procurement process, and ensuring the data's operability and compliance. Specifically, this includes: If the approval result is "automatic approval", the decision approval data is the approval certificate, with a compliance verification pass detail and low risk confirmation statement, supporting the procurement business to directly enter the order execution stage; If the approval result is "tiered routing approval", the decision approval data clearly defines the allocation of approval nodes (such as "procurement supervisor reviews the reasonableness of prices + finance reviews the compliance of invoices"), access permissions for decision support data, and approval time limits. It also includes risk warnings and historical reference cases to help approvers make quick decisions. If the approval result is "forced interception", the decision approval data will be a process suspension notice and a special investigation instruction, which will be clearly pushed to the compliance department / audit department, along with complete risk analysis data, evidence chain index and key investigation directions (such as "verify whether there is an equity relationship between the supplier and the person in charge"). For procurement operations that require rectification, the decision-making and approval data also includes a list of rectification requirements, review nodes and standards after rectification, ensuring that the rectified data can be re-entered into the approval process, which aligns with the compliance management logic of "risk interception-rectification-review".

[0093] The resulting decision-making and approval data comprehensively covers the entire process of "approval-routing-interception-rectification," providing clear data support and execution basis for the compliant advancement of procurement operations.

[0094] Optionally, before step S101, which involves constructing a knowledge graph of the procurement process based on entity parameters, rule parameters, relationship parameters, and risk parameters corresponding to the procurement data, such as... Figure 7 As shown, the method also includes: Step S701: Obtain the document files corresponding to the procurement process and the review process, and determine the parsing strategy for the document files.

[0095] First, it is necessary to comprehensively collect all types of documents involved in the entire procurement and review process, covering purchase requisitions, supplier qualification documents (business licenses, industry qualification certificates, etc.), contracts, quotations, purchase orders, invoices (electronic invoices, paper scans), acceptance forms, payment vouchers, etc., encompassing structured, semi-structured, and unstructured formats (PDF, images, scans, Word documents, etc.). Combining multimodal document deep semantic parsing technology, targeted parsing strategies are determined based on document format characteristics and content type: for PDF and image documents, the LayoutLM multimodal model can be used, integrating text, layout (coordinate information), and visual (image features) three-dimensional parsing capabilities; for scanned documents, text and coordinate information can be extracted first using Tesseract / PaddleOCR tools, and then visual features can be extracted using the ResNet model; for structured documents (such as Excel format quotations), table recognition technology is used to quickly locate data areas; simultaneously, dedicated parsing templates are preset for different document types (contracts, invoices, qualification documents) to ensure that the parsing strategy adapts to document characteristics and improves the accuracy of information extraction.

[0096] Step S702: Extract the target information contained in the document file using the parsing strategy; wherein the target information includes at least: procurement entity information, procurement target information, time information, contract terms information, and invoice information.

[0097] Based on a defined parsing strategy, target information is systematically extracted from various documents using OCR+NLP fusion technology: Procurement entity information: The BERT-NER named entity recognition technology is used to extract the applicant, approver, supplier name, consignee, and handling department, and simultaneously identify related data such as supplier business registration information, tax number, and contact information; Procurement target information: Accurately extract the name, specifications, quantity, unit price, total price, and category of goods / services, and establish the relationship between the target and the procurement entity and suppliers using relationship extraction technology; Time information: Extract key time nodes such as purchase request time, contract signing date, delivery deadline, payment period, invoice issuance time, and acceptance time, and ensure the legality of date formats through rule validation; Contract terms information: Analyze core terms such as payment methods, warranty period, liability for breach of contract, special agreements, and qualification requirements, and use text classification technology to identify unfavorable or legally risky clauses; Invoice information: Extract invoice number, tax number, amount, tax amount, tax rate, invoice header, invoice authenticity verification mark, etc., and combine them with the three-invoice matching logic to associate order and warehouse entry information; Expand target information: Extract additional information such as supplier qualification validity, inquiry and price comparison records, logistics tracking clues, and acceptance record details to support the improvement of subsequent procurement data and risk identification.

[0098] After extraction, post-processing is performed through knowledge graph verification (such as matching supplier names with the access list) and format rule verification (such as the legality of amount values) to filter out erroneous or redundant information.

[0099] Step S703: Determine procurement data based on target information.

[0100] Based on the extracted and verified target information, multi-dimensional structured integration is performed to form complete and standardized procurement data, providing core data support for subsequent knowledge graph construction (step S101): By linking and integrating scattered target information such as procurement entities, targets, and time, a logical link chain of "procurement entity - procurement target - supplier - contract - invoice" is established, which aligns with the entity relationship requirements of knowledge graphs. Mapping target information into the basic data dimensions required for knowledge graph construction: original data of entity parameters corresponding to information such as procurement entity / supplier / target, basic data of rule parameters corresponding to compliance requirements and approval process specifications in contract terms, original data of relationship parameters corresponding to the association information between entity and target, supplier and blacklist, and basic data of risk parameters corresponding to information such as abnormal invoice characteristics and supplier qualification defects. The integrated data is standardized to unify the data format (such as monetary units, date formats, and coding rules), and missing key fields are supplemented (such as supplementing supplier business information through external data sources) to ensure the integrity, consistency, and availability of the procurement data, laying the foundation for the accurate determination of subsequent entity parameters, rule parameters, etc.

[0101] Deep semantic parsing employs LayoutLM to process PDF / image documents. Model architecture: LayoutLM, combining text, layout (coordinates), and visual (image) modalities. Input: Document image; text and coordinates are extracted via OCR (Tesseract / PaddleOCR), and visual features are extracted via ResNet. Encoding: Text embedding + position embedding (2D coordinate normalization) + visual embedding are concatenated and input to a Transformer. Tasks: Document classification (contracts / invoices / certificates); key information extraction (contract amount, signing date, supplier name); table recognition (extracting table structure and content). Training: Using public datasets such as FUNSD and CORD, plus 5000 self-built procurement documents, achieving an F1 score of 92.8%. Post-processing: Rule-based validation (e.g., amount format, date validity) and knowledge graph validation (e.g., supplier name in the database).

[0102] Optionally, after step S105, which determines the decision support data corresponding to the procurement data based on the approval strategy and uses the decision support data to determine the decision approval data corresponding to the procurement data, as follows: Figure 8 As shown, the method also includes: Step S801: Obtain real-time information change data, litigation data, public opinion data, and financial monitoring data corresponding to suppliers.

[0103] By establishing a multi-source data real-time connection mechanism, we continuously collect dynamic data from suppliers across all dimensions, ensuring that the data covers four core dimensions: supplier operational status, legal compliance, social reputation, and financial health, as detailed below: Information Change Data: Connect with relevant enterprise credit information disclosure systems and other business data sources to obtain supplier business change records in real time, including key information such as changes in legal representatives, adjustments to equity structures, changes in business scope, changes in registered capital, relocation of business addresses, and cancellation / revocation status; Litigation data: By linking with relevant judicial disclosure platforms (such as the platform for querying information on persons subject to enforcement), we can capture in real time judicial risk data such as supplier-related litigation cases, enforcement records, information on the identification of dishonest persons subject to enforcement, and administrative penalties (such as environmental penalties and tax penalties); Public opinion data: Through web crawlers and public opinion monitoring tools, we cover mainstream news media, social platforms, industry forums and other channels to capture negative public opinion related to suppliers in real time, including public opinion information such as product quality accidents, service complaints, exposure of violations, and social responsibility disputes. Financial monitoring data: Connects with tax systems, corporate credit rating agencies, and third-party financial data platforms to obtain real-time financial dynamics of suppliers, including records of inclusion in the list of abnormal operations, abnormal tax declarations, fluctuations in key financial statement indicators (such as a significant decline in revenue or excessive debt ratio), and financial risk data such as financing defaults.

[0104] Data cleaning and format standardization are carried out simultaneously during data collection to ensure data accuracy and usability, providing high-quality input for risk assessment.

[0105] Step S802: Use information change data, litigation data, public opinion data, and financial monitoring data to determine the corresponding risk level of the supplier.

[0106] This step involves in-depth analysis and risk assessment of the collected multi-dimensional dynamic data to accurately determine the latest risk level of the supplier. First, feature extraction is performed on various types of data. For example, equity change data is associated with the feature of "related party transaction risk", records of dishonest persons subject to enforcement are associated with the feature of "serious compliance risk", the breadth of negative public opinion is associated with the feature of "reputation risk diffusion", and tax abnormalities are associated with the feature of "financial performance capability risk". It can call upon pre-trained ensemble learning risk models (such as Stacking multi-model fusion architecture), input the extracted features into the model for quantitative scoring, and combine the SHAP value to analyze the contribution of each feature to the risk score (such as "the record of dishonest person subject to enforcement contributes +40 points" and "the continuous decline in revenue contributes +25 points"). Based on the preset risk threshold (in line with the "low / medium / high" three-level risk classification standard), the latest risk level of the supplier is determined by the comprehensive quantitative scoring results. At the same time, the risk level is compared with the historical risk level to clarify the "upgrade", "downgrade" or "maintain" status of the risk level, generate a risk level change report, and mark the core risk triggering factors.

[0107] Step S803: Update the approval strategy based on the risk level results.

[0108] Based on the latest risk level results of suppliers, the original approval strategy will be adjusted and iterated in a targeted manner to ensure that the approval strategy is dynamically adapted to the real-time risk status of suppliers: If a supplier's risk level is upgraded (e.g., from medium risk to high risk): the risk layer data of that supplier in the knowledge graph will be updated synchronously (e.g., marked with a "high-risk judicial" label), and the approval strategy will be adjusted to "mandatory interception + special review"; if it involves an ongoing procurement contract, an additional compliance review process may be triggered. If a supplier’s risk level is downgraded (e.g., from high risk to medium risk): the approval strategy is optimized to “tiered routing approval + key monitoring”, simplifying some non-core verification steps, retaining key compliance and risk verification items, and shortening the approval cycle, such as routing to joint approval by the procurement manager and the finance department, without the need for prior intervention by the compliance department. If the supplier's risk level remains unchanged but there are changes in characteristics (such as low risk but slight negative public opinion): update the risk warning content in the approval strategy, add an explanation of the change data to the approval process to assist the approver in decision-making; After the strategy is updated, the compliance verification rule base and knowledge graph are linked in sync to ensure that the new approval strategy is automatically synchronized to the compliance review process of subsequent procurement business, forming a closed loop of "data collection-risk assessment-strategy optimization" to continuously improve the accuracy of the approval strategy and risk control capabilities.

[0109] As can be seen from the above procurement review process control method, this method determines the compliance verification strategy of the review process by constructing a knowledge graph of the procurement process, and then uses the structured rule processing of the knowledge graph to automatically execute and reason about the procurement process. In addition, this method uses the compliance verification results and risk assessment results of the procurement data to determine the approval strategy of the review process, improves the accuracy of the review process, can identify complex risk points in the procurement review process, and can conduct full-cycle collaborative monitoring of the supplier's compliance status, thereby achieving full-process coverage of the procurement review process.

[0110] Corresponding to the above embodiments of the procurement review process control method, this invention also provides a procurement review process control system, such as... Figure 9 As shown, the system includes: The knowledge graph construction module 910 is used to construct a knowledge graph corresponding to the procurement process based on entity parameters, rule parameters, relationship parameters and risk parameters corresponding to the procurement data. The compliance verification result determination module 920 is used to determine the compliance verification strategy corresponding to the audit process using a knowledge graph, and to determine the compliance verification result corresponding to the procurement data through the compliance verification strategy. The risk assessment result determination module 930 is used to determine the risk assessment strategy corresponding to the audit process based on the risk identification data corresponding to the procurement process, and to determine the risk assessment result corresponding to the procurement data through the risk assessment strategy. The approval strategy determination module 940 is used to determine the approval strategy corresponding to the review process based on the compliance verification results and risk assessment results. The decision review and control module 950 is used to determine the decision support data corresponding to the procurement data based on the approval strategy, and to determine the decision approval data corresponding to the procurement data using the decision support data.

[0111] As can be seen from the above procurement review process control system, the system determines the compliance verification strategy of the review process by constructing a knowledge graph of the procurement process. It then uses the structured rule processing of the knowledge graph to automatically execute and reason about the procurement process. In addition, the system uses the compliance verification results and risk assessment results of the procurement data to determine the approval strategy of the review process, thereby improving the accuracy of the review process. It can identify complex risk points in the procurement review process and can conduct full-cycle collaborative monitoring of the supplier's compliance status, thus achieving full coverage of the procurement review process.

[0112] The procurement review process control system provided in this embodiment of the invention has the same implementation principle and technical effects as the aforementioned procurement review process control method embodiment. For the sake of brevity, any parts not mentioned in the system embodiment can be referred to the corresponding content in the aforementioned procurement review process control method embodiment.

[0113] This embodiment also provides an electronic device, the structural schematic diagram of which is shown below. Figure 10 As shown, the device includes a processor 101 and a memory 102; wherein, the memory 102 is used to store one or more computer instructions, which are executed by the processor to implement the steps of the above-mentioned procurement review process control method.

[0114] Figure 10 The electronic device shown also includes a bus 103 and a communication interface 104, with the processor 101, communication interface 104 and memory 102 connected via the bus 103.

[0115] The memory 102 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. The bus 103 may be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 10 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0116] The communication interface 104 is used to connect to at least one user terminal and other network units through a network interface, and to send encapsulated IPv4 packets or IPv4 packets to the user terminal through the network interface.

[0117] Processor 101 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 101 or by instructions in software form. The processor 101 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this disclosure. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this disclosure can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 102, and processor 101 reads the information in memory 102 and, in conjunction with its hardware, completes the steps of the method described in the foregoing embodiments.

[0118] This invention also provides a storage medium storing a computer program, which, when executed by a processor, performs the steps of the procurement review process control method described in the foregoing embodiments.

[0119] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, devices, and methods can be implemented in other ways. The system embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0120] 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.

[0121] 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.

[0122] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion 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, electronic device, or network device, etc.) to execute all or part of the steps of the methods described in 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.

[0123] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the claims.

Claims

1. A method for controlling the procurement review process, characterized in that, The method includes: A knowledge graph corresponding to the procurement process is constructed based on entity parameters, rule parameters, relationship parameters, and risk parameters corresponding to the procurement data. The knowledge graph is used to determine the compliance verification strategy corresponding to the audit process, and the compliance verification result corresponding to the procurement data is determined through the compliance verification strategy. Based on the risk identification data corresponding to the procurement process, determine the risk assessment strategy corresponding to the audit process, and determine the risk assessment result corresponding to the procurement data through the risk assessment strategy; The approval strategy corresponding to the review process is determined based on the compliance verification results and the risk assessment results. Based on the approval strategy, the decision support data corresponding to the procurement data is determined, and the decision support data is used to determine the decision approval data corresponding to the procurement data.

2. The procurement review process control method according to claim 1, characterized in that, The steps for constructing a knowledge graph of the procurement process based on entity parameters, rule parameters, relationship parameters, and risk parameters corresponding to procurement data include: Entity parameters are determined based on the supplier data, product data, contract data, order data, invoice data, employee data, and department data contained in the procurement data; The rule parameters are determined based on the procurement authority data, supplier access data, price compliance data, contract terms data, invoice compliance data, and payment approval data contained in the procurement data. The relationship parameters are determined based on the supplier association data, market price association data, product category association data, employee permission association data, and blacklist association data contained in the procurement data; Risk parameters are determined based on the related-party transaction risk data, bid rigging risk data, fraudulent procurement risk data, and abnormal pricing risk data contained in the procurement data; A knowledge graph corresponding to the procurement process is constructed using the entity attribute parameters corresponding to the entity parameters, the rule logic parameters corresponding to the rule parameters, the relation semantic parameters corresponding to the relation parameters, and the risk extraction parameters corresponding to the risk parameters.

3. The procurement review process control method according to claim 1, characterized in that, The steps of determining the compliance verification strategy corresponding to the audit process using the knowledge graph, and determining the compliance verification result corresponding to the procurement data using the compliance verification strategy, include: Based on the applicant's procurement authority, approval process authority, and monetary authority determined by the knowledge graph, the corresponding authority verification strategy for the review process is determined. Based on the supplier access data, supplier blacklist data, supplier qualification data, and related conflict data determined by the knowledge graph, the supplier verification strategy corresponding to the review process is determined; Based on the market price comparison results, historical price comparison results, and price comparison record results determined by the knowledge graph, the price verification strategy corresponding to the review process is determined. Based on the contract verification data, process verification data, and invoice verification data determined by the knowledge graph, the compliance verification strategy corresponding to the audit process is determined. The compliance verification strategy corresponding to the audit process is determined using the permission verification strategy, the supplier verification strategy, the price verification strategy, and the compliance verification strategy. The compliance verification strategies are used to obtain the compliance verification results corresponding to the procurement data under the permission verification strategy, the supplier verification strategy, the price verification strategy, and the compliance verification strategy, respectively.

4. The procurement review process control method according to claim 1, characterized in that, The steps of determining the risk assessment strategy corresponding to the audit process based on the risk identification data corresponding to the procurement process, and determining the risk assessment result corresponding to the procurement data through the risk assessment strategy, include: Based on the procurement process, a relationship graph between suppliers and employees is determined, and the related risk identification data corresponding to the relationship graph is used to determine the related transaction assessment strategy corresponding to the audit process. Obtain the similarity of quotation features among the suppliers under the procurement process, and use the collusion risk identification data corresponding to the quotation feature similarity to determine the collusion and bid-rigging evaluation strategy corresponding to the review process; Based on the reasonableness data, authenticity data, logistics trajectory data and acceptance record data of the supplier under the procurement process, the false procurement risk identification data corresponding to the procurement process is determined, and the false procurement risk identification data is used to determine the false procurement assessment strategy corresponding to the audit process. Using the product pricing data and historical product data corresponding to the procurement process, determine the pricing anomaly risk identification data for the product, and use the pricing anomaly risk identification data to determine the pricing anomaly assessment strategy corresponding to the review process; Fraud risk identification data is determined by using the procurement data and resignation data of the suppliers under the procurement process, and fraud assessment strategies are determined by using the fraud risk identification data for the audit process. Based on the related-party transaction assessment strategy, the bid-rigging assessment strategy, the fraudulent procurement assessment strategy, the pricing anomaly assessment strategy, and the fraud assessment strategy, the risk assessment strategy corresponding to the review process is determined; The risk assessment strategies are used to obtain the risk assessment results corresponding to the procurement data under the related party transaction assessment strategy, the bid rigging assessment strategy, the fraudulent procurement assessment strategy, the pricing anomaly assessment strategy, and the fraud assessment strategy, respectively.

5. The procurement review process control method according to claim 1, characterized in that, The steps for determining the approval strategy corresponding to the review process based on the compliance verification results and the risk assessment results include: The compliance status assessment result corresponding to the procurement process is determined using the compliance verification result, and the risk level assessment result corresponding to the procurement process is determined using the risk assessment result. The procurement process is generated based on the compliance status assessment results and the risk level assessment results. Based on the decision support information, historical statistical data, supplier profile data, price comparison data, and risk analysis data corresponding to the procurement process are determined. The approval strategy corresponding to the review process is determined based on the historical statistical data, the supplier profile data, the price comparison data, and the risk analysis data.

6. The procurement review process control method according to claim 5, characterized in that, The steps of determining the decision support data corresponding to the procurement data based on the approval strategy, and determining the decision approval data corresponding to the procurement data using the decision support data, include: The decision support data corresponding to the procurement data is determined by using the historical statistical data, supplier profile data, price comparison data, and risk analysis data corresponding to the approval strategy. The decision support data is used to determine the approval results and decision results corresponding to the approval strategy, and the approval results, decision results, historical statistical data, supplier profile data, price comparison data and risk analysis data are saved to a preset database; The decision approval data corresponding to the procurement data is determined based on the approval results and the decision results.

7. The procurement review process control method according to claim 1, characterized in that, Before the step of constructing a knowledge graph corresponding to the procurement process based on entity parameters, rule parameters, relationship parameters, and risk parameters corresponding to the procurement data, the method further includes: Obtain the document files corresponding to the procurement process and the review process, and determine the parsing strategy corresponding to the document files; The target information contained in the document file is extracted using the parsing strategy described above; wherein the target information includes at least: information on the procuring entity, information on the procuring object, time information, contract terms information, and invoice information. The procurement data is determined based on the target information.

8. The procurement review process control method according to claim 1, characterized in that, After determining the decision support data corresponding to the procurement data based on the approval strategy, and determining the decision approval data corresponding to the procurement data using the decision support data, the method further includes: Real-time acquisition of supplier-related information change data, litigation data, public opinion data, and financial monitoring data; The risk level of the supplier is determined by using the information change data, the litigation data, the public opinion data, and the financial monitoring data. The approval strategy is updated based on the risk level results.

9. A procurement review process control system, characterized in that, The system includes: The knowledge graph construction module is used to build a knowledge graph corresponding to the procurement process based on entity parameters, rule parameters, relationship parameters and risk parameters corresponding to procurement data. The compliance verification result determination module is used to determine the compliance verification strategy corresponding to the audit process using the knowledge graph, and to determine the compliance verification result corresponding to the procurement data through the compliance verification strategy. The risk assessment result determination module is used to determine the risk assessment strategy corresponding to the audit process based on the risk identification data corresponding to the procurement process, and to determine the risk assessment result corresponding to the procurement data through the risk assessment strategy. The approval strategy determination module is used to determine the approval strategy corresponding to the review process based on the compliance verification results and the risk assessment results. The decision review and control module is used to determine the decision support data corresponding to the procurement data based on the approval strategy, and to determine the decision approval data corresponding to the procurement data using the decision support data.

10. An electronic device, characterized in that, The electronic device includes a processor and a memory, the memory storing computer-executable instructions that can be executed by the processor, the processor executing the computer-executable instructions to implement the steps of the procurement review process control method according to any one of claims 1 to 8.

Images