Contract performance process risk management method and system combined with multi-source data integration analysis

Abstract

The application discloses a method and system for risk management of performance process combined with multi-source data integration analysis, and mainly relates to the technical field of data processing. The method comprises the following steps: obtaining contract data and supply chain data, performing data cleaning and standardization processing, obtaining a performance plan information set and a performance execution event set; obtaining a first heavy perception risk identification result and a second heavy perception risk identification result; performing implicit alignment and feature weighting on the first heavy perception risk identification result and the second heavy perception risk identification result to construct a double risk perception result; determining a performance process risk management strategy based on the double risk perception result, and performing risk management according to the performance process risk management strategy. The application has the beneficial effect of solving the technical problem of low risk management reliability and high missed detection rate caused by the lack of cross-stage and cross-event correlation analysis of multi-source data in the prior art, and achieving the technical effect of improving the reliability of performance process risk management.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically to a method and system for risk management of performance processes that combines multi-source data integration and analysis. Background Technology

[0002] In supply chain fulfillment management, companies typically rely on multiple business systems, such as contract management systems, logistics systems, and shipping systems, to record contract requirements and the actual fulfillment process. However, these data sources often suffer from inconsistencies in format, different recording granularities, and significant differences in business semantics, making it difficult to form a coherent chain of information during the fulfillment process and resulting in the difficulty in timely detection of fulfillment deviations. Furthermore, traditional fulfillment risk identification often relies on single-dimensional data comparisons, such as judging solely by the pure time difference between the contract delivery date and the actual arrival date, which fails to identify hidden risks from the complex and ever-changing supply chain data.

[0003] Existing technologies generally lack the ability to jointly analyze multi-source data. For discrepancies between shipping records and contract fulfillment, traditional methods typically employ rule-based comparison, which cannot further uncover implicit semantic relationships between events such as batch splitting, route changes, and abnormal remarks. Furthermore, they cannot effectively detect weak anomalies across stages and events, leading to untimely risk warnings and a high rate of missed detections. Summary of the Invention

[0004] This application provides a risk management method and system for the performance process that combines multi-source data integration and analysis. It is used to address the technical problem in the prior art that the lack of cross-stage and cross-event correlation analysis of multi-source data leads to low reliability and high false negative rate in risk management.

[0005] In view of the above problems, this application provides a method and system for risk management of the performance process that combines multi-source data integration and analysis.

[0006] The first aspect of this application provides a method for risk management of performance processes that combines multi-source data integration and analysis, the method comprising: Acquire contract and supply chain data, perform data cleaning and standardization, and obtain a set of performance plan information and a set of performance execution events; Risk perception is performed on the set of performance plan information and the set of performance execution events according to the dual inconsistency perception risk identification mechanism to obtain the first perception risk identification result and the second perception risk identification result. Implicit alignment and feature weighting are performed on the first-level risk identification result and the second-level risk identification result to construct a dual-risk identification result; Based on the results of the dual risk perception, a risk management strategy for the performance process is determined, and risk management is carried out according to the risk management strategy for the performance process.

[0007] In one possible embodiment, risk perception is performed on the performance plan information set and the performance execution event set according to a dual inconsistency perception risk identification mechanism to obtain a first-level perception risk identification result and a second-level perception risk identification result, including: According to the preset explicit comparison items, the first-level perception risk identification is performed on the performance plan information set and the performance execution event set to obtain the first-level perception risk identification result; The set of performance events is subjected to sequential semantic association memory iteration to obtain performance execution memory; Based on the performance execution memory, a second layer of perceived risk identification is performed to obtain the second layer of perceived risk identification result.

[0008] In one possible embodiment, according to preset explicit comparison items, a first-level risk identification is performed on the performance plan information set and the performance execution event set to obtain the first-level risk identification result, including: The set of performance events is subjected to structured execution information extraction to obtain a structured data set; According to the preset explicit comparison items, the performance plan information set is compared with the structured data set item by item to obtain multiple explicit deviation results and multiple explicit deviation degrees. The multiple explicit deviation results and multiple explicit deviation degrees are used as the first-level perceived risk identification results; The preset explicit comparison items include delivery time comparison items, quantity consistency comparison items, batch structure comparison items, event sequence comparison items, logistics node integrity comparison items, material consistency comparison items, and delivery location comparison items.

[0009] In one possible embodiment, the set of performance execution events is subjected to sequential semantic association memory iteration to obtain performance execution memory, including: The set of performance execution events is sorted in chronological order to obtain the performance execution event sequence; Traverse the sequence of performance execution events, and take each performance execution event as the center to perform forward correlation feature identification to obtain a forward correlation feature sequence; Traverse the sequence of performance execution events, and take each performance execution event as the center to perform backward correlation feature identification to obtain the backward correlation feature sequence; The forward-associated feature sequence and the backward-associated feature sequence are mapped and aligned to obtain a forward-backward mapped associated feature sequence; Based on the forward-backward mapping associated feature sequence, the forward and backward semantic association memory is iterated to obtain the performance execution memory.

[0010] In one possible embodiment, the forward-backward mapping associated feature sequence is used to perform forward-backward semantic association memory iteration to obtain the performance execution memory, including: Extract the first forward-backward mapping association feature and the second forward-backward mapping association feature from the forward-backward mapping association feature sequence, perform forward-backward semantic association recognition, and add the recognition result into the first semantic association memory unit; Based on the first semantic association memory unit, the third forward-backward mapping association feature in the forward-backward mapping association feature sequence is subjected to forward-backward semantic association memory iteration to obtain the second semantic association memory unit; This process continues until the last element of the forward-backward mapping associated feature sequence is reached, at which point the performance memory is obtained.

[0011] In one possible embodiment, a second-level risk identification is performed based on the performance execution memory to obtain the second-level risk identification result, including: Multiple sample performance execution memories and multiple sample second-level perception risk identification results are used as training samples; The framework built on the feedforward neural network is trained under supervision based on the training samples until the training converges, and the trained second-level perceptual risk identifier is obtained. The second-level risk identification device is used to perform risk perception on the performance execution memory to obtain the second-level risk identification result.

[0012] In one possible embodiment, the first-level risk identification result and the second-level risk identification result are implicitly aligned and weighted to construct a dual-risk identification result, including: Implicit alignment is performed on risks of the same type in the first and second levels of risk identification results to obtain implicit alignment results. The implicit alignment results are weighted according to preset weights to obtain the dual risk perception results.

[0013] A second aspect of this application provides a performance process risk management system that integrates multi-source data analysis, the system comprising: The data processing module is used to acquire contract data and supply chain data, perform data cleaning and standardization processing, and obtain a set of performance plan information and a set of performance execution events. The risk perception module is used to perceive the risk of the performance plan information set and the performance execution event set according to the dual inconsistency perception risk identification mechanism, and obtain the first perception risk identification result and the second perception risk identification result. The feature weighting module is used to implicitly align and weight the first-level risk identification result and the second-level risk identification result to construct a dual risk identification result; The risk management module is used to determine the risk management strategy for the performance process based on the results of the dual risk perception, and to carry out risk management according to the risk management strategy for the performance process.

[0014] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application acquires contract data and supply chain data, performs data cleaning and standardization to obtain a set of performance plan information and a set of performance execution events. It then uses a dual inconsistency perception risk identification mechanism to perceive risks in these sets, obtaining first-level and second-level risk identification results. Implicit alignment and feature weighting are applied to these two results to construct a dual risk perception result. Based on this dual risk perception result, a performance process risk management strategy is determined, and risk management is implemented according to this strategy. This achieves the technical effect of dual-perception risk identification of multi-source data, improving the reliability of performance process risk management. Attached Figure Description

[0015] Appendix Figure 1 This is a schematic diagram of the performance process risk management method that combines multi-source data integration and analysis, provided in an embodiment of the present invention.

[0016] Appendix Figure 2 This is a schematic diagram of the performance process risk management system structure provided by an embodiment of the present invention, which integrates and analyzes multi-source data.

[0017] The labels shown in the attached diagram: Data processing module 11, risk perception module 12, feature weighting module 13, risk management module 14. Detailed Implementation

[0018] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims. It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to these processes, methods, products, or devices.

[0019] Example 1, as shown in the appendix Figure 1 As shown, this application provides a method for risk management of the performance process that combines multi-source data integration and analysis, wherein the method includes: Step S100: Obtain contract data and supply chain data, perform data cleaning and standardization processing, and obtain a set of performance plan information and a set of performance execution events; In one possible embodiment, contract data refers to the performance constraint information recorded in the supply chain contract, including delivery plans, planned time points, planned batches, quantity requirements, material specifications, supplier obligations, etc.; supply chain data refers to the execution data generated by business systems such as the shipping system, transportation system, and warehousing system during the actual performance process, such as outbound records, batch information, transportation node events, exception notes, route change records, and arrival confirmations. The performance plan information set refers to the structured set of processed contract constraint information. The performance execution event set is the set of execution data divided by events after the structured decomposition of supply chain execution behaviors.

[0020] The system retrieves contract-related information from the company's internal contract management system, such as planned data like 500 units of a certain material to be delivered by the 15th of this month. At the same time, it collects raw execution information related to contract fulfillment in real time from interfaces such as the shipping system, logistics tracking system, and warehousing system, such as when the goods leave the warehouse, changes in nodes during transportation, exception descriptions, and final arrival time.

[0021] Because this data comes from different systems, the formats are often inconsistent. For example, the time format might be 2024 / 05 / 01 in one system and 05-01-24 in another. Material codes might also be written as MAT001 in system A and 001-M in system B. Therefore, it is necessary to first standardize these formats, organize all information into fields that can correspond to each other, and clean up obviously erroneous or duplicate data, such as missing timestamps or the same outbound record being repeated twice.

[0022] After the cleanup is completed, the planned content in the contract is organized into a structured plan set, such as standard fields for delivery date, quantity, batch, etc. Similarly, the supply chain execution behavior is broken down into individual events, such as shipment on 2024 / 5 / 10, arrival at transit warehouse on 2024 / 5 / 11, and transportation delay on 2024 / 5 / 12, and these are placed into the fulfillment execution event set.

[0023] By unifying, clarifying, and making traceable performance-related information that was originally scattered across multiple systems and had chaotic formats, the technology achieves the effect of providing a reliable data foundation for subsequent comparison between planning and execution, as well as risk identification.

[0024] Step S200: Perform risk perception on the performance plan information set and the performance execution event set according to the dual inconsistency perception risk identification mechanism to obtain the first perception risk identification result and the second perception risk identification result; Furthermore, risk perception is performed on the performance plan information set and the performance execution event set according to the dual inconsistency perception risk identification mechanism to obtain the first perception risk identification result and the second perception risk identification result. In this embodiment, step S200 further includes: According to the preset explicit comparison items, the first-level perception risk identification is performed on the performance plan information set and the performance execution event set to obtain the first-level perception risk identification result; The set of performance events is subjected to sequential semantic association memory iteration to obtain performance execution memory; Based on the performance execution memory, a second layer of perceived risk identification is performed to obtain the second layer of perceived risk identification result.

[0025] In one embodiment, the dual inconsistency perception risk identification mechanism refers to using two different dimensions to identify risks: one is explicit deviation identification based on rules and contractual requirements, and the other is implicit risk identification based on the semantic relationships between supply chain execution events. Predefined explicit comparison items refer to some predefined hard rules, such as whether delivery time is delayed, whether quantity is consistent, and whether batch structure conforms to the contract, used to determine obvious performance deviations. Specifically, the predefined explicit comparison items include delivery time comparison items, quantity consistency comparison items, batch structure comparison items, event sequence comparison items, logistics node integrity comparison items, material consistency comparison items, and delivery location comparison items.

[0026] Using pre-defined explicit comparison items, the performance plan is compared item by item with the performance execution, such as checking for delivery delays, insufficient shipment quantities, missing logistics nodes, and abnormal batch splitting. This allows us to identify directly observable problems; these findings constitute the first layer of perceived risk identification.

[0027] Furthermore, all fulfillment events are arranged chronologically, and the semantic and behavioral relationships before and after each event are connected. For example, if the quantity of shipments decreases in a certain period and then transportation stagnates, it may indicate a replenishment risk. If the warehouse's exception notes include words like "traffic jam" or "transfer," and subsequent nodes experience delays, then these pieces of information are semantically related. By iterating over these relationships before and after events, all related features are gradually accumulated into a fulfillment memory.

[0028] Furthermore, this allows us to assess the potential risks hidden within the event chain. For example, a minor delay may not be considered a risk in itself, but the combination of a slight delay → route change → anomaly report → delivery fluctuations constitutes a typical weak-signal chain of risks. Ultimately, these identification results form the second layer of perceived risk identification.

[0029] This technology not only identifies surface problems but also analyzes deeper issues, achieving a more comprehensive, precise, and predictive technical effect in risk identification.

[0030] Furthermore, according to preset explicit comparison items, a first-level risk identification is performed on the performance plan information set and the performance execution event set to obtain the first-level risk identification result. Step S200 in this embodiment further includes: The set of performance events is subjected to structured execution information extraction to obtain a structured data set; According to the preset explicit comparison items, the performance plan information set is compared with the structured data set item by item to obtain multiple explicit deviation results and multiple explicit deviation degrees. The multiple explicit deviation results and multiple explicit deviation degrees are used as the first-level perceived risk identification results; The preset explicit comparison items include delivery time comparison items, quantity consistency comparison items, batch structure comparison items, event sequence comparison items, logistics node integrity comparison items, material consistency comparison items, and delivery location comparison items.

[0031] In one embodiment, the raw, irregular performance records in the supply chain system are transformed into calculable and comparable standard fields to obtain a structured dataset. This structured dataset is a collection of all performance records after they have been structured, facilitating subsequent comparison against contract requirements. Explicit deviation results refer to the type of deviation, such as insufficient quantity received or delayed shipment, while explicit deviation severity refers to the degree of deviation, such as a 2-day delay or a quantity difference of 30 units.

[0032] During the first level of risk identification, if the logistics system returns a message describing a traffic jam and anticipated delay, key fields are extracted: event type (transportation anomaly), time (the exact time the anomaly occurred), and location (e.g., transfer station location), related batches, and related materials. This transforms all events into structured data that can be directly compared by field. For example, free text like "shipped on 2024 / 05 / 11" can be broken down into clear fields: event type = shipment, time = 2024 / 05 / 11, quantity = 300, batch = batch 01, etc.

[0033] The system performs a step-by-step comparison based on pre-defined explicit comparison criteria. For example, it checks whether the delivery time exceeds the contractually required delivery time, whether the quantity shipped matches the contract quantity, whether goods that should have been delivered in one batch were split into multiple batches, and whether any nodes in the transportation chain are missing (e.g., a transit warehouse → destination warehouse route where no transit warehouse record is found). Additionally, it checks whether the material code matches the contract and whether the actual delivery location is as specified in the contract.

[0034] Then, multiple explicit bias results and multiple explicit bias degrees are output as the first-level perceived risk identification results, thus providing basic explicit bias identification.

[0035] Furthermore, the semantic association memory of the performance execution event set is iterated before and after to obtain the performance execution memory. Step S200 in this embodiment of the application also includes: The set of performance execution events is sorted in chronological order to obtain the performance execution event sequence; Traverse the sequence of performance execution events, and take each performance execution event as the center to perform forward correlation feature identification to obtain a forward correlation feature sequence; Traverse the sequence of performance execution events, and take each performance execution event as the center to perform backward correlation feature identification to obtain the backward correlation feature sequence; The forward-associated feature sequence and the backward-associated feature sequence are mapped and aligned to obtain a forward-backward mapped associated feature sequence; Based on the forward-backward mapping associated feature sequence, the forward and backward semantic association memory is iterated to obtain the performance execution memory.

[0036] Furthermore, based on the forward-backward mapping associated feature sequence, iterative semantic association memory is performed to obtain the performance execution memory. In this embodiment, step S200 further includes: Extract the first forward-backward mapping association feature and the second forward-backward mapping association feature from the forward-backward mapping association feature sequence, perform forward-backward semantic association recognition, and add the recognition result into the first semantic association memory unit; Based on the first semantic association memory unit, the third forward-backward mapping association feature in the forward-backward mapping association feature sequence is subjected to forward-backward semantic association memory iteration to obtain the second semantic association memory unit; This process continues until the last element of the forward-backward mapping associated feature sequence is reached, at which point the performance memory is obtained.

[0037] In one embodiment, the fulfillment event sequence refers to a list of all fulfillment events, such as outbound shipments, transportation milestones, exception descriptions, route changes, and arrivals, arranged chronologically to reflect the development sequence of the fulfillment process. Forward correlation feature identification involves tracing back from the current point in time to find preceding event features related to the current event, such as whether the current transportation delay is related to previous late shipments. Backward correlation feature identification involves inferring future event features that may be affected by the current event, such as whether splitting the current batch might cause delays in subsequent nodes. A forward-backward mapping correlation feature sequence maps forward and backward correlation features one-to-one to determine the causal relationship or strength of connection between individual events in the entire fulfillment chain.

[0038] For example, is the current transportation node delay related to previous batch splitting or outbound delays? If the logistics notes mention traffic congestion, "congestion" is used as a semantic feature associated with the delay event. Backward correlation feature recognition is used to analyze whether the current event may affect subsequent events. For example, if there is a route change, subsequent nodes may be delayed, or a current shortage may lead to future replenishment. These two types of information form forward correlation feature sequences and backward correlation feature sequences, respectively.

[0039] By mapping and aligning forward and backward features bit by bit, essentially matching cause and effect, the model can understand the causal relationships between events. Then, iterative learning is performed on the mapped feature sequence. First, the two most prominent related features are extracted from the sequence, and semantic analysis is used to determine whether there is a cooperative or causal relationship between them. The identification result is written into the first memory unit. The second iteration then analyzes the third related feature based on the first memory unit, continuously reinforcing existing memories and supplementing new features to construct an increasingly complete event association graph. This process continues until the end of the sequence, ultimately yielding the performance execution memory—a complete performance event association graph that reflects how events influence each other and are linked together in the entire performance process.

[0040] Furthermore, based on the performance execution memory, a second layer of perceived risk identification is performed to obtain the second layer of perceived risk identification result. In this embodiment, step S200 further includes: Multiple sample performance execution memories and multiple sample second-level perception risk identification results are used as training samples; The framework built on the feedforward neural network is trained under supervision based on the training samples until the training converges, and the trained second-level perceptual risk identifier is obtained. The second-level risk identification device is used to perform risk perception on the performance execution memory to obtain the second-level risk identification result.

[0041] In one embodiment, performance execution memories generated from a large amount of historical performance data are collected, along with the actual results of whether potential risks occurred during these performance execution processes. This yields multiple sample performance execution memories and multiple sample second-level risk identification results. These training samples are then input into a risk identification framework composed of a feedforward neural network for training. Each training iteration allows the model to better understand which event combinations lead to which risks. For example, the model automatically learns that batch splitting + route changes + abnormal notes indicating replenishment signifies replenishment risk. Or it learns that slight delays + multi-node stall patterns often indicate transportation anomaly risk. After continuous training, the model eventually reaches training convergence, indicating that it has a stable ability to identify implicit risks.

[0042] Finally, in actual operation, the performance execution memory generated during the current performance process is input into the trained second-level risk perception identifier, which automatically determines whether there are hidden risks. For example, the model may identify that although there is no obvious delay on the surface, similar combinations of past events usually cause delivery deviations, thus issuing an early warning of implicit risks. By identifying risks that cannot be seen through direct rule comparison through machine learning models, deeper risk signals are provided for subsequent risk management decisions.

[0043] Step S300: Implicitly align and weight features of the first and second risk identification results to construct a dual risk identification result; Furthermore, implicit alignment and feature weighting are performed on the first-level risk identification result and the second-level risk identification result to construct a dual risk identification result. Step S300 in this embodiment of the application further includes: Implicit alignment is performed on risks of the same type in the first and second levels of risk identification results to obtain implicit alignment results. The implicit alignment results are weighted according to preset weights to obtain the dual risk perception results.

[0044] As mentioned above, the first level of risk identification results is obtained through explicit comparisons, such as time difference, quantity difference, and missing nodes, to determine the specific type and degree of deviation, for example, a 2-day delay or 50 fewer items shipped. The second level of risk identification results are potential risks identified through performance execution memory and implicit models, such as weak signal chain risks.

[0045] Since explicit and implicit risks often provide different levels of information for the same risk category, the first step is to check whether they involve the same type of risk. For example, the first level of identification might detect insufficient shipments, while the second level might predict that the final quantity may still be insufficient due to replenishment risks. This process maps both types of results to the same risk category and automatically performs semantic unification, essentially aligning the actual and potential problems into a single risk perspective, resulting in an implicit alignment.

[0046] Subsequently, these aligned risks are weighted according to preset weights. These weights can be set based on risk severity, model confidence, and historical impact. For example, explicit risks that have already occurred and are clearly defined may be given higher weights, while implicit, predictable risks may be given appropriate weights based on model confidence. This weighting process yields a final dual-risk perception result that more accurately reflects the actual impact of the risks. By integrating explicit biases and implicit risks into a unified risk assessment system, companies can comprehensively understand all the risks they face in the current performance process from a single viewpoint, thus providing an accurate basis for subsequent strategy formulation.

[0047] Step S400: Determine the performance process risk management strategy based on the dual risk perception results, and carry out risk management according to the performance process risk management strategy.

[0048] Preferably, a comprehensive assessment is conducted of the risk level, risk type, and stage of risk occurrence under the current performance status. For example, if explicit risk indicates a delayed shipment, while implicit risk indicates an abnormal trend in the subsequent transportation chain, the overall performance risk will be judged to be high. Based on this risk information, risk management strategies will be automatically formulated according to preset risk strategy templates or adaptive strategy generation rules, such as: immediately notifying the supplier to resend goods, adjusting the transportation route at the arrival node, reducing the monitoring frequency of transportation nodes, and triggering manual intervention.

[0049] Risk management actions are automatically implemented based on pre-defined strategies. For example, when a potential replenishment risk is identified, the replenishment process may be triggered in advance; when a route change is detected that may cause delays, a rerouting suggestion may be issued to the logistics carrier; when the risk is judged to be low, only light monitoring and alerts may be issued.

[0050] Example 2, based on the same inventive concept as the performance process risk management method combining multi-source data integration and analysis in the foregoing examples, as shown in the appendix. Figure 2 As shown, this application provides a performance process risk management system that integrates multi-source data analysis. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0051] Data processing module 11 is used to acquire contract data and supply chain data, perform data cleaning and standardization processing, and obtain a set of performance plan information and a set of performance execution events; Risk perception module 12 is used to perceive the risk of the performance plan information set and the performance execution event set according to the dual inconsistency perception risk identification mechanism, and obtain the first perception risk identification result and the second perception risk identification result. The feature weighting module 13 is used to implicitly align and weight the first-level risk recognition result and the second-level risk recognition result to construct a dual risk recognition result; Risk management module 14 is used to determine the performance process risk management strategy based on the dual risk perception results, and to perform risk management according to the performance process risk management strategy.

[0052] Furthermore, the risk perception module 12 is used to perform the following steps: According to the preset explicit comparison items, the first-level perception risk identification is performed on the performance plan information set and the performance execution event set to obtain the first-level perception risk identification result; The set of performance events is subjected to sequential semantic association memory iteration to obtain performance execution memory; Based on the performance execution memory, a second layer of perceived risk identification is performed to obtain the second layer of perceived risk identification result.

[0053] Furthermore, the risk perception module 12 is used to perform the following steps: The set of performance events is subjected to structured execution information extraction to obtain a structured data set; According to the preset explicit comparison items, the performance plan information set is compared with the structured data set item by item to obtain multiple explicit deviation results and multiple explicit deviation degrees. The multiple explicit deviation results and multiple explicit deviation degrees are used as the first-level perceived risk identification results; The preset explicit comparison items include delivery time comparison items, quantity consistency comparison items, batch structure comparison items, event sequence comparison items, logistics node integrity comparison items, material consistency comparison items, and delivery location comparison items.

[0054] Furthermore, the risk perception module 12 is used to perform the following steps: The set of performance execution events is sorted in chronological order to obtain the performance execution event sequence; Traverse the sequence of performance execution events, and take each performance execution event as the center to perform forward correlation feature identification to obtain a forward correlation feature sequence; Traverse the sequence of performance execution events, and take each performance execution event as the center to perform backward correlation feature identification to obtain the backward correlation feature sequence; The forward-associated feature sequence and the backward-associated feature sequence are mapped and aligned to obtain a forward-backward mapped associated feature sequence; Based on the forward-backward mapping associated feature sequence, the forward and backward semantic association memory is iterated to obtain the performance execution memory.

[0055] Furthermore, the risk perception module 12 is used to perform the following steps: Extract the first forward-backward mapping association feature and the second forward-backward mapping association feature from the forward-backward mapping association feature sequence, perform forward-backward semantic association recognition, and add the recognition result into the first semantic association memory unit; Based on the first semantic association memory unit, the third forward-backward mapping association feature in the forward-backward mapping association feature sequence is subjected to forward-backward semantic association memory iteration to obtain the second semantic association memory unit; This process continues until the last element of the forward-backward mapping associated feature sequence is reached, at which point the performance memory is obtained.

[0056] Furthermore, the risk perception module 12 is used to perform the following steps: Multiple sample performance execution memories and multiple sample second-level perception risk identification results are used as training samples; The framework built on the feedforward neural network is trained under supervision based on the training samples until the training converges, and the trained second-level perceptual risk identifier is obtained. The second-level risk identification device is used to perform risk perception on the performance execution memory to obtain the second-level risk identification result.

[0057] Furthermore, the feature weighting module 13 is used to perform the following steps: Implicit alignment is performed on risks of the same type in the first and second levels of risk identification results to obtain implicit alignment results. The implicit alignment results are weighted according to preset weights to obtain the dual risk perception results.

[0058] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0059] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0060] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A risk management method for contract performance processes that combines multi-source data integration and analysis, characterized in that: The method includes: Acquire contract and supply chain data, perform data cleaning and standardization, and obtain a set of performance plan information and a set of performance execution events; Risk perception is performed on the set of performance plan information and the set of performance execution events according to the dual inconsistency perception risk identification mechanism to obtain the first perception risk identification result and the second perception risk identification result. Implicit alignment and feature weighting are performed on the first and second risk identification results to construct a dual risk identification result; Based on the results of the dual risk perception, a risk management strategy for the performance process is determined, and risk management is carried out according to the risk management strategy for the performance process.

2. The performance process risk management method combining multi-source data integration and analysis as described in claim 1, characterized in that, Risk perception is performed on the performance plan information set and the performance execution event set according to the dual inconsistency perception risk identification mechanism to obtain the first perception risk identification result and the second perception risk identification result, including: According to the preset explicit comparison items, the first-level perception risk identification is performed on the performance plan information set and the performance execution event set to obtain the first-level perception risk identification result; The set of performance events is subjected to sequential semantic association memory iteration to obtain performance execution memory; Based on the performance execution memory, a second layer of perceived risk identification is performed to obtain the second layer of perceived risk identification result.

3. The performance process risk management method combining multi-source data integration and analysis as described in claim 2, characterized in that, According to preset explicit comparison items, the first-level perceived risk identification is performed on the performance plan information set and the performance execution event set to obtain the first-level perceived risk identification result, including: The set of performance events is subjected to structured execution information extraction to obtain a structured data set; According to the preset explicit comparison items, the performance plan information set is compared with the structured data set item by item to obtain multiple explicit deviation results and multiple explicit deviation degrees. The multiple explicit deviation results and multiple explicit deviation degrees are used as the first-level perceived risk identification results; The preset explicit comparison items include delivery time comparison items, quantity consistency comparison items, batch structure comparison items, event sequence comparison items, logistics node integrity comparison items, material consistency comparison items, and delivery location comparison items.

4. The performance process risk management method combining multi-source data integration and analysis as described in claim 2, characterized in that, The set of performance execution events is subjected to sequential semantic association memory iteration to obtain performance execution memory, including: The set of performance execution events is sorted in chronological order to obtain the performance execution event sequence; Traverse the sequence of performance execution events, and take each performance execution event as the center to perform forward correlation feature identification to obtain a forward correlation feature sequence; Traverse the sequence of performance execution events, and take each performance execution event as the center to perform backward correlation feature identification to obtain the backward correlation feature sequence; The forward-associated feature sequence and the backward-associated feature sequence are mapped and aligned to obtain a forward-backward mapped associated feature sequence; Based on the forward-backward mapping associated feature sequence, the forward and backward semantic association memory is iterated to obtain the performance execution memory.

5. The performance process risk management method combining multi-source data integration and analysis as described in claim 4, characterized in that, Based on the aforementioned forward-backward mapping associated feature sequence, iterative forward-backward semantic association memory is performed to obtain the performance execution memory, including: Extract the first forward-backward mapping association feature and the second forward-backward mapping association feature from the forward-backward mapping association feature sequence, perform forward-backward semantic association recognition, and add the recognition result into the first semantic association memory unit; Based on the first semantic association memory unit, the third forward-backward mapping association feature in the forward-backward mapping association feature sequence is subjected to forward-backward semantic association memory iteration to obtain the second semantic association memory unit; This process continues until the last element of the forward-backward mapping associated feature sequence is reached, at which point the performance memory is obtained.

6. The performance process risk management method combining multi-source data integration and analysis as described in claim 2, characterized in that, Based on the performance execution memory, a second layer of perceived risk identification is performed to obtain the second layer of perceived risk identification results, including: Multiple sample performance execution memories and multiple sample second-level perception risk identification results are used as training samples; The framework built on the feedforward neural network is trained under supervision based on the training samples until the training converges, and the trained second-level perceptual risk identifier is obtained. The second-level risk identification device is used to perform risk perception on the performance execution memory to obtain the second-level risk identification result.

7. The performance process risk management method combining multi-source data integration and analysis as described in claim 1, characterized in that, Implicit alignment and feature weighting are performed on the first-level and second-level risk identification results to construct a dual-risk perception result, including: Implicit alignment is performed on risks of the same type in the first and second levels of risk identification results to obtain implicit alignment results. The implicit alignment results are weighted according to preset weights to obtain the dual risk perception results.

8. A performance process risk management system that integrates and analyzes multi-source data, characterized in that: The system is used to implement the performance process risk management method according to any one of claims 1-8, which combines multi-source data integration and analysis. The system includes: The data processing module is used to acquire contract data and supply chain data, perform data cleaning and standardization processing, and obtain a set of performance plan information and a set of performance execution events. The risk perception module is used to perceive the risk of the performance plan information set and the performance execution event set according to the dual inconsistency perception risk identification mechanism, and obtain the first perception risk identification result and the second perception risk identification result. The feature weighting module is used to implicitly align and weight the first-level risk identification result and the second-level risk identification result to construct a dual risk identification result; The risk management module is used to determine the risk management strategy for the performance process based on the results of the dual risk perception, and to carry out risk management according to the risk management strategy for the performance process.

Images