Multi-source heterogeneous data fused supply chain demand prediction model construction method and system

Abstract

The invention relates to a project management and data processing technology, in particular to a supply chain demand prediction model construction method and system fusing multi-source heterogeneous data, and the method comprises the steps: presetting risk buffer time for a task depending on a heterogeneous data source, the risk buffer time being associated with a risk attribute of the heterogeneous data source, an activation condition based on the running state of the heterogeneous data source is set; obtaining instant information of the running state of the heterogeneous data source, wherein the instant information comprises a data transmission state, an interface response state and / or data accuracy; when the instant information meets the activation condition of the corresponding task, the activated risk buffer time associated with the task is adjusted to update the task plan construction period; and after the task plan construction period is updated, recalculating the task key path, and outputting early warning information indicating task progress change. The method has the advantages that dynamic and active project progress management can be realized, and project predictability and control capability are improved.

Description

Technical Field

[0001] The present application relates to project management and data processing technology, and in particular to a method and system for constructing a supply chain demand forecasting model that integrates multi-source heterogeneous data. Background Art

[0002] In the process of enterprise informatization and digital transformation, building accurate supply chain demand forecasting models is crucial for optimizing inventory management, improving operational efficiency, and reducing costs. These model building projects often rely on integrating heterogeneous data from multiple sources to enhance forecast accuracy and robustness. These data sources include not only structured data from internal sales, inventory, and production systems, but also potentially include capacity data from external suppliers, transportation data from logistics carriers, industry reports from market research firms, macroeconomic indicators, and even unstructured or semi-structured data such as social media and news.

[0003] When initiating such projects, project management teams typically develop a detailed project schedule based on established project management processes. This plan meticulously outlines every step, including estimated durations and dependencies, from data collection, data cleaning and conversion, data fusion, feature engineering, model design and training, model validation and tuning, to final model deployment. However, during project implementation, especially when relying on multi-source, heterogeneous data input, project schedule management often quickly encounters severe challenges. Data acquisition often takes longer than planned due to significant differences in acquisition methods, update frequencies, data formats, and quality standards across different data sources. This makes data preparation a major bottleneck not only in the early stages of a project but also throughout its lifecycle.

[0004] Delays in the data preparation phase inevitably squeeze the time reserved for subsequent key technical steps, such as model design, model training, and model validation, like a domino effect. Some heterogeneous data sources, crucial to model prediction results, face significant uncertainty regarding their inherent stability and the timeliness of data updates. When these key data sources experience intermittent outages, significant fluctuations in data quality, or far lower-than-expected update frequency, not only does this directly impact the progress of the current data acquisition task, but it also necessitates the suspension of subsequent model training and validation based on this incomplete, inaccurate, or outdated data, potentially even requiring rework of already completed work. This chain reaction, triggered by specific data source issues, further disrupts the original project schedule, making tracking, controlling, and predicting project progress extremely difficult.

[0005] In view of the above problems, the existing technology is in urgent need of improvement. Summary of the Invention

[0006] In order to address the deficiencies of the existing technology, the present application provides a method and system for constructing a supply chain demand forecasting model that integrates multi-source heterogeneous data, which has the advantages of being able to achieve dynamic and proactive project progress management and improve project predictability and control capabilities.

[0007] This application provides a method for constructing a supply chain demand forecasting model that integrates multi-source heterogeneous data, the method comprising: Preset risk buffer time for tasks that rely on heterogeneous data sources. The risk buffer time is associated with the risk attributes of the heterogeneous data sources. Activation conditions are set based on the operating status of the heterogeneous data sources. The activation conditions are related to the availability, data quality, and update frequency of the heterogeneous data sources. Obtain real-time information on the operational status of heterogeneous data sources, including data transmission status, interface response status, and / or data accuracy; When the instant information meets the activation conditions of the corresponding task, the activated risk buffer time associated with the task is adjusted to update the task schedule; After the task plan duration is updated, the task critical path is recalculated and warning information indicating changes in task progress is output.

[0008] Through the above solution, dynamic and proactive project progress management can be achieved, and project predictability and control capabilities can be improved.

[0009] To further address the problem, this application also proposes presetting a risk buffer time for tasks that rely on heterogeneous data sources. The risk buffer time is associated with the risk attributes of the heterogeneous data sources, and activation conditions are set based on the operating status of the heterogeneous data sources, including: When introducing new heterogeneous data sources, perform a phased trial process on the heterogeneous data sources, record the operational performance data and technical interaction information, and set corresponding trial conditions in multiple stages of the phased trial process; Based on the operational performance data and technical interaction information, an initial risk assessment result is formed. The risk assessment result includes the initial risk attributes of the new heterogeneous data source; Configure the corresponding risk buffer time based on the initial risk assessment results, and set activation conditions based on the operating status of heterogeneous data sources.

[0010] Through the above solution, a structured new data source risk assessment and configuration process is provided.

[0011] To improve the solution, this application also proposes that the method further includes: During the execution of the task, continuously monitor changes in the initial assessment basic information related to heterogeneous data sources, including data interface documents, data quality history reports and / or supplier service commitments; The risk attributes are updated according to the change information, and the risk buffer time that has not yet been activated is adjusted according to the updated risk attributes.

[0012] Through the above solution, dynamic updates of risk attributes and buffer times based on continuous monitoring are achieved.

[0013] To further solve the problem, the present application further proposes that, when the instant information meets the activation conditions of the corresponding task, the activated risk buffer time associated with the task is adjusted to update the task's planned duration, including: When the instant information meets the activation conditions of the corresponding task, the status information of the problem existing in the corresponding heterogeneous data source is obtained, and the status information includes the operation status of the data source, network problems and / or interface interruption; Based on the status information, the expected duration or impact of the problem is assessed to obtain a problem assessment result, which includes the number of days of delay and / or the scope of impact; Compare the problem assessment results with the activated risk buffer time. If there is a deviation, adjust the risk buffer time again to update the corresponding task plan duration.

[0014] Through the above solution, task duration can be accurately adjusted based on real-time problem status and impact assessment.

[0015] To improve the solution, this application also proposes that status information is derived from multiple message channels to obtain status information of problems existing in corresponding heterogeneous data sources, including: Preset reliability indicators for each message channel, which reflect the accuracy, timeliness and consistency of the corresponding information channel; When conflicting information exists in multiple message channels, a predefined conflict handling strategy is applied based on the timeliness of acquisition and the reliability identifier corresponding to the message channel to disambiguate the conflicting information and form a unified evaluation basis for the status information of the problems existing in the corresponding heterogeneous data sources.

[0016] Through the above solution, the accuracy and reliability of data source status information are improved.

[0017] To further solve the problem, this application also proposes that the conflict handling strategy includes: Preset conflict mode library, store multiple conflict modes, and configure disambiguation rule sets for each mode; When the conflict information meets a certain conflict pattern, the corresponding disambiguation rule set is called to perform disambiguation processing.

[0018] Through the above scheme, a systematic conflict information processing mechanism is provided.

[0019] To improve the solution, this application also proposes: Obtaining disambiguation results and feedback information of the disambiguation rule set during the corresponding disambiguation process, including the disambiguation success rate and result accuracy; When the feedback information indicates that the disambiguation results do not meet the preset quality standards, analyze the reasons for not meeting them; Adjust existing rules in the disambiguation rule set or add new rules to the disambiguation rule set based on the reasons that are not met.

[0020] Through the above solution, continuous optimization of conflict handling rules is achieved.

[0021] To further solve the problem, this application also proposes to preset reliability identifiers corresponding to each message channel, including: Preset the reliability mark of each message channel; During the execution of the task, continuously monitor the success rate, timeliness and consistency of the information provided by the message channel corresponding to the reliability indicator; Based on performance results, reliability indicators are evaluated and updated.

[0022] Through the above scheme, the real-time and accuracy of information channel reliability assessment are ensured.

[0023] To improve the solution, this application also proposes to evaluate and update the reliability mark based on the performance results, including: Get preset performance indicators of performance results; Compare the performance results with the preset performance indicators, calculate the comprehensive score or reliability level of the message channel, and update the reliability mark.

[0024] Through the above scheme, a quantitative channel reliability assessment method is provided.

[0025] To further solve the problem, this application also proposes a supply chain demand forecasting model construction system that integrates multi-source heterogeneous data, including: The risk buffer configuration module is used to preset the risk buffer time for tasks that rely on heterogeneous data sources. The risk buffer time is associated with the risk attributes of the heterogeneous data sources, and the activation conditions based on the operating status of the heterogeneous data sources are set. The activation conditions are related to the availability, data quality, and update frequency of the heterogeneous data sources. A status monitoring module is used to obtain real-time information on the operating status of heterogeneous data sources, including data transmission status, interface response status and / or data accuracy; A plan adjustment module is used to adjust the activated risk buffer time associated with the task to update the task schedule when the instant information meets the activation conditions of the corresponding task; The calculation and warning module is used to recalculate the critical path of the task after the task plan duration is updated, and output warning information indicating the change in task progress.

[0026] Through the above scheme, a system carrier for implementing the above method is provided.

[0027] In summary, the present application provides a method and system for constructing a supply chain demand forecasting model that integrates multi-source heterogeneous data. By introducing a method for dynamically adjusting the risk buffer time based on the risk and real-time status of heterogeneous data sources and updating the task plan duration, it solves the problem of project schedule management caused by the uncertainty of multi-source heterogeneous data that is difficult to effectively deal with in the existing technology. It has the advantages of being able to achieve dynamic and proactive project schedule management and improve project predictability and control capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A flowchart of a method for constructing a supply chain demand forecasting model that integrates multi-source heterogeneous data is provided in one embodiment of the present application.

[0029] Figure 2 One of the flow charts of a method for constructing a supply chain demand forecasting model that integrates multi-source heterogeneous data provided in another embodiment of the present application.

[0030] Figure 3 The second flowchart of a method for constructing a supply chain demand forecasting model by integrating multi-source heterogeneous data is provided in another embodiment of the present application.

[0031] Figure 4 The third flowchart of a method for constructing a supply chain demand forecasting model that integrates multi-source heterogeneous data is provided in another embodiment of the present application.

[0032] Figure 5 This is a fourth flow chart of a method for constructing a supply chain demand forecasting model that integrates multi-source heterogeneous data, provided in another embodiment of the present application.

[0033] Figure 6 This is a fifth flow chart of a method for constructing a supply chain demand forecasting model that integrates multi-source heterogeneous data, provided in another embodiment of the present application.

[0034] Figure 7 FIG6 is a flowchart of a method for constructing a supply chain demand forecasting model by integrating multi-source heterogeneous data, provided in another embodiment of the present application.

[0035] Figure 8 FIG7 is a flowchart of a method for constructing a supply chain demand forecasting model that integrates multi-source heterogeneous data, provided in another embodiment of the present application.

[0036] Figure 9 FIG8 is a flowchart of a method for constructing a supply chain demand forecasting model by integrating multi-source heterogeneous data, provided in another embodiment of the present application.

[0037] Figure 10 A flowchart of a supply chain demand forecasting model construction system that integrates multi-source heterogeneous data, provided in another embodiment of the present application.

[0038] In the figure: 1. Risk buffer configuration module; 2. Status monitoring module; 3. Plan adjustment module; 4. Calculation and early warning module. DETAILED DESCRIPTION

[0039] The technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments. The components of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for which protection is claimed, but merely represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of this application.

[0040] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and should not be understood as indicating or implying relative importance.

[0041] Traditional supply chain demand forecasting models that integrate heterogeneous data from multiple sources face significant challenges in actual implementation. These models are prone to sudden changes, such as interface interruptions, data delays, and significant quality fluctuations, as well as extreme sensitivity to the quality and timeliness of specific data sources. These challenges pose significant challenges to project schedule management. Traditional project management tools are inadequate in detecting and responding to these data-driven risks. They struggle to detect changes in data source status in real time, accurately assess the cascading schedule impact of single or multiple data source issues on all subsequent related tasks, dynamically identify resulting critical path drift and emerging schedule bottlenecks, and provide project managers with timely and effective decision-making to adjust project plans and allocate limited expert resources. This makes it difficult to achieve dynamic and effective control of the overall project schedule, directly impacting on-time and budgetary project delivery, and hindering the strategic goal of leveraging advanced forecasting models to improve supply chain efficiency.

[0042] Reference Figure 1 In this regard, this application proposes a method for constructing a supply chain demand forecasting model that integrates multi-source heterogeneous data, including: S1000: Preset a risk buffer time for tasks that rely on heterogeneous data sources. The risk buffer time is associated with the risk attributes of the heterogeneous data sources. Activation conditions based on the operating status of the heterogeneous data sources are also set. The activation conditions are related to the availability, data quality, and update frequency of the heterogeneous data sources. S2000: Acquire real-time information on the operating status of heterogeneous data sources, including data transmission status, interface response status, and / or data accuracy; S3000: When the instant information meets the activation condition of the corresponding task, the activated risk buffer time associated with the task is adjusted to update the task schedule; S4000: After the task plan duration is updated, the task critical path is recalculated and warning information indicating changes in task progress is output.

[0043] Among them, in this embodiment, risk buffer time refers to the additional time reserved for tasks that rely on a specific data source, which can be determined based on the historical performance of the data source, the supplier service level agreement or the experience of domain experts, and is mainly to deal with possible delays or interruptions of the data source; the risk attribute of heterogeneous data sources refers to the assessment of the uncertainty or potential problems of the data source in terms of availability, data quality, update frequency, etc., which can be expressed by risk scores, risk levels or risk categories, and is mainly to quantify the potential impact of different data sources on project progress; activation conditions refer to specific events or status thresholds that trigger the adjustment of risk buffer time, which can be set as the number of consecutive failures of the data source interface, the data update delay duration or the data quality check failure rate, and is mainly to ensure that the dynamic adjustment mechanism is only activated when there is actually a problem with the data source; among them, the real-time information on the operating status of the heterogeneous data source refers to the real-time monitoring data of the current working status of the data source, which may include network connectivity, API response time or data record Record integrity, which is mainly to timely perceive abnormal situations of data sources; adjusting the activated risk buffer time associated with the task means modifying the risk buffer time that has previously taken effect due to the satisfaction of activation conditions based on the real-time status information of the data source. It can increase or decrease the buffer time according to the severity of the problem or the expected duration. It is mainly to make the planned task duration more accurately reflect the actual situation; among them, recalculating the task critical path means using project management algorithms (such as the critical path method) to reanalyze the dependencies and durations of all tasks after the planned task duration changes. It can be implemented using project management software or customized algorithms. It is mainly to identify the task sequence that affects the total project duration; among them, outputting early warning information indicating changes in task progress means that the system generates and sends notifications to inform the project team or relevant parties of changes in task duration or critical path. It can be in the form of email, message push or dashboard display. It is mainly to promptly remind managers to pay attention and take countermeasures.

[0044] The core innovation of this application is that by combining the real-time monitoring of the operating status of heterogeneous data sources with a dynamic risk buffer mechanism preset based on risk attributes, and linking the real-time adjustment of task planning duration and critical path, it solves the problem of dynamic project progress management caused by the uncertainty of multi-source heterogeneous data that traditional methods are difficult to cope with, thereby achieving the effect of improving the accuracy of project progress prediction and realizing proactive risk control.

[0045] The solution of this application addresses the uncertainty of heterogeneous data sources by establishing a dynamic project schedule management mechanism. First, for tasks that rely on different heterogeneous data sources, corresponding risk buffers are pre-set based on the inherent risk characteristics of the data sources. A set of activation conditions based on the operating status of each data source is also defined. These conditions reflect key indicators of data source availability, data quality, or update frequency. During project execution, the system continuously obtains real-time information on the operating status of each heterogeneous data source, such as whether data transmission is smooth, whether the interface responds promptly, and whether the acquired data is accurate. When real-time information from a data source indicates that its operating status meets the pre-set activation conditions, the system triggers an adjustment process for the risk buffer for tasks that rely on that data source. This adjustment is dynamic, modifying the activated risk buffer based on the specific issues (such as the type, severity, or expected duration). The adjusted risk buffer is directly used to update the planned duration of the task. Changes in task durations affect the entire project plan. Therefore, after the planned duration of a task is updated, the system immediately recalculates the critical path of the entire project to identify new schedule bottlenecks or changes in the total project duration. Finally, the system outputs early warning information indicating these progress changes, promptly notifying project managers and teams so they can quickly respond and take necessary countermeasures, such as reallocating resources or adjusting subsequent task schedules. Through this mechanism of real-time monitoring, conditional triggering, dynamic adjustment, and coordinated feedback, this solution effectively perceives and addresses the uncertainties introduced by heterogeneous data sources, enabling refined and proactive management of project progress.

[0046] In a specific embodiment, a project management system module can be constructed to implement this method. This module can include a data source risk configuration interface, allowing project managers to set risk attributes (e.g., high risk, medium risk, low risk) for each heterogeneous data source (e.g., an external supplier's API interface). Based on this, the project manager can automatically or manually configure the initial risk buffer for tasks that rely on that data source (e.g., a 3-day buffer for tasks with high-risk data sources, and a 1-day buffer for tasks with medium-risk data sources). Furthermore, within this interface or another configuration interface, activation conditions can be set for each data source. For example, for an API interface data source, the activation condition can be set as "three consecutive API call failures" or "the data update timestamp is 24 hours later than expected." The system can include a data source monitoring agent, deployed at the interface layer that interacts with each heterogeneous data source. This agent continuously obtains real-time operational status information from the data source and sends this information to the project management system module. When the real-time information reported by the monitoring agent meets the activation conditions for a data source that a task depends on, the project management system module initiates the plan adjustment process. This process can first attempt to obtain more detailed information about the issue's status (for example, through error log analysis or communication with the data source provider). It then uses pre-set rules or algorithms to estimate the issue's expected duration or impact. For example, if the error log indicates a temporary network fluctuation, the estimated delay might be half a day; if it's a data source server maintenance notification, the estimated delay might be one day. The system compares this estimate with the task's activated risk buffer and adjusts the task's scheduled duration accordingly. For example, if the estimated delay is one day and the activated buffer is two days, the system might extend the task's duration by one day and mark the task as having used the one-day buffer. If the estimated delay is three days, the system might extend the task's duration by three days and mark the task as having used the full two-day buffer and incurred an additional one-day delay. After the task duration is updated, the system's built-in project scheduling engine automatically reruns the critical path algorithm to update the project's total duration and critical task sequence. The system then outputs early warning information indicating task progress changes and potential project delays through built-in notification services, such as sending messages to the project team's collaboration platform or updating progress status on the project dashboard.

[0047] Reference Figure 2 Furthermore, in another embodiment of the present application, step S1000 includes: S1100: When a new heterogeneous data source is introduced, a phased trial process is performed on the heterogeneous data source, and operational performance data and technical interaction information are recorded. The phased trial process sets corresponding trial conditions in multiple stages; S1200: Based on the operational performance data and technical interaction information, an initial risk assessment result is generated. The risk assessment result includes initial risk attributes of the new heterogeneous data source. S1300: Configure the corresponding risk buffer time according to the initial risk assessment result, and set the activation conditions based on the operating status of the heterogeneous data source.

[0048] In this embodiment, the phased trial process involves dividing the testing and verification of new heterogeneous data sources into multiple, sequential, and clearly defined phases. Each phase gradually increases the complexity and data volume of the test to simulate actual application scenarios and observe performance. This can be achieved by conducting small-scale data access tests in a sandbox or isolated environment, functional and performance verification on medium-sized datasets, and stability testing under data volumes and concurrent requests close to those in actual production environments.

[0049] Performance data refers to data collected during phased trials on the operational status and performance of heterogeneous data sources, such as data transfer rates, interface response times, data processing delays, system resource utilization, and error rates. This data can be obtained using automated monitoring tools, log analysis systems, or manual recording. Its purpose is to provide quantifiable, objective data reflecting the actual operational status of the new data source.

[0050] Technical interaction information refers to information generated during the phased trial process when conducting technical integration and troubleshooting with heterogeneous data sources, such as interface call logs, error codes, exception stack traces, communication records with data source providers, key information in technical documentation, and compatibility test results. This information can be obtained through logging systems, issue tracking systems, document management systems, or manual recording.

[0051] Trial conditions refer to the pre-set criteria or requirements for each stage of the phased trial process used to measure the success of that stage, such as a data transmission success rate of 95%, an average interface response time of less than 500 milliseconds, a specific data type error rate of less than 1%, and a system stability of 99.9%. These criteria can be based on business needs, technical specifications, or industry standards. Their purpose is to provide clear criteria or requirements for each trial stage, ensuring the accuracy of the assessment process. The initial risk assessment results refer to the conclusions drawn from a preliminary analysis and assessment of the potential risks of new heterogeneous data sources based on the operational performance data and technical interaction information collected during the phased trial. These can be expressed using risk scoring models, expert ratings, or qualitative descriptions. Their purpose is to transform observations made during the trial into a preliminary quantitative or qualitative understanding of the risks. Initial risk attributes refer to the elements in the initial risk assessment results that reflect the inherent risk characteristics of new heterogeneous data sources, such as data availability risk, data quality risk, technical integration risk, and vendor support risk. These can be expressed using risk levels, risk categories, or specific risk descriptions. Its purpose is to identify and summarize the main risk points of new data sources.

[0052] Through the synergistic effect of the aforementioned steps, the solution of this application enables a systematic assessment and pre-defined response measures for the risks of newly introduced heterogeneous data sources. When the system needs to introduce a new heterogeneous data source to support the construction of a supply chain demand forecasting model, a structured assessment process is initiated to mitigate the uncertainties and risks that this new data source may introduce. First, a phased trial is conducted on the new data source. This process is not a simple one-time test, but rather is broken down into multiple, sequential, objectively defined phases, each with pre-defined trial conditions. During the trial, the system continuously records operational performance data about the data source, such as data transmission stability, interface responsiveness, and data processing efficiency. It also records technical interaction information generated during technical integration and troubleshooting, such as error types encountered, solutions, and details of communication with the data source provider. This recorded data and information forms the basis for an objective assessment. After the phased trial is complete, the system conducts a comprehensive analysis based on the collected operational performance data and technical interaction information to form an initial risk assessment. This assessment result specifically reflects the initial risk attributes of the new heterogeneous data source, such as its potential risk level or specific issues in terms of availability, data quality, and technical integration. Based on this initial risk assessment, the system configures a corresponding risk buffer for tasks that rely on the new data source. The higher the risk attribute, the longer the configured risk buffer may be. The system also sets activation conditions based on the operational status of the data source. These activation conditions are related to the data source's availability, data quality, and update frequency. For example, if the assessment results indicate that the data source has a medium risk in terms of data transmission stability, a moderate risk buffer may be configured, activated when the data transmission success rate falls below a certain threshold. Through this series of steps, the system can fully assess the risks of the new data source before officially using it for model construction and pre-configure countermeasures, thereby mitigating the impact of the introduction of the new data source on the overall project schedule. This pre-setting method provides the foundation and prerequisite for dynamically adjusting the activated risk buffer during task execution based on the real-time operational status of the heterogeneous data source. This ensures that the pre-set risk buffer time and activation conditions are based on the actual assessment results of the new data source, rather than being arbitrarily set, making subsequent dynamic adjustments targeted and effective. This combination enables the entire method to not only dynamically respond to risks during task execution, but also scientifically evaluate potential risks and preset response plans before task execution, forming a complete risk management closed loop and improving the ability of supply chain demand forecasting model construction projects to deal with the uncertainty of heterogeneous data sources.

[0053] The following illustrates this with a specific example. When a new external market price data source is needed to enhance the accuracy of a supply chain demand forecasting model, the system initiates a phased trial of this new data source. For example, this trial process can be divided into three phases. The first phase can be conducted in an isolated sandbox environment, with the trial conditions set to test basic data interface connectivity and the ability to correctly parse a small amount of historical data. During this phase, the system records the success rate and response time of interface calls, as well as any errors or warnings encountered during data parsing, as performance data and technical interaction information. If the first phase passes, the system proceeds to the second phase, testing the data transmission speed, completeness, and preliminary data quality on a medium-sized, recent dataset. For example, this checks for a large number of missing values or outliers. The trial conditions for this phase can be set as data transmission speed reaching a certain threshold, data completeness exceeding a certain percentage, and anomaly rates for key fields falling below a certain percentage. The system continues to record performance data and technical interaction information. If the second phase also passes, the system proceeds to the third phase, conducting long-term stability testing with data volumes and concurrent requests close to those of a production environment, simulating the high load conditions of a real-world application scenario. The trial conditions for this phase can be set as continuous uninterrupted operation for a certain period of time, average response time that meets requirements, and an extremely low error rate. After the entire phased trial process is completed, the system will conduct a comprehensive analysis based on all recorded operational performance data and technical interaction information, such as interface stability records, data quality reports, and technical communication records with the data source provider, to form an initial risk assessment. This assessment will include the initial risk attributes of the market price data source, such as a medium risk for availability and a medium-to-high risk for data timeliness. Finally, based on this initial risk assessment, the system will configure a corresponding risk buffer for data acquisition and cleaning tasks that rely on this market price data source, such as adding an additional two days to the planned duration. Furthermore, the system will set activation conditions based on the data source's operational status. For example, the previously configured risk buffer will be activated if the data source's interface timeout exceeds three consecutive times, data update delays exceed 12 hours, or the key field missing rate exceeds 5%.

[0054] Reference Figure 3 Furthermore, another embodiment of the present application proposes: S5000: During task execution, continuously monitor changes in the initial assessment basic information related to heterogeneous data sources. The initial assessment basic information includes data interface documents, data quality history reports, and / or supplier service commitments. S6000: Update risk attributes according to the change information, and adjust the risk buffer time that has not yet been activated according to the updated risk attributes.

[0055] In this embodiment, "initial assessment basic information" refers to the baseline data and agreements regarding heterogeneous data sources used to form the initial risk assessment results. Specifically, it may include the technical specifications of the data source, historical performance records, and details of the commitments made by the service provider. Its purpose is to provide a preliminary understanding of the inherent characteristics and potential risks of the data source. "Change information" refers to any modifications or deviations detected in the initial assessment basic information during task execution. Specifically, it may be reflected in document version updates, changes in trends in reports, or revisions to the terms of service. Its purpose is to capture the evolution of the basic state of the data source. "Updating risk attributes" refers to revising the risk assessment results associated with the heterogeneous data source based on the detected change information. Specifically, this may be achieved by adjusting the risk level, risk score, or related risk parameters. Its purpose is to ensure that the risk assessment results reflect the current state of the data source. "Adjusting unactivated risk buffer time" refers to modifying the duration of tasks that have not yet been actually triggered or used and are reserved to address potential data source issues based on the updated risk attributes. Specifically, this may be achieved by increasing or decreasing the buffer time of specific tasks. Its purpose is to proactively manage potential future delays based on the latest risk assessment results.

[0056] The solution in this application continuously monitors the underlying fundamental information of heterogeneous data sources throughout the entire task execution cycle, such as their interface specifications, historical quality performance, and service provider commitments. When this fundamental information changes, the system promptly detects and identifies these changes. Based on this information, the system reassesses the risk attributes associated with the data source, such as the stability of data acquisition, the reliability of data quality, or the likelihood of service interruption. This continuous monitoring of changes in fundamental information and the dynamic updating of risk attributes enable this application to proactively identify potential increases or decreases in risk. Based on this, the system adjusts the preset risk buffer times for tasks that have not yet begun execution or encountered actual problems, based on the updated risk attributes. For example, if changes in fundamental information indicate an increase in risk for a data source, the system can increase the buffer time for future tasks that rely on that data source; conversely, if the risk decreases, the buffer time can be appropriately reduced. This mechanism, combined with a reactive mechanism that adjusts activated buffer times in the event of real-time operational status issues (such as data transmission interruptions or interface response delays), forms a more comprehensive risk management strategy. By continuously monitoring changes in basic information and proactively adjusting buffer times that have not yet been activated, this application can respond to potential problems earlier and reduce unexpected delays caused by changes in basic information, thereby improving the accuracy and reliability of task planning durations and effectively supporting dynamic control of project progress.

[0057] In some preferred embodiments, a task that relies on an external vendor's API (Application Programming Interface) to obtain data can be described. During the initial assessment phase, this task is assigned a risk buffer and activation conditions are set based on API response time and data accuracy. During task execution, the system can configure an automated tool to regularly review the vendor's API documentation update log and obtain weekly quality reports for this API data source from the internal data quality monitoring system. For example, if the automated tool detects an incompatible change in the data format described in the API documentation, or if the weekly quality report shows a significant increase in the data missing rate, these changes are identified as changes to the basic information used for the initial assessment. Based on these changes, the system can automatically increase the risk attribute of the vendor's API data source, for example, adjusting its data quality risk level from "low" to "medium," according to pre-set rules or models. The system then identifies all tasks that rely on this vendor's API data source and have not yet started or completed. The risk buffer associated with these tasks is considered the inactive risk buffer. Based on the updated risk attributes (e.g., a "medium" data quality risk level), the system can increase the buffer time for these unactivated risks according to a pre-set policy (e.g., increasing the buffer time by X days for each level of risk increase). This way, even before the actual data acquisition task encounters real-time issues caused by format changes or quality degradation, the system has already reserved more buffer time for subsequent data cleaning and processing tasks to address potential risks.

[0058] Reference Figure 4 Furthermore, in another embodiment of the present application, step S3000 includes: S3100: When the instant information meets the activation condition of the corresponding task, status information of the problem existing in the corresponding heterogeneous data source is obtained, the status information including the operation status of the data source, network problems and / or interface interruption; S3200: Based on the status information, the expected duration or impact of the problem is evaluated to obtain a problem evaluation result, which includes the number of days of delay and / or the scope of impact. S3300: Compare the problem assessment result with the activated risk buffer time. If there is a deviation, adjust the risk buffer time again to update the corresponding task plan duration.

[0059] In this embodiment, the status information of the problem existing in the heterogeneous data source refers to a specific description of the current operating status of the data source, which may include whether the data source is accessible, whether data transmission is normal, whether the interface is responsive, whether the data content is complete and accurate, etc., and may be obtained through methods such as monitoring tool collection, log analysis, and error report parsing. Based on the status information, assessing the expected duration or impact of the problem refers to using analytical methods to predict the possible duration of the problem or the scope of its impact on task execution based on the acquired status information. This may be performed through methods such as historical data analysis, expert system judgment, and machine learning model prediction. The problem assessment result refers to the quantitative or qualitative conclusion on the impact of the data source problem on the task, which may include the expected number of days of task delay, the number or type of tasks affected, etc. Comparing the problem assessment result with the activated risk buffer time refers to comparing the assessed problem impact result with the risk buffer time currently allocated and activated for the task, which may be performed through methods such as numerical comparison and range judgment. Adjusting the risk buffer time again refers to modifying the currently activated risk buffer time based on the comparison result, which may be adjusted by increasing, decreasing, or resetting.

[0060] The solution of this application first obtains status information about issues existing in heterogeneous data sources when the activation conditions are met. This provides detailed information on the nature and severity of the issues. Based on this status information, the expected duration or impact of the issue is further assessed, resulting in a quantitative or qualitative problem assessment. This assessment, based on the actual issue status, is then compared with the risk buffer time currently activated for the task. This comparison mechanism ensures that adjustments to the risk buffer time are not blind, but are based on an accurate assessment of the actual issue impact. Further adjustments are only made if there is a discrepancy between the assessment result and the existing buffer time. This refined adjustment process ensures that the risk buffer time more accurately reflects actual task delays. Building on the previous solution's pre-set risk buffer time, setting activation conditions, and adjusting it upon activation, this solution further refines the post-activation adjustment process, upgrading risk management from simple trigger-based adjustments to dynamic, precise adjustments based on the issue status. This allows for more accurate updates to task schedules and effectively addresses the complex impacts of heterogeneous data source issues.

[0061] In some preferred embodiments, a task in a supply chain demand forecasting model construction project relies on a real-time inventory data API provided by an external vendor. This task has a preset risk buffer, with activation criteria set to 10 consecutive minutes of API unresponsiveness. When the API is unresponsive for 15 consecutive minutes, the activation criteria are met. The system further obtains status information about issues with the corresponding heterogeneous data source and discovers that the API server returns a 503 error. Simultaneously, network monitoring indicates a sharp increase in network latency to the vendor server. Based on this status information, the system estimates the expected duration of the issue. Combining historical experience, the system predicts that the issue could last three hours, resulting in a projected delay of 0.5 days for the task. The resulting risk buffer is currently activated for the task at 0.2 days. Comparing the estimated risk buffer of 0.5 days with the actual activated risk buffer of 0.2 days reveals a discrepancy. Based on this discrepancy, the system adjusts the risk buffer again, adding 0.3 days to bring the total activated buffer to 0.5 days. Finally, the task's scheduled duration is updated based on the adjusted risk buffer.

[0062] Reference Figure 5 Furthermore, in another embodiment of the present application, the sub-step of step S3200: obtaining status information of problems existing in the corresponding heterogeneous data source includes: S3210: Preset reliability indicators for each message channel. The reliability indicators reflect the accuracy, timeliness, and consistency of the corresponding information channel. S3220: When conflicting information exists in multiple message channels, a predefined conflict handling strategy is applied based on the timeliness of acquisition and the reliability identifier corresponding to the message channel to disambiguate the conflicting information and form a unified evaluation basis for the status information of the problems existing in the corresponding heterogeneous data sources.

[0063] In this embodiment, the status information comes from multiple message channels. Message channels refer to different sources that provide heterogeneous data source status information. Specifically, they can be implemented using automated monitoring systems, log analysis tools, manual reporting systems, or third-party service interfaces. The purpose is to obtain information about data source status from multiple sources. Reliability identification refers to a mark used to measure the credibility of information provided by each message channel. It can be implemented by numerical scoring, grade classification or Boolean flag. Its purpose is to quantify the credibility of different message sources and provide a basis for conflict resolution. Conflicting information refers to the situation where multiple message channels provide inconsistent status descriptions of the same heterogeneous data source problem. Acquisition timeliness refers to the time difference between the time the system receives the information provided by the message channel and the current time. Its purpose is to reflect the timeliness of the information and serve as a reference factor for conflict resolution. Conflict resolution strategy refers to the method or set of rules used to resolve information conflicts between multiple message channels. It can be implemented by priority-based rules, weighted average algorithm or majority voting mechanism. Its purpose is to provide rules or methods for resolving information conflicts.

[0064] Disambiguation is the process of applying conflict resolution strategies to eliminate inconsistencies in conflicting information and produce a single, definitive state information result. Its purpose is to eliminate information inconsistencies and produce a single, definitive result. A unified assessment basis refers to standardized, conflict-free state information that, after disambiguation, is used to subsequently assess the impact of heterogeneous data source issues. Its purpose is to provide reliable, standardized information for subsequent assessments.

[0065] The solution of the present application obtains status information of problems existing in heterogeneous data sources from multiple message channels, and in response to possible conflicts in this information, pre-sets a reliability identifier for each message channel that reflects its accuracy, timeliness and consistency. Precisely because the status information comes from multiple channels and the reliability of each channel is different, when an information conflict is detected, the system can comprehensively consider the timeliness of information acquisition and the reliability identifier of the corresponding message channel, and apply a predefined conflict handling strategy for disambiguation. It is through this conflict disambiguation mechanism based on reliability and timeliness that the final status information is a unified, verified, and more reliable evaluation basis. This processed and accurate status information provides a solid foundation for subsequent assessment of the impact of problems based on status information and adjustment of risk buffer time, making subsequent task plan duration adjustments more accurate, thereby improving the robustness and effectiveness of the entire method.

[0066] In some preferred embodiments, the present application is implemented as follows. When obtaining problem status information about an external vendor's API data source, the system can configure three message channels to monitor the API's status: an automated API health check service (Channel A), a backend service that analyzes API call logs (Channel B), and an interface for receiving vendor email notifications (Channel C). The system can preset reliability indicators for these channels. For example, the reliability indicator for Channel A can be set to 0.9, Channel B can be set to 0.7, and Channel C can be set to 0.5. At a certain point in time, if Channel A reports a prolonged API response time (Status: Warning, Timeliness: T0), Channel B reports a large number of connection timeout errors in the log (Status: Severe Error, Timeliness: T0+2 minutes), and Channel C reports receiving an email notification from the vendor that the API is undergoing emergency maintenance and is expected to be down for 2 hours (Status: Down, Timeliness: T0+5 minutes), the system detects that these messages conflict. The system can then use a preset conflict resolution strategy, for example, stipulating that when "Down" status information is present, the final status will prioritize "Down," while also considering timeliness and reliability. The system recognizes that channel C has reported an "outage" status. The system can either accept the "outage" information from channel C or combine it with other information to make a judgment. After disambiguation, a unified status message can be generated, such as "API outage, expected to last 2 hours." This unified status message can be used to subsequently assess the impact on the task's scheduled duration.

[0067] Reference Figure 6 Furthermore, in another embodiment of the present application, it is proposed that the conflict handling strategy includes: A1: A preset conflict pattern library stores multiple conflict patterns and configures a disambiguation rule set for each pattern; A2: When conflicting information matches a certain conflict pattern, the corresponding disambiguation rule set is called to perform disambiguation processing.

[0068] Among them, in this embodiment, the conflict pattern library refers to a collection of multiple predefined conflict patterns, which can be implemented by a database table, a configuration file or a data structure in memory, and its purpose is to classify and manage possible conflict types; a conflict pattern refers to a specific type of conflict information expression, which can be defined by describing differences in information sources, inconsistent data values or timestamp conflicts, and its purpose is to identify and classify received conflict information; a disambiguation rule set refers to a series of processing rules set for a specific conflict pattern, which can be implemented by a conditional judgment sequence, a decision tree logic or a weighted scoring algorithm, and its purpose is to provide a specific method for resolving the corresponding conflict pattern.

[0069] The solution of the present application solves the problem that fixed strategies cannot effectively handle diverse conflicts by introducing a refined conflict handling mechanism. Specifically, the system pre-builds a conflict pattern library, which defines in detail a variety of possible conflict scenarios and configures a special disambiguation rule set for each scenario. When the system receives conflicting information about the status of heterogeneous data sources from multiple message channels, it first analyzes the specific manifestations of these conflicting information and matches them with the patterns stored in the conflict pattern library. Once it is identified that the conflicting information meets a specific conflict pattern, the system will accurately call the disambiguation rule set associated with the pattern. These rule sets are designed for this specific conflict pattern and may take into account the reliability identification of the message channel, the timeliness of the information or other relevant factors, so that the logic that best suits the current conflict type can be applied for disambiguation processing. It is precisely because of the ability to dynamically select and apply different disambiguation rule sets based on the conflict pattern that the system can more accurately and effectively resolve various complex conflicts and form a more reliable and unified basis for evaluating the status of heterogeneous data sources. This processing method based on pattern matching and rule set calling significantly improves the accuracy of disambiguation compared to a single fixed strategy, thereby providing a more solid foundation for subsequent problem assessment and task plan adjustment based on status information.

[0070] In some preferred embodiments, specifically, assume that the system receives conflicting information about the status of a certain data source: message channel A (with a higher reliability indicator) reports "data interface interruption," while message channel B (with a lower reliability indicator) reports "data transmission delay." The system first matches this conflict information with a preset conflict pattern library. The library may contain a conflict pattern called "interface status and transmission status conflict," which defines the scenario when the operating status of a data source reported by different channels is inconsistent. The disambiguation rule set associated with this pattern may include the following rule: If one channel reports an interface interruption and the other reports a transmission delay, and the channel reporting the interface interruption has a higher reliability than the channel reporting the transmission delay, then the interface interruption information is adopted. The system recognizes that the current conflict conforms to this pattern and calls the corresponding rule set. According to the rule, since channel A has a higher reliability indicator than channel B, the system adopts the "data interface interruption" reported by channel A as the final unified status information. For example, if the same message channel C reports "Data quality degraded (error rate 5%)" and "Data quality degraded (error rate 10%)" within a short period of time, the conflict pattern library may contain a pattern called "Same-source data quality report conflict," whose rule set may dictate that the report with the latest timestamp be used. The system matches this pattern and, based on the rules, adopts the latter report, "Data quality degraded (error rate 10%)," as the unified status information. This allows the system to flexibly apply preset rules based on the specific conflict type, avoiding misjudgments that could result from simplistic, fixed rules.

[0071] Reference Figure 7 Furthermore, another embodiment of the present application further includes: A3: Obtain the disambiguation results and feedback information of the disambiguation rule set in the corresponding disambiguation process. The feedback information includes the disambiguation success rate and result accuracy. A4: When the feedback information indicates that the disambiguation results do not meet the preset quality standards, analyze the reasons for not meeting them; A5: Adjust the existing rules in the disambiguation rule set or add new rules to the disambiguation rule set based on the reasons that are not met.

[0072] In this embodiment, the disambiguation rule set refers to a set of predefined rules for resolving conflicts in multi-source information, which can be implemented by using rules set based on expert experience, rules generated based on machine learning models, or a combination of the two.

[0073] Among them, the disambiguation process refers to the process of applying the disambiguation rule set to handle conflicting information, which can be achieved by sequential execution of rules, parallel execution of rules, or priority-based rule matching and execution. The disambiguation result refers to the unified information obtained after the disambiguation process, which can be achieved by majority voting results, weighted average results, or results selected based on specific rules. Feedback information refers to data for evaluating the disambiguation effect, which includes the disambiguation success rate and result accuracy. The disambiguation success rate refers to the proportion of successfully resolved conflicts, and the result accuracy refers to the degree of conformity between the disambiguation result and the actual situation or expected result. Feedback information can be obtained by manual evaluation, comparison with known real data, or based on subsequent task performance evaluation.

[0074] The preset quality standard refers to the threshold or indicator for measuring whether the disambiguation result is acceptable. It can be set by a minimum disambiguation success rate, a minimum result accuracy rate, or an upper limit on the error rate in a specific scenario. The reason for non-satisfaction refers to the specific factors that lead to the disambiguation result not meeting the standard. It can be analyzed by methods such as insufficient rule coverage, rule conflicts, overly strict or overly loose rules, etc. Adjusting existing rules in the disambiguation rule set refers to modifying existing rules in the rule set. It can be achieved by modifying the rule conditions, rule actions, or rule priorities.

[0075] Adding new rules to the disambiguation rule set means adding new rules to the rule set, which can be achieved by manually writing new rules, mining new rules from historical data, or generating new rules based on model learning.

[0076] The solution of this application obtains performance data of the disambiguation rule set during the actual disambiguation process, namely, disambiguation results and feedback information. This feedback information quantifies the effectiveness of the rule set, such as the disambiguation success rate and result accuracy. Based on this feedback information, the system can determine whether the current disambiguation results meet the preset quality standards. If the standards are not met, the specific reasons for the poor disambiguation results are further analyzed, such as whether certain conflict patterns do not have corresponding rules, or whether the existing rules are too general or contain errors. It is precisely because of the in-depth analysis of the reasons for non-satisfaction that the disambiguation rule set can be optimized in a targeted manner. Based on the specific reasons analyzed, the rule set can be improved by adjusting existing rules (such as modifying rule conditions and priorities) or adding new rules (such as writing rules for new conflict patterns). This continuous evaluation, analysis, and optimization cycle enables the disambiguation rule set to be continuously improved as actual application scenarios change and problems are exposed, significantly improving the accuracy and reliability of multi-source information conflict disambiguation. Compared with the basic solution that only relies on static preset rules for disambiguation, this application adds the adaptive optimization capability of the rule set, which can effectively deal with complex and changeable actual data conflict situations, and solves the problem that the static rule set may cause the disambiguation results to be inaccurate or substandard.

[0077] In some preferred embodiments, the present application is implemented as follows: Suppose that when building a supply chain demand forecasting model, it is necessary to integrate estimated delivery date information from multiple channels, which may be conflicting. The system has a preset set of disambiguation rules, such as "prioritize information from channel A" and "take the average when channels B and C conflict." During actual processing, the system applies these rules to disambiguate conflicting estimated delivery date information, resulting in a unified estimated delivery date. Simultaneously, the system records the results of each disambiguation process and collects feedback information, such as calculating the percentage of successfully resolved conflicts as the disambiguation success rate. The accuracy of the results is assessed by comparing them with actual delivery dates or by manual sampling. Assume that the preset quality standards require a disambiguation success rate of 98% and an accuracy of 95%. If, after a period of operation, the feedback indicates a disambiguation success rate of only 95% and an accuracy of only 90%, these standards are not met, the system will analyze the reasons for the failure to meet the standards. For example, log analysis may reveal that some conflicting information involves new data sources not covered by the existing rule set, or that the existing rules are ineffective in handling certain types of conflicts (such as conflicting information about delays caused by extreme weather). Based on the analysis results, the system can adjust existing rules in the disambiguation rule set, such as modifying a rule's priority or conditions to make it more suitable for a specific scenario. It can also add new rules, such as writing new disambiguation logic for new data sources or specific types of conflicts. This way, the disambiguation rule set is continuously optimized, improving the accuracy and reliability of subsequent disambiguation processing.

[0078] Reference Figure 8 Furthermore, in another embodiment of the present application, the sub-step of step S3210: presetting the reliability identifier corresponding to each message channel includes: S3211: preset reliability identifiers for each message channel; S3212: During the execution of the task, continuously monitor the success rate, timeliness, and consistency of the information provided by the message channel corresponding to the reliability indicator; S3213: Evaluate and update reliability indicators based on performance results.

[0079] In this embodiment, message channels refer to different sources that provide status information on issues with heterogeneous data sources. These can be implemented using APIs, emails, manual reports, system logs, and other means. The success rate refers to the frequency or proportion of successful information acquisition by a message channel. This can be measured by the ratio of successful information acquisition attempts to the total number of attempts. Timeliness refers to the speed or immediacy with which a message channel provides information. This can be measured by the time delay between information generation and system retrieval. Consistency refers to the degree to which the information provided by a message channel conforms to information from other sources or historical information. This can be measured by the degree to which the information content matches known facts or multi-source cross-validation results. Performance results refer to the data set obtained by monitoring the success rate, timeliness, and consistency of a message channel during task execution. This can be implemented using structured logging, performance statistics, or real-time data stream monitoring. Evaluating and updating reliability indicators refers to calculating or adjusting the reliability indicator value or level corresponding to a message channel based on the monitored performance results. This can be implemented using methods such as rule-based adjustments, statistical analysis, or machine learning models.

[0080] The solution of this application presets reliability indicators for each message channel, providing an initial weight or priority for subsequent information disambiguation. During task execution, the system continuously monitors the actual performance of information provided by each message channel, including the success rate, timeliness, and consistency of information acquisition. This monitoring data reflects the reliability of the message channel in real-world operating environments. Based on these performance results, the system evaluates and adjusts the initially preset reliability indicators. For example, if a message channel performs well over a period of time, its reliability indicator can be increased; conversely, if it performs poorly, its reliability indicator can be lowered. This dynamic update mechanism enables the reliability indicator to reflect the actual reliability of the message channel in real time. When multiple message channels provide conflicting information about issues with heterogeneous data sources, the system can use the updated, more accurate reliability indicators for disambiguation. Information provided by the more reliable message channel is assigned a higher weight or priority, allowing for a more accurate assessment of the true status of the issue. This dynamic adjustment of reliability indicators and their application in disambiguation improves the accuracy of problem status assessment, thereby making task schedule adjustments and warning information output based on this status information more reliable. Compared with relying solely on the initial preset identification, this dynamic mechanism can adapt to changes in the reliability of the message channel, improving the robustness and accuracy of the entire method.

[0081] In some preferred embodiments, the implementation is as follows: There are three messaging channels: API, email notification, and manual data entry. The initial preset reliability ratings are: 0.9 for the API, 0.7 for email notification, and 0.5 for manual data entry. During task execution, the system continuously monitors the performance of these three channels. For example, the system monitors the API call log to record the success and response time of each call; monitors the email server reception log to record the arrival time of emails; and monitors the operation log of the manual data entry system to record the timeliness of the entered data and subsequent verification results. Suppose, after a period of monitoring, it is found that the API success rate has dropped to 80%, the average response time has increased, the arrival delay of email notifications has increased, and the consistency of manually entered data is relatively high. Based on these performance results, the system evaluates and updates the reliability ratings. For example, a comprehensive score can be calculated based on metrics such as success rate, timeliness, and consistency, and the reliability ratings can be adjusted proportionally. After the adjustments, the reliability ratings of the API might drop to 0.8, email notifications to 0.6, and manual data entry to 0.6. These updated reliability ratings will be used to resolve conflicting information.

[0082] Reference Figure 9 Furthermore, in another embodiment of the present application, step S3213 includes: S32131: Get the preset performance indicators of the performance results; S32132: Compare the performance results with the preset performance indicators, calculate the comprehensive score or reliability level of the message channel, and update the reliability indicator.

[0083] Among them, in this embodiment, the preset performance indicators of the performance results refer to the standards or thresholds used to measure the performance of the message channel in providing information, which can be set in the form of minimum requirements for success rate, maximum delay time for timeliness, minimum compliance ratio for consistency, etc.; the performance results refer to the actual data or statistical values of the success rate, timeliness and consistency of the information provided by the message channel continuously monitored during the execution of the task; comparing the performance results with the preset performance indicators refers to comparing and analyzing the actual performance data of the message channel with the preset performance standards; calculating the comprehensive score or reliability level of the message channel refers to obtaining a quantitative value or classification mark based on the comparison results through certain calculation methods or rules to represent the overall reliability of the message channel, which can be achieved by weighted average calculation, fuzzy logic evaluation, rule-based grading, etc.; updating the reliability mark refers to modifying the reliability status mark currently associated with the message channel according to the calculated comprehensive score or reliability level.

[0084] The solution of this application introduces preset performance indicators to provide an objective, quantitative benchmark for evaluating the performance of message channels. After obtaining the actual performance results of a message channel, these results are compared with the preset performance indicators. Based on the comparison results, a comprehensive score or reliability grade is calculated. This score or grade comprehensively and quantitatively evaluates the performance of the message channel across multiple dimensions, such as success rate, timeliness, and consistency. Finally, the reliability indicator of the message channel is dynamically updated based on the calculated comprehensive score or reliability grade. This process enables data-driven, quantitative evaluation and dynamic adjustment of the message channel's reliability. Accurately updated reliability indicators more accurately reflect the current status of the message channel. When conflicting information exists across multiple message channels, a predefined conflict resolution strategy is applied based on the obtained timeliness and reliability indicators corresponding to the message channels to resolve the conflicting information, generating a unified assessment basis for the status of the problem across heterogeneous data sources. Based on this status information, the expected duration or impact of the problem is assessed to obtain a problem assessment result. The problem assessment result is compared with the activated risk buffer time. If there is a discrepancy, the risk buffer time is adjusted again to update the corresponding task schedule. After the task's planned duration is updated, the critical path is recalculated and an early warning message indicating changes in task progress is output. This entire process forms a closed feedback loop, enabling project progress management to more accurately respond to changes in data source status.

[0085] In some preferred embodiments, the performance indicators for message channel A can be preset as follows: a success rate of no less than 95%, a timeliness delay of no more than 100 milliseconds, and a consistency of no less than 98%. During task execution, the performance of message channel A over the past period is continuously monitored, with the following results: an average success rate of 92%, an average timeliness delay of 120 milliseconds, and an average consistency of 99%. The performance results are compared with the preset performance indicators. For example, a success rate of 92% is lower than 95%, a timeliness delay of 120 milliseconds is higher than 100 milliseconds, and a consistency of 99% is higher than 98%. A weighted average method can be used to calculate the overall score, for example, setting a weight of 0.4 for the success rate, 0.3 for the timeliness, and 0.3 for the consistency. When calculating the score, the actual value can be normalized or penalized with the indicator value. For example, if the success rate falls below the indicator, points will be deducted, and if the timeliness exceeds the indicator, points will be deducted. The calculated overall score for message channel A is 75 points. According to the preset reliability grading rules, for example, a score of 80 or above is considered high reliability, a score of 60-80 is considered medium reliability, and a score below 60 is considered low reliability. The reliability indicator of message channel A is updated to "medium reliability".

[0086] Reference Figure 10 Furthermore, another embodiment of the present application proposes a supply chain demand forecasting model construction system that integrates multi-source heterogeneous data, including: The risk buffer configuration module is used to preset the risk buffer time for tasks that rely on heterogeneous data sources. The risk buffer time is associated with the risk attributes of the heterogeneous data sources, and the activation conditions based on the operating status of the heterogeneous data sources are set. The activation conditions are related to the availability, data quality, and update frequency of the heterogeneous data sources. A status monitoring module is used to obtain real-time information on the operating status of heterogeneous data sources, including data transmission status, interface response status and / or data accuracy; A plan adjustment module is used to adjust the activated risk buffer time associated with the task to update the task schedule when the instant information meets the activation conditions of the corresponding task; The calculation and warning module is used to recalculate the critical path of the task after the task plan duration is updated, and output warning information indicating the change in task progress.

[0087] Among them, in this embodiment, the risk buffer configuration module refers to a module for pre-setting additional time reserves for tasks that rely on heterogeneous data sources. The time reserves are associated with the risk attributes of the heterogeneous data sources, and trigger conditions based on the operating status of the heterogeneous data sources are set. The trigger conditions are related to the availability, data quality, and update frequency of the heterogeneous data sources. It can be implemented by using the functional modules responsible for configuration management and rule setting in the software system. Its purpose is to provide response time for potential data source problems and reduce the risk of task delays; the status monitoring module refers to a module for obtaining real-time information on the operating status of heterogeneous data sources. The real-time information includes data transmission status, interface response status and / or data accuracy. It can be implemented by using the functional modules responsible for data source connection management and rule setting in the software system. The function module of status collection is used to timely grasp the operation status of the data source; the plan adjustment module is used to adjust the activated risk buffer time associated with the task according to the preset logic when the instant information meets the activation conditions of the corresponding task, so as to update the planned duration of the task. It can be implemented by the function module responsible for project progress management and plan update in the software system, and its purpose is to dynamically adjust the task progress according to the data source status; the calculation and warning module is used to recalculate the task critical path after the task planned duration is updated, and output warning information indicating the change of task progress. It can be implemented by the function module responsible for project critical path analysis and information notification in the software system, and its purpose is to timely identify progress risks and issue warnings to users.

[0088] Finally, it should be noted that although the above embodiments have been described in the specification and drawings of this application, this does not limit the scope of patent protection of this application. All technical solutions generated by replacing or modifying equivalent structures or equivalent processes based on the essential concepts of this application using the contents recorded in the specification and drawings of this application, as well as directly or indirectly implementing the technical solutions of the above embodiments in other related technical fields, are included in the scope of patent protection of this application.

Claims

1. A method for constructing a supply chain demand forecasting model that integrates multi-source heterogeneous data, characterized in that: include: Preset a risk buffer time for tasks that rely on heterogeneous data sources. The risk buffer time is associated with the risk attributes of the heterogeneous data sources. Set activation conditions based on the operating status of the heterogeneous data sources. The activation conditions are related to the availability, data quality, and update frequency of the heterogeneous data sources. Acquiring real-time information on the operating status of the heterogeneous data source, the real-time information including data transmission status, interface response status and / or data accuracy; When the instant information meets the activation condition of the corresponding task, adjusting the activated risk buffer time associated with the task to update the planned duration of the task; After the planned duration of the task is updated, the critical path of the task is recalculated, and warning information indicating changes in the task progress is output.

2. The method for constructing a supply chain demand forecasting model integrating multi-source heterogeneous data according to claim 1 is characterized in that: The risk buffer time for tasks that rely on heterogeneous data sources is preset, the risk buffer time is associated with the risk attribute of the heterogeneous data source, and activation conditions are set based on the operating status of the heterogeneous data source, including: When a new heterogeneous data source is introduced, a phased trial process is performed on the heterogeneous data source, and operational performance data and technical interaction information are recorded, wherein the phased trial process sets corresponding trial conditions in multiple stages; forming an initial risk assessment result based on the operational performance data and the technical interaction information, wherein the risk assessment result includes an initial risk attribute of the new heterogeneous data source; The corresponding risk buffer time is configured according to the initial risk assessment result, and the activation condition based on the operating status of the heterogeneous data source is set.

3. The method for constructing a supply chain demand forecasting model integrating multi-source heterogeneous data according to claim 2 is characterized in that: The method further comprises: During task execution, continuously monitor changes in the initial assessment basic information related to the heterogeneous data sources, wherein the initial assessment basic information includes data interface documents, data quality history reports, and / or supplier service commitments; The risk attribute is updated according to the change information, and the risk buffer time that has not been activated is adjusted according to the updated risk attribute.

4. The method for constructing a supply chain demand forecasting model integrating multi-source heterogeneous data according to claim 1 is characterized in that: When the instant information satisfies the activation condition of the corresponding task, adjusting the activated risk buffer time associated with the task to update the planned duration of the task includes: When the instant information satisfies the activation condition of the corresponding task, status information of the problem existing in the corresponding heterogeneous data source is obtained, the status information including the operation status of the data source, network problems and / or interface interruption; Based on the status information, assess the expected duration or impact of the problem to obtain a problem assessment result, wherein the problem assessment result includes the number of days of delay and / or the scope of impact; The problem assessment result is compared with the activated risk buffer time. If there is a deviation, the risk buffer time is adjusted again to update the corresponding task plan duration.

5. The method for constructing a supply chain demand forecasting model integrating multi-source heterogeneous data according to claim 4 is characterized in that: The status information is derived from multiple message channels, and the obtaining of the status information of the corresponding problem existing in the heterogeneous data source includes: Presetting a reliability identifier corresponding to each of the message channels, wherein the reliability identifier reflects the accuracy, timeliness and consistency of the corresponding information channel; When conflicting information exists in multiple message channels, a predefined conflict handling strategy is applied based on the timeliness of acquisition and the reliability identifier corresponding to the message channel to disambiguate the conflicting information, thereby forming a unified evaluation basis for the status information of the problem existing in the corresponding heterogeneous data sources.

6. The method for constructing a supply chain demand forecasting model integrating multi-source heterogeneous data according to claim 5 is characterized in that: The conflict handling strategies include: Preset conflict mode library, store multiple conflict modes, and configure disambiguation rule sets for each mode; When the conflict information conforms to a certain conflict pattern, the corresponding disambiguation rule set is called to perform disambiguation processing.

7. The method for constructing a supply chain demand forecasting model integrating multi-source heterogeneous data according to claim 6 is characterized in that: Also includes: Obtaining disambiguation results and feedback information of the disambiguation rule set in the corresponding disambiguation processing process, wherein the feedback information includes a disambiguation success rate and result accuracy; When the feedback information indicates that the disambiguation result does not meet the preset quality standard, analyzing the reasons for not meeting the standard; According to the unsatisfied reason, an existing rule in the disambiguation rule set is adjusted or a new rule in the disambiguation rule set is added.

8. The method for constructing a supply chain demand forecasting model integrating multi-source heterogeneous data according to claim 5 is characterized in that: The preset reliability identifiers corresponding to the respective message channels include: Presetting the reliability identifier of each of the message channels; During the execution of the task, continuously monitor the performance results of the success rate, timeliness and consistency of the information provided by the message channel corresponding to the reliability identifier; The reliability indicator is evaluated and updated according to the performance result.

9. The method for constructing a supply chain demand forecasting model integrating multi-source heterogeneous data according to claim 8 is characterized in that: The evaluating and updating the reliability indicator according to the performance result includes: Obtaining a preset performance indicator of the performance result; Compare the performance result with the preset performance indicator, calculate the comprehensive score or reliability level of the message channel, and update the reliability indicator.

10. A supply chain demand forecasting model construction system integrating multi-source heterogeneous data, characterized by: include: A risk buffer configuration module is used to preset a risk buffer time for tasks that rely on heterogeneous data sources, wherein the risk buffer time is associated with the risk attributes of the heterogeneous data sources, and to set activation conditions based on the operating status of the heterogeneous data sources, wherein the activation conditions are related to the availability, data quality, and update frequency of the heterogeneous data sources; A status monitoring module, configured to obtain real-time information on the operating status of the heterogeneous data source, wherein the real-time information includes data transmission status, interface response status, and / or data accuracy; A plan adjustment module, configured to adjust the activated risk buffer time associated with the task to update the planned duration of the task when the instant information meets the activation condition of the corresponding task; The calculation and warning module is used to recalculate the critical path of the task after the planned duration of the task is updated, and output warning information indicating the change in task progress.

Images