A smart supplier expansion management method based on technology stack and big data

Abstract

The present invention relates to the field of supply chain management technology, and in particular to a smart supplier expansion management method based on technology stack and big data. It comprises the following steps: S1: obtaining the approval data and delivery data of historical suppliers, and defining them as first approval data and first delivery data; dividing the first approval data and the first delivery data into structured data and unstructured data according to the timeline; S2: extracting the number of audits and the audit interval of the audit department based on the division result of the first approval data; S3: based on the number of audits and the audit interval, obtaining the audit efficiency evaluation value of the enterprise by quantifying the repeatability and time delay in the audit process. The present invention integrates supplier approval and delivery data, improves parsing efficiency, quantifies audit bottlenecks, dynamically evaluates contract performance capabilities, optimizes management processes, realizes full-process automated control, enhances collaboration efficiency and scientific decision-making, and ensures efficient operation of the supply chain.

Description

Technical Field

[0001] The present invention relates to the technical field of supply chain management, and in particular to an intelligent supplier expansion management method based on technology stack and big data. Background Art

[0002] In recent years, with the increasing complexity of corporate supply chains, supplier expansion management has gradually become a core part of corporate operations. However, traditional management models generally rely on manual operations and decentralized data management, resulting in increasingly prominent problems such as a lack of core data support in the approval process, lengthy cross-departmental communication links, and difficulty in data traceability. In existing technologies, supplier access, grading, and performance management mostly use offline or semi-automated processes, which make it difficult to integrate and analyze multi-source heterogeneous data in real time, resulting in delayed decision-making and inefficiency. In addition, the lack of unstructured data processing capabilities makes it impossible to fully tap the potential value of historical data, further exacerbating management costs and risks. In response to the above pain points, there is an urgent need for a data-driven intelligent supplier management system that can run through the entire life cycle to achieve process automation, information transparency, and scientific decision-making. Summary of the Invention

[0003] In order to overcome the shortcomings of inefficient approval and difficult data tracing, the present invention provides an intelligent supplier expansion management method based on technology stack and big data.

[0004] The technical implementation scheme of the present invention is: a smart supplier expansion management method based on technology stack and big data, comprising the following steps:

[0005] S1: Acquire historical supplier approval data and delivery data, and define them as first approval data and first delivery data; divide the first approval data and the first delivery data into structured data and unstructured data according to the timeline;

[0006] S2: Based on the classification results of the first approval data, extract the number of audits and the audit interval of the audit department;

[0007] S3: Based on the number of audits and the audit interval, the enterprise's audit efficiency evaluation value is obtained by quantifying the repetitiveness and time delay in the audit process; based on the classification results of the first delivery data, the supplier's credibility is obtained by quantifying the stability and change frequency of the delivery data;

[0008] S4: Based on the audit efficiency evaluation value and the supplier credibility, the enterprise audit process is optimized and new suppliers are screened; based on the optimized enterprise audit process and the screened new suppliers, the approval data and delivery data of the new suppliers are extracted and defined as second approval data and second delivery data;

[0009] S5: Determine an approval management process for a new supplier based on the number of reviews and review intervals for the first approval data and the second approval data; and determine an extended management process for the new supplier based on the variance and change frequency of structured data and unstructured data between the first delivery data and the second delivery data.

[0010] The first approval data and the first delivery data are divided into structured data and unstructured data according to the timeline, including:

[0011] The structured data refers to standardized information with a unified format and capable of direct quantitative analysis;

[0012] The unstructured data refers to multimodal information that is heterogeneous in form and requires technical analysis;

[0013] Obtaining the review timeline of the first approval data, dividing the review data on the review timeline into structured data and unstructured data, and obtaining a division result of the review timeline;

[0014] A delivery timeline of the first delivery data is obtained, and the delivery data on the delivery timeline is divided into structured data and unstructured data to obtain a division result of the delivery timeline.

[0015] Preferably, the obtaining of the historical supplier's approval data and delivery data, which are defined as first approval data and first delivery data, includes:

[0016] The approval data refers to the static qualification information submitted by the supplier and independently reviewed by the enterprise;

[0017] The delivery data refers to the multi-dimensional information generated by the dynamic interaction between suppliers and enterprises during the order fulfillment process.

[0018] Preferably, the obtaining of the review timeline of the first approval data, dividing the review data on the review timeline into structured data and unstructured data, and obtaining the division result of the review timeline includes:

[0019] The audit timeline refers to the timeline from when the supplier submits the audit data to when the audit is completed;

[0020] On the audit timeline, obtain the audit department corresponding to the retrospective timeline, extract the structured data and unstructured data of the audit department, and obtain the division result of the audit timeline;

[0021] Backtracking timeline: refers to the abnormal path in the approval process where the process rolls back to a previous node due to failure of review or incomplete information.

[0022] Preferably, extracting the number of audits and audit intervals of the audit department based on the division result of the first approval data includes:

[0023] Record the number of times the audit department appears on the audit timeline, and use the start to end of each approval process as a sub-audit timeline to obtain several sub-audit timelines;

[0024] Based on the plurality of sub-audit timelines, the number of audits of the audit departments on the plurality of sub-audit timelines is extracted, and at the same time, the audit interval of the same audit department that appears the most times in adjacent sub-audit timelines is extracted.

[0025] Preferably, the acquiring of the delivery timeline of the first delivery data, dividing the delivery data on the delivery timeline into structured data and unstructured data, and obtaining a division result of the delivery timeline, includes:

[0026] The delivery timeline refers to the time from when the supplier starts delivering an order to when the delivery is completed;

[0027] Obtain N delivery timelines, extract delivery data of the same order on the N delivery timelines, divide the delivery data into structured data and unstructured data, record the changes in structured data and unstructured data of the same order, and obtain the division results of the delivery timelines.

[0028] Preferably, the audit efficiency evaluation value of the enterprise is obtained by quantifying the repetitiveness and time delay in the audit process based on the audit times and audit intervals, including: the approval efficiency evaluation formula is as follows:

[0029]

[0030] in, To review the efficiency evaluation value, For the The maximum number of repetitions of the same audit department in each audit timeline, For the The average review interval between each sub-review timeline and the previous process for the same department, is the industry benchmark interval time constant, The total number of sub-review timelines.

[0031] Preferably, the supplier credibility is obtained by quantifying the stability and change frequency of the delivery data based on the division result of the first delivery data, including: the supplier credibility formula is as follows,

[0032] in, is the supplier credibility value, ∈(0,1], is the variance of structured data for the same order in N deliveries, is the change frequency of unstructured data in N deliveries of the same order, 、 is the penalty coefficient.

[0033] Preferably, based on the audit efficiency evaluation value and the supplier credibility, the enterprise audit process is optimized and new suppliers are screened; based on the optimized enterprise audit process and the screened new suppliers, the approval data and delivery data of the new suppliers are extracted and defined as second approval data and second delivery data, including:

[0034] If the audit efficiency evaluation value is greater than or equal to the preset audit efficiency evaluation threshold, the duplicate audit departments are merged;

[0035] If the audit efficiency evaluation value is less than the preset audit efficiency evaluation threshold, the original audit process is maintained;

[0036] If the supplier credibility is greater than or equal to a preset supplier credibility threshold, the supplier is determined to be an extended supplier; based on the extended supplier, the approval data and delivery data of the extended supplier are extracted and defined as second approval data and second delivery data;

[0037] If the supplier credibility is less than the preset supplier credibility threshold, the supplier is re-screened.

[0038] Preferably, determining the approval management process for a new supplier based on the number of reviews and review intervals of the first approval data and the second approval data; and determining the extended management process for a new supplier based on the variance and change frequency of structured data and unstructured data of the first delivery data and the second delivery data include:

[0039] The approval management process for new suppliers is determined through the approval decision formula. The approval decision formula is as follows:

[0040] in, is the approval decision value, For the second approval data The maximum number of repeated audits for the same department in a sub-audit timeline, is the average review interval of the second approval data, The total number of sub-review timelines for the second approval data. For the first approval data The maximum number of repeated audits for the same department in a sub-audit timeline, is the average review interval of the first approval data, The total number of sub-review timelines for the first approval data. is the regulating factor;

[0041] The expansion management process for new suppliers is determined through the expansion management formula. The expansion management formula is as follows:

[0042] in, To extend the management value, The variance of the structured data for the second delivery data, The change frequency of unstructured data for the second delivery data, The variance of the structured data for the first delivery data, The frequency of changes to unstructured data for first-time delivery.

[0043] Beneficial effects: The present invention integrates the approval data and delivery data of historical suppliers to build a data management system for the entire life cycle, providing accurate data support for supplier expansion management; by dividing the approval data and delivery data into structured data and unstructured data according to the timeline, it significantly improves the parsing efficiency and integration capabilities of multi-source heterogeneous data, and effectively solves the problems of data islands and information fragmentation in the traditional model; the audit efficiency evaluation value calculation based on the number of audits and the audit interval time can quantify the efficiency bottleneck of the enterprise audit process, and provide a scientific basis for optimizing the configuration of the audit department and reducing redundant links; at the same time, the supplier credibility formula By analyzing the structured variance and unstructured change frequency of delivery data for the same order, suppliers' performance capabilities can be dynamically assessed, enhancing the objectivity and reliability of new supplier screening. By comparing the approval management process and delivery data variance ratio of new and old suppliers, the expansion management process can be dynamically adjusted and optimized to ensure the flexibility and adaptability of supplier expansion decisions. Finally, based on the deep integration of technology stacks and big data, the solution implements automated control of the entire process of supplier access, grading, and performance, significantly improving cross-departmental collaboration efficiency and the scientific nature of business decisions, and providing systematic technical support for cost reduction, efficiency improvement, and risk prevention and control in supply chain management. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a flow chart of the smart supplier expansion management method based on technology stack and big data of the present invention;

[0045] Figure 2 A flowchart of the process of dividing the review timeline for the present invention;

[0046] Figure 3 A flowchart illustrating the process of dividing the delivery timeline of the present invention. DETAILED DESCRIPTION

[0047] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0048] A smart supplier expansion management method based on technology stack and big data, such as Figure 1 As shown, the following steps are included:

[0049] S1: Acquire historical supplier approval data and delivery data, and define them as first approval data and first delivery data; divide the first approval data and the first delivery data into structured data and unstructured data according to the timeline;

[0050] S2: Based on the classification results of the first approval data, extract the number of audits and the audit interval of the audit department;

[0051] S3: Based on the number of audits and the audit interval, the enterprise's audit efficiency evaluation value is obtained by quantifying the repetitiveness and time delay in the audit process; based on the classification results of the first delivery data, the supplier's credibility is obtained by quantifying the stability and change frequency of the delivery data;

[0052] S4: Based on the audit efficiency evaluation value and the supplier credibility, the enterprise audit process is optimized and new suppliers are screened; based on the optimized enterprise audit process and the screened new suppliers, the approval data and delivery data of the new suppliers are extracted and defined as second approval data and second delivery data;

[0053] S5: Determine an approval management process for a new supplier based on the number of reviews and review intervals for the first approval data and the second approval data; and determine an extended management process for the new supplier based on the variance and change frequency of structured data and unstructured data between the first delivery data and the second delivery data.

[0054] The first approval data and the first delivery data are divided into structured data and unstructured data according to the timeline, including:

[0055] The structured data refers to standardized information with a unified format and capable of direct quantitative analysis;

[0056] The unstructured data refers to multimodal information that is heterogeneous in form and requires technical analysis;

[0057] Obtaining the review timeline of the first approval data, dividing the review data on the review timeline into structured data and unstructured data, and obtaining a division result of the review timeline;

[0058] A delivery timeline of the first delivery data is obtained, and the delivery data on the delivery timeline is divided into structured data and unstructured data to obtain a division result of the delivery timeline.

[0059] It's important to note that structured and unstructured data differ significantly in their analytical dimensions and technical processing. Structured data facilitates direct quantitative analysis and model training; unstructured data, because it contains multimodal information like semantics and images, requires analysis through natural language processing (NLP) or computer vision (CV). This segmentation allows for targeted optimization of data processing efficiency and depth.

[0060] Audit timeline: records the complete cycle from supplier qualification submission to audit completion, used to identify audit efficiency bottlenecks (such as repeated approval steps);

[0061] Delivery timeline: Tracks the entire process of an order from initiation to fulfillment, used to analyze delivery stability (such as delay frequency and abnormal fluctuations).

[0062] Based on the dynamic division of timelines, data is deeply bound to business scenarios. For example, in the audit timeline, structured data (such as audit time and department turnover times) is used to quantify efficiency, and unstructured data (such as audit opinion text and scanned attachments) is used to extract risk keywords through semantic analysis. In the delivery timeline, structured data (such as delivery punctuality rate and quality inspection pass rate) is directly involved in credibility calculation, and unstructured data (such as logistics communication records and abnormal feedback emails) is used to assist decision-making through sentiment analysis.

[0063] For example, in a supplier's review timeline, structured data includes "qualification review took 3 days and was transferred through 2 departments," and unstructured data includes the environmental certification PDF file submitted by the supplier (requiring OCR parsing). In the delivery timeline, structured data includes "order A delivery delay rate 5%," and unstructured data includes the supplier's logistics anomaly explanation email (requiring NLP to extract key reasons). Combining these two data allows for an accurate assessment of the supplier's compliance risk and performance capabilities, providing a multi-dimensional basis for expansion decisions.

[0064] Obtain historical supplier approval and delivery data, and define them as the first approval and first delivery data, including:

[0065] The approval data refers to the static qualification information submitted by the supplier and independently reviewed by the enterprise;

[0066] The delivery data refers to the multi-dimensional information generated by the dynamic interaction between suppliers and enterprises during the order fulfillment process.

[0067] It should be noted that traditional supplier management relies on manual review and scattered data, making it difficult to integrate multi-source information, resulting in delayed decision-making and inefficiency. By defining approval and delivery data in a structured manner, a standardized analysis foundation is established to support full life cycle management.

[0068] Approval data: reflects the supplier's qualification compliance and requires quantification of audit efficiency and risks;

[0069] Delivery data: reflects the supplier's ability to fulfill its obligations, and requires dynamic evaluation of quality and stability.

[0070] Approval data: Extract static qualification documents (such as business licenses and certification certificates) from the enterprise ERP and SRM systems, and automatically collect them through API interfaces or direct database connections;

[0071] Delivery data: Through the order management system (OMS) and Internet of Things (IoT) devices, interaction records during order fulfillment (such as delivery time, quality inspection reports, and logistics tracks) are captured in real time, and natural language processing (NLP) is used to parse unstructured communication logs.

[0072] For example: When a supplier submits ISO certification (approval data), the data is automatically archived and marked with the review time. During a supplier's delivery process, the delivery delay rate of a batch of products (structured data) and after-sales communication emails (unstructured data) are recorded in real time. After integration, a supplier performance profile is generated to provide a basis for subsequent expansion decisions.

[0073] By centrally storing multi-source heterogeneous data (such as qualification documents and delivery logs) in the Hadoop data lake, the Spark engine is used to perform batch analysis on structured data (review time, delivery timeliness), and natural language processing (NLP) and machine learning models (such as LSTM) are used to parse unstructured communication records in real time, extract risk features (such as high-frequency rejection keywords), and dynamically optimize review rules and supplier profiles.

[0074] Obtain the review timeline of the first approval data, divide the review data on the review timeline into structured data and unstructured data, and obtain the division result of the review timeline, including:

[0075] The audit timeline refers to the timeline from when the supplier submits the audit data to when the audit is completed;

[0076] On the audit timeline, obtain the audit department corresponding to the retrospective timeline, extract the structured data and unstructured data of the audit department, and obtain the division result of the audit timeline;

[0077] Backtracking timeline: refers to the abnormal path in the approval process where the process rolls back to a previous node due to failure of review or incomplete information.

[0078] It should be noted that if Figure 2 As shown, the backtracking timeline refers to an abnormal path in the approval process where the process rolls back to the previous node due to failure in the review or incomplete information. For example, the normal process should be "submit → preliminary review → final review", but in reality, due to rejection of the preliminary review, the "submit → preliminary review → submit → preliminary review → final review" process is repeated, forming a redundant link.

[0079] By tracing such back-off paths, we can extract structured data (such as the number of back-offs and time consumption) and unstructured data (such as the text of the rejection reason) of repeated nodes, identify process bottlenecks (such as frequently rejected departments), and provide a basis for merging or optimizing redundant steps.

[0080] For example, during a supplier audit, the initial review was rejected twice due to "incomplete environmental protection qualifications." The retrospective timeline showed "submission → initial review (rejection) → submission → initial review (pass) → final review." Structured data statistics showed that the initial review took three days to complete, and unstructured data analysis of the rejection opinion PDF revealed a missing ISO certificate. After optimizing the pre-review rules based on this, the process was shortened to "submission → final review," creating a synergistic effect with the aforementioned delivery data optimization.

[0081] Based on the division results of the first approval data, the number of audits and the audit interval of the audit department are extracted, including:

[0082] Record the number of times the audit department appears on the audit timeline, and use the start to end of each approval process as a sub-audit timeline to obtain several sub-audit timelines;

[0083] Based on the plurality of sub-audit timelines, the number of audits of the audit departments on the plurality of sub-audit timelines is extracted, and at the same time, the audit interval of the same audit department that appears the most times in adjacent sub-audit timelines is extracted.

[0084] It should be noted that the split review timeline refers to two independent review cycles in chronological order (for example, the first process is "submission → preliminary review → rejection", and the second process is "submission → preliminary review → final review").

[0085] Audit frequency extraction: Count the total number of times each audit department participates in all sub-audit timelines (for example, the initial audit department appears 3 times in 5 processes).

[0086] Audit interval time extraction:

[0087] High-frequency department screening: For each pair of adjacent sub-review timelines, count the number of times each department appears (for example, the preliminary review department appears twice in adjacent processes) and select the department with the highest number of appearances.

[0088] Interval time calculation: Extract the last review timestamp of the selected department in adjacent processes and calculate the time difference (for example, the initial review time of the first process is Day 1, the initial review time of the second process is Day 3, and the interval time is 2 days);

[0089] Tie handling: If the number of times a department appears is the same, the department that appears first in the process will be given priority.

[0090] Example: A supplier went through two separate review timelines due to incomplete qualifications. After merging the preliminary and final review rules, the number of reviews was reduced from 2 to 1, and the interval was reset to zero.

[0091] Obtain a delivery timeline of the first delivery data, divide the delivery data on the delivery timeline into structured data and unstructured data, and obtain a delivery timeline division result, including:

[0092] The delivery timeline refers to the time from when the supplier starts delivering an order to when the delivery is completed;

[0093] Obtain N delivery timelines, extract delivery data of the same order on the N delivery timelines, divide the delivery data into structured data and unstructured data, record the changes in structured data and unstructured data of the same order, and obtain the division results of the delivery timelines.

[0094] It should be noted that if Figure 3 As shown, the division of the delivery timeline aims to quantify the stability and abnormal fluctuations of the supplier's fulfillment behavior by tracking the entire cycle of the order from initiation to completion; structured data (such as delivery time, quality inspection pass rate) can directly participate in the model calculation, while unstructured data (such as logistics communication records, abnormal feedback text) needs to use NLP or OCR to analyze semantic and image information. The combination of the two can comprehensively evaluate delivery quality and potential risks.

[0095] For example, in the delivery timeline for Order A from a supplier, structured data indicated a "delivery delay rate of 5%." Unstructured data included an exception email from the logistics company (with the keyword "weather" parsed). Combined with the initial review rules from the previous example, the supplier's lax pre-qualification process led to frequent delivery anomalies. By strengthening the pre-qualification process, the supplier's subsequent order delay rate dropped to 2%, achieving closed-loop optimization of review and delivery data.

[0096] Based on the number of audits and the audit interval, the audit efficiency evaluation value of the enterprise is obtained by quantifying the repetitiveness and time delay in the audit process, including: the approval efficiency evaluation formula is as follows,

[0097] in, To review the efficiency evaluation value, For the The maximum number of repetitions of the same audit department in each audit timeline, For the The average review interval between each sub-review timeline and the previous process for the same department, is the industry benchmark interval time constant, The total number of sub-review timelines.

[0098] It should be noted that the approval efficiency evaluation formula reflects the enterprise's audit efficiency by quantifying the duplication and time delay in the audit process; : No. The maximum number of repeated reviews by the same department in a sub-process (e.g. the initial review is triggered twice); : The average interval time of the department in adjacent sub-processes (e.g., 1 day between two preliminary reviews); : Industry benchmark interval (e.g. the industry standard is 1 day); : Total number of sub-processes.

[0099] Number of repetitions ( ) is higher, the efficiency is lower; the interval time ( ) is longer, the greater the efficiency loss; the natural logarithm ln is used to weaken the influence of extreme values ​​and ensure calculation stability.

[0100] Example: Assume that among the three sub-processes of a company, Department A 2, 1, and 3 times respectively. 2, 1, 3 days, = 1 day; calculated: =2ln(3)+1ln(2)+3ln(4)≈7.05; if the industry efficiency threshold is <5, the current value of 7.05 indicates that optimization is needed: reduce repeated review: merge the preliminary review and final review rules, Reduce to 1; and shorten the interval time: through automated approval, near .

[0101] Based on the division results of the first delivery data, the supplier credibility is obtained by quantifying the stability and change frequency of the delivery data, including: the supplier credibility formula is as follows:

[0102] in, is the supplier credibility value, ∈(0,1], is the variance of structured data for the same order in N deliveries, is the change frequency of unstructured data in N deliveries of the same order, 、 is the penalty coefficient.

[0103] It should be noted that the supplier credibility formula evaluates the supplier's performance reliability by quantifying the stability and change frequency of delivery data. (Structured data variance): Acquisition method: For N deliveries of the same order, extract structured indicators (such as delivery time, quality inspection pass rate), and calculate the variance. For example, the delivery time of an order is 10 days, 12 days, and 11 days respectively, and the average is 11 days. The variance is = . (Unstructured data change frequency): Acquisition method: Count the number of modifications or content changes per unit time of unstructured data (such as logistics communication records, abnormality explanation emails) during the same order delivery process. For example, if the logistics record is changed twice due to "weather reasons" and "equipment failure", then =2; 、 : Penalty coefficient, set according to business needs (such as =0.5, =0.3), which is used to adjust the influence weight of the two types of data on credibility; (variance of structured data) and (Unstructured data change frequency) is standardized (such as normalization or z-score) and penalized by the penalty coefficient 、 Balanced impact.

[0104] Example calculation: A supplier completes 3 deliveries. =0.67, =2, substitute into the formula: =1 / 1.935≈0.517, credibility ∈(0,1], the closer the value is to 1, the more reliable it is; in this case, 0.517 indicates that the delivery stability needs to be optimized (reducing ) and reduce changes (reduce ).

[0105] It should be noted that in order to achieve dynamic optimization of the above review process (such as merging redundant review departments), the Flowable workflow engine is used to build an automated approval process based on the BPMN2.0 standard:

[0106] Visual modeling: Map qualification review and cross-departmental flow nodes into configurable processes (such as "preliminary review → final review"), supporting drag-and-drop adjustments (such as merging preliminary review and final review nodes);

[0107] Automated event triggering: Process nodes automatically execute operations (such as notifying the final review department after passing the preliminary review and sending email reminders when rejected), reducing manual intervention;

[0108] Full-link traceability: Through distributed transaction management (Seata) and event tracing, logs such as rejection reasons and review time are recorded, supporting process card point analysis (such as high-frequency rejection departments);

[0109] Quality verification and monitoring: embed a rule engine in the node (such as qualification integrity check), and monitor node time consumption in real time through Prometheus, triggering warnings in case of abnormalities (such as audit timeout).

[0110] Based on the audit efficiency evaluation value and supplier credibility, the enterprise audit process is optimized and new suppliers are screened. Based on the optimized enterprise audit process and the screened new suppliers, the approval data and delivery data of the new suppliers are extracted and defined as second approval data and second delivery data, including:

[0111] If the audit efficiency evaluation value is greater than or equal to the preset audit efficiency evaluation threshold, the duplicate audit departments are merged;

[0112] If the audit efficiency evaluation value is less than the preset audit efficiency evaluation threshold, the original audit process is maintained;

[0113] If the supplier credibility is greater than or equal to a preset supplier credibility threshold, the supplier is determined to be an extended supplier; based on the extended supplier, the approval data and delivery data of the extended supplier are extracted and defined as second approval data and second delivery data;

[0114] If the supplier credibility is less than the preset supplier credibility threshold, the supplier is re-screened.

[0115] It should be noted that the threshold setting method, audit efficiency assessment threshold: based on historical data or industry benchmarks, by statistically analyzing the average efficiency value of the company's past efficient audit processes (such as =5), set the threshold (such as 20% redundancy) based on the business tolerance (such as =6). Supplier credibility threshold: Based on the lowest reliability level of historical suppliers (e.g. ≥0.7), or it can be derived by inferring the characteristics of high-quality suppliers through machine learning models (such as 0.65 as the dividing line).

[0116] Dynamically trigger process optimization and supplier screening through thresholds; when the audit efficiency value exceeds the standard, merge duplicate departments to shorten the cycle; when the credibility meets the standard, lock in high-quality suppliers and collect high-quality supplier data (second data) to form a closed-loop feedback.

[0117] Example: A company sets =6, the audit efficiency value of a supplier =7.05 (previous calculation value), triggering the merger of the preliminary review and final review departments. After optimization, dropped to 4.2; at the same time, the supplier's credibility =0.517 (lower than the threshold of 0.65), automatically excluded and re-screened; the credibility of the new supplier B =0.82, and was included in the extended supplier. The approval and delivery data of the new supplier B was defined as the second data for subsequent process iterations.

[0118] Determining an approval management process for a new supplier based on the number of reviews and review intervals of the first approval data and the second approval data; and determining an extended management process for a new supplier based on the variance and change frequency of structured data and unstructured data of the first delivery data and the second delivery data, including:

[0119] The approval management process for new suppliers is determined through the approval decision formula. The approval decision formula is as follows:

[0120] in, is the approval decision value, For the second approval data The maximum number of repeated audits for the same department in a sub-audit timeline, is the average review interval of the second approval data, The total number of sub-review timelines for the second approval data. For the first approval data The maximum number of repeated audits for the same department in a sub-audit timeline, is the average review interval of the first approval data, The total number of sub-review timelines for the first approval data. is the regulating factor;

[0121] It should be noted that the approval decision formula quantifies the degree of optimization of the new process by comparing the audit efficiency differences between new and old suppliers. : The ratio of the number of repeated audits of new and old suppliers. The smaller the ratio, the less redundancy there is in the new process. : The ratio of the average review interval between the new and old processes. The smaller the ratio, the faster the new process. : Adjustment factor, used to control decision sensitivity (such as =0.1 focuses on efficiency, = 0.2 focuses on stability).

[0122] Decision logic: →1 (close to 1): indicates that the new process has less repeated review and shorter time, and can be adopted; →0 (close to 0): Indicates that the efficiency of the new process has not been improved and needs to be adjusted.

[0123] Example: Old supplier data: =10, =5 days; New supplier data: =8, =4 days; take = 0.1, we get: ≈0.852, because >0.7 (preset threshold), the new process is judged to be efficient and is determined as the standard approval management process.

[0124] The expansion management process for new suppliers is determined through the expansion management formula. The expansion management formula is as follows:

[0125] in, To extend the management value, The variance of the structured data for the second delivery data, The change frequency of unstructured data for the second delivery data, The variance of the structured data for the first delivery data, The frequency of changes to unstructured data for first-time delivery.

[0126] It should be noted that the extended management formula quantifies the degree of improvement in the reliability of the new supplier by comparing the delivery stability differences between the old and new suppliers. 、 : The structured variance and unstructured change frequency of the old supplier, representing the historical delivery volatility; 、 : The corresponding value of the new supplier, reflecting the new supplier's performance stability; : Compress the fluctuation ratio difference, avoid the influence of extreme values, and ensure the robustness of the results; and Greater than or equal to 0; in the extended management formula, 、 、 and The maximum and minimum values ​​have been normalized to ensure that the value range is [0,1] to eliminate the impact of dimensional differences on volatility comparison.

[0127] Decision logic: The larger the value, the closer the fluctuation ratio between the new and old suppliers is to 0, indicating that the stability of the new supplier is significantly better than the historical level and cooperation can be expanded. The smaller the value, the greater the new supplier's fluctuation ratio is than the preset fluctuation ratio threshold, indicating that the new supplier has not met expectations and the management process needs to be adjusted.

[0128] Example: Old supplier data: =2.5, =3; New supplier data: =1.2, =1; calculated: =1 / ln(1.4)≈2.47, if the threshold is set to If the stability of the new supplier is >2, the new supplier is judged to have met the stability standards, designated as an extended supplier, and an optimized management process is adopted (such as shortening the delivery cycle).

[0129] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A smart supplier expansion management method based on technology stack and big data, characterized by: The following steps are involved: S1: Obtain the historical supplier's approval data and delivery data, and define them as the first approval data and the first delivery data; dividing the first approval data and the first delivery data into structured data and unstructured data according to a timeline; S2: Based on the classification results of the first approval data, extract the number of audits and the audit interval of the audit department; S3: Based on the number of audits and the audit intervals, obtain an audit efficiency assessment value for the enterprise by quantifying the repetitiveness and time delay in the audit process; Based on the classification results of the first delivery data, the supplier credibility is obtained by quantifying the stability and change frequency of the delivery data; S4: Based on the audit efficiency evaluation value and supplier credibility, optimize the enterprise audit process and screen new suppliers; Based on the optimized enterprise review process and screened new suppliers, the approval data and delivery data of new suppliers are extracted and defined as the second approval data and second delivery data; S5: Determine the new supplier approval management process based on the number of reviews and review intervals of the first approval data and the second approval data; determining an expansion management process for a new supplier based on variances and change frequencies of structured data and unstructured data between the first delivery data and the second delivery data; The first approval data and the first delivery data are divided into structured data and unstructured data according to the timeline, including: The structured data refers to standardized information with a unified format and capable of direct quantitative analysis; The unstructured data refers to multimodal information that is heterogeneous in form and requires technical analysis; Obtaining the review timeline of the first approval data, dividing the review data on the review timeline into structured data and unstructured data, and obtaining a division result of the review timeline; A delivery timeline of the first delivery data is obtained, and the delivery data on the delivery timeline is divided into structured data and unstructured data to obtain a division result of the delivery timeline.

2. The intelligent supplier expansion management method based on technology stack and big data according to claim 1 is characterized in that: The acquisition of the historical supplier's approval data and delivery data, which are defined as the first approval data and the first delivery data, includes: The approval data refers to the static qualification information submitted by the supplier and independently reviewed by the enterprise; The delivery data refers to the multi-dimensional information generated by the dynamic interaction between suppliers and enterprises during the order fulfillment process.

3. The intelligent supplier expansion management method based on technology stack and big data according to claim 1 is characterized in that: The step of obtaining the review timeline of the first approval data and dividing the review data on the review timeline into structured data and unstructured data to obtain a division result of the review timeline includes: The audit timeline refers to the time from when the supplier submits the audit data to when the audit is completed; On the audit timeline, obtain the audit department corresponding to the retrospective timeline, extract the structured data and unstructured data of the audit department, and obtain the division result of the audit timeline; Backtracking timeline: refers to the abnormal path in the approval process where the process rolls back to a previous node due to failure of review or incomplete information.

4. The method for intelligent supplier expansion management based on technology stack and big data according to claim 1 is characterized in that: The extracting of the number of audits and the audit interval of the audit department based on the division result of the first approval data includes: Record the number of times the audit department appears on the audit timeline, and use the start to end of each approval process as a sub-audit timeline to obtain several sub-audit timelines; Based on the plurality of sub-audit timelines, the number of audits of the audit departments on the plurality of sub-audit timelines is extracted, and at the same time, the audit interval of the same audit department that appears the most times in adjacent sub-audit timelines is extracted.

5. The intelligent supplier expansion management method based on technology stack and big data according to claim 1 is characterized in that: The obtaining of the delivery timeline of the first delivery data, dividing the delivery data on the delivery timeline into structured data and unstructured data, and obtaining a division result of the delivery timeline includes: The delivery timeline refers to the time from when the supplier starts delivering an order to when the delivery is completed; Obtain N delivery timelines, extract delivery data of the same order on the N delivery timelines, divide the delivery data into structured data and unstructured data, record the changes in structured data and unstructured data of the same order, and obtain the division results of the delivery timelines.

6. The method for intelligent supplier expansion management based on technology stack and big data according to claim 1 is characterized in that: Based on the number of audits and the audit interval, the audit efficiency evaluation value of the enterprise is obtained by quantifying the repetitiveness and time delay in the audit process, including: the approval efficiency evaluation formula is as follows: in, To review the efficiency evaluation value, For the The maximum number of repetitions of the same audit department in each audit timeline, For the The average review interval between each sub-review timeline and the previous process for the same department, is the industry benchmark interval time constant, The total number of sub-review timelines.

7. The method for intelligent supplier expansion management based on technology stack and big data according to claim 1 is characterized in that: The supplier credibility is obtained by quantifying the stability and change frequency of the delivery data based on the division result of the first delivery data, including: the supplier credibility formula is as follows: in, is the supplier credibility value, ∈(0,1], is the variance of structured data for the same order in N deliveries, is the change frequency of unstructured data in N deliveries of the same order, 、 is the penalty coefficient.

8. The method for intelligent supplier expansion management based on technology stack and big data according to claim 1 is characterized in that: Based on the audit efficiency evaluation value and the supplier credibility, the enterprise audit process is optimized and new suppliers are screened; based on the optimized enterprise audit process and the screened new suppliers, the approval data and delivery data of the new suppliers are extracted and defined as second approval data and second delivery data, including: If the audit efficiency evaluation value is greater than or equal to the preset audit efficiency evaluation threshold, the duplicate audit departments are merged; If the audit efficiency evaluation value is less than the preset audit efficiency evaluation threshold, the original audit process is maintained; If the supplier credibility is greater than or equal to a preset supplier credibility threshold, the supplier is determined to be an extended supplier; based on the extended supplier, the approval data and delivery data of the extended supplier are extracted and defined as second approval data and second delivery data; If the supplier credibility is less than the preset supplier credibility threshold, the supplier is re-screened.

9. The method for intelligent supplier expansion management based on technology stack and big data according to claim 1 is characterized in that: determining the approval management process for the new supplier based on the number of reviews and review intervals of the first approval data and the second approval data; Based on the variance and change frequency of structured data and unstructured data of the first delivery data and the second delivery data, determine an expansion management process for the new supplier, including: The approval management process for new suppliers is determined through the approval decision formula. The approval decision formula is as follows: in, is the approval decision value, For the second approval data The maximum number of repeated audits for the same department in each audit timeline, is the average review interval of the second approval data, The total number of sub-review timelines for the second approval data. For the first approval data The maximum number of repeated audits for the same department in each audit timeline, is the average review interval of the first approval data, The total number of sub-review timelines for the first approval data. is the regulating factor; The expansion management process for new suppliers is determined through the expansion management formula. The expansion management formula is as follows: in, To extend the management value, The variance of the structured data for the second delivery data, The frequency of changes to the unstructured data of the second delivery data, The variance of the structured data for the first delivery data, The frequency of changes to unstructured data for first-time delivery.

Images