Self-adaptive decision-making logistics product management method and device

Abstract

The invention relates to the technical field of information management, and discloses an adaptive decision logistics product management method. Comprising the steps of integrating a heterogeneous data cleaning algorithm, performing feature extraction and standardization on structured and unstructured data by adopting a Transform-based cross-modal alignment model, and embedding an anomaly detection model; a self-adaptive rule generator based on reinforcement learning constructs a reward function, and dynamically optimizes a subsidy rule strategy; a time sequence prediction model is combined to predict a future work clothes application demand trend, and an inventory management system is linked to realize a pre-distribution strategy; a graph neural network is adopted to construct an organizational structure knowledge graph, and an optimal approval path is generated; an approval risk prediction model is introduced, approval risks are predicted, and a grading early warning mechanism is triggered; generating a personalized chemical clothes recommendation list based on a mixed recommendation model of collaborative filtering and a knowledge graph; integrating a computer vision model, and automatically matching the size of the work clothes; the method has the advantages of improving employee satisfaction and reducing company cost.

Description

Technical Field

[0001] This invention relates to the field of information management methods, and in particular to an adaptive decision-making method and apparatus for logistics product management. Background Technology

[0002] In the daily operations of a company, the requisition and management of essential work items for employees is a crucial task. However, traditional methods of requisitioning and managing work items have several problems. Unclear subsidy rules mean employees are often unaware of prices and subsidies when applying, potentially leading to conflicts between employees and the company. Manual processing of the requisition process is inefficient and error-prone, resulting in long processing times and disrupting employees' normal work. Furthermore, as companies grow, managing work item information becomes increasingly complex. The lack of a unified information management system means work item information is scattered across various departments or individuals, making centralized management and effective retrieval difficult. This not only increases the difficulty of company management but also hinders the rational allocation of work item resources and cost control.

[0003] Therefore, an adaptive decision-making method and apparatus for logistics product management are provided to solve the above problems. Summary of the Invention

[0004] The main objective of this invention is to address the following technical issues in the existing technology: unclear subsidy rules, employees being unaware of prices and subsidy details when applying, which can easily lead to conflicts between employees and the company; manual application processes are inefficient and prone to errors, resulting in long application cycles and affecting employees' normal work; and as companies expand, the management of work item information becomes increasingly complex, lacking a unified information management system, with work item information scattered across various departments or personnel, making centralized management and effective retrieval difficult.

[0005] The first aspect of this invention provides an adaptive decision-making method for logistics product management, the adaptive decision-making method for logistics product management comprising: It integrates heterogeneous data cleaning algorithms and adopts a Transformer-based cross-modal alignment model to extract and standardize features from structured and unstructured data in human resources, finance, and workwear inventory systems. It also embeds an anomaly detection model to identify and correct outliers in applicant information in real time. An adaptive rule generator based on reinforcement learning is used to construct a reward function and dynamically optimize subsidy rule strategies. Combined with a time series forecasting model, it predicts future demand trends for work uniforms and links with the inventory management system to implement pre-allocation strategies. A graph neural network is used to construct an organizational knowledge graph, dynamically analyze departmental hierarchical relationships, approver workload status and process urgency, and generate the optimal approval path; an approval risk prediction model is introduced to predict approval risks based on the applicant's credit rating and historical violation records, triggering a tiered early warning mechanism. A hybrid recommendation model based on collaborative filtering and knowledge graphs generates a personalized list of recommended work clothes; an integrated computer vision model automatically matches work clothes sizes.

[0006] Optionally, the integrated heterogeneous data cleaning algorithm employs a Transformer-based cross-modal alignment model to extract and standardize features from structured and unstructured data in the human resources system, financial system, and uniform inventory system. It also embeds an anomaly detection model to identify and correct outliers in applicant information in real time, including: Acquire data from the human resources system, the financial system, and the uniform inventory system. The data from the human resources system includes basic employee information and attendance and performance information. The data from the financial system includes income and expenditure details and budget accounting information. The data from the uniform inventory system includes uniform procurement information, inventory quantity information, and distribution record information. Data from the human resources system, the financial system, and the uniform inventory system are cleaned using a heterogeneous data cleaning algorithm. By using the Transformer cross-modal alignment model, we can extract key features from the data in the human resources system, the financial system, and the work clothes inventory system. The extracted information is converted according to a unified data format and standard; An anomaly detection model is embedded, and the Isolation Forest algorithm is used to monitor applicant information in real time, mark abnormal information, generate anomaly correction instructions, and send them to the application end. Obtain the error correction information from the application end and correct the marked error information.

[0007] Optionally, the reinforcement learning-based adaptive rule generator constructs a reward function and dynamically optimizes the subsidy rule strategy: An adaptive rule generator based on reinforcement learning generates subsidy rules; A reward function is constructed based on historical approval records, employee satisfaction feedback, and cost control targets to determine the applicability of subsidy rules. Iterate through the generated subsidy rules, determine the applicability of the subsidy rules, adjust the subsidy rules, and dynamically optimize the subsidy rule strategy.

[0008] Optionally, the step of combining a time series forecasting model to predict future trends in work uniform requisition demand and linking it with the inventory management system to implement a pre-allocation strategy includes: Utilizing LSTM networks to build time series forecasting models; Analyze historical work uniform application data and, in conjunction with time factors, predict future trends in work uniform application demand. The inventory management system is linked to implement a pre-allocation strategy.

[0009] Optionally, the step of using graph neural networks to construct an organizational knowledge graph, dynamically analyzing departmental hierarchical relationships, approver workload status, and process urgency to generate the optimal approval path includes: Graph neural networks are used to construct an organizational knowledge graph to obtain departmental hierarchical relationships, approver workload status, and process urgency. The approval path is designed based on the hierarchical relationship between departments, and the approvers for each step of the approval path are designed based on the urgency of the process and the workload of the approvers, thus generating the optimal approval path.

[0010] Optionally, the introduction of an approval risk prediction model, which predicts approval risks based on the applicant's credit rating and historical violation records, and triggers a tiered early warning mechanism, includes: An approval risk prediction model is introduced, which uses the XGBoost classifier to predict approval risks by combining the applicant's credit rating and historical violation records. When the applicant's credit rating is the first preset rating and the number of historical violations is less than the first preset number, the predicted approval risk is low and the approval will proceed normally. When an applicant's credit rating is the second preset rating and the number of historical violations is greater than the first preset number but less than the second preset number, the predicted approval risk is risky, and the review process will be strengthened. When an applicant's credit rating is the third preset rating and the number of historical violations is greater than the second preset number, the predicted approval risk is high, and additional supporting materials are required to assist in the approval process. The credit ratings are arranged from highest to lowest as follows: first preset rating, second preset rating, and third preset rating.

[0011] Optionally, the hybrid recommendation model based on collaborative filtering and knowledge graph generates a personalized list of recommended work clothes; integrating a computer vision model to automatically match work clothes sizes includes: A hybrid recommendation model based on collaborative filtering and knowledge graph is used to generate a list of recommended work uniforms by obtaining employee job characteristics, seasonal factors and historical wearing preferences. The system integrates the ResNet-50 computer vision model to acquire employee body photos, measures body dimensions based on these photos, and matches these dimensions with uniform size standards.

[0012] A second aspect of the present invention provides an adaptive decision-making logistics product management device, the adaptive decision-making logistics product management device comprising: The data cleaning module is used to integrate heterogeneous data cleaning algorithms. It adopts a Transformer-based cross-modal alignment model to extract and standardize features from structured and unstructured data in human resources, finance, and workwear inventory systems. It also embeds an anomaly detection model to identify and correct outliers in applicant information in real time. The subsidy rule generation module is used to construct a reward function and dynamically optimize the subsidy rule strategy based on an adaptive rule generator using reinforcement learning; combined with a time series forecasting model, it predicts the future demand trend of work clothes and links with the inventory management system to realize a pre-allocation strategy. The approval path generation module is used to construct an organizational structure knowledge graph using graph neural networks, dynamically analyze departmental hierarchical relationships, approver workload status and process urgency, and generate the optimal approval path; it also introduces an approval risk prediction model to predict approval risks based on the applicant's credit rating and historical violation records, triggering a tiered early warning mechanism. The work uniform recommendation and matching module is used to generate a personalized work uniform recommendation list based on a hybrid recommendation model of collaborative filtering and knowledge graph; it also integrates a computer vision model to automatically match work uniform sizes.

[0013] A third aspect of the present invention provides an electronic device, the electronic device comprising a memory and at least one processor, wherein the memory stores instructions; The at least one processor invokes the instructions in the memory to cause the electronic device to perform the various steps of the adaptive decision-making logistics product management method as described above.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the various steps of the adaptive decision-making logistics product management method as described above.

[0015] The technical solution of this invention provides an adaptive decision-making logistics product management method, which integrates a heterogeneous data cleaning algorithm to achieve feature extraction and standardization of data. It also generates subsidy rules through a rule generator and dynamically optimizes the subsidy rules based on employee feedback, thereby improving employee satisfaction and controlling company costs. Furthermore, it provides an approval path generation method that can quickly generate the optimal approval path and predict approval risks, thereby improving approval quality and reducing enterprise operational risks. By integrating a computer vision model to recommend the most suitable uniform size, it solves the problem of employees having to change uniforms multiple times due to unsuitable sizes, thereby improving employee wearing experience and satisfaction. Attached Figure Description

[0016] Figure 1 A flowchart of an adaptive decision-making logistics product management method provided in the first embodiment of the present invention; Figure 2 A schematic diagram of the structure of the adaptive decision-making logistics product management device provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0017] This invention provides an adaptive decision-making method for logistics product management, including: integrating heterogeneous data cleaning algorithms; employing a Transformer-based cross-modal alignment model to extract and standardize features from structured and unstructured data in human resources, finance, and workwear inventory systems; embedding an anomaly detection model to identify and correct outliers in applicant information in real time; constructing a reward function based on reinforcement learning to dynamically optimize subsidy rule strategies; combining a time series forecasting model to predict future workwear application demand trends and linking with the inventory management system to implement pre-allocation strategies; using graph neural networks to construct an organizational structure knowledge graph to dynamically analyze departmental hierarchical relationships, approver workload, and process urgency to generate the optimal approval path; and introducing... The approval risk prediction model predicts approval risks based on the applicant's credit rating and historical violation records, triggering a tiered early warning mechanism. A hybrid recommendation model based on collaborative filtering and knowledge graphs generates a personalized work uniform recommendation list. An integrated computer vision model automatically matches work uniform sizes. This invention addresses several issues in existing technologies, including unclear subsidy rules, employees' lack of understanding of prices and subsidies during application, which can easily lead to conflicts between employees and companies; inefficient and error-prone manual application processes resulting in long application cycles and disruption to normal employee work; and the increasing complexity of work item information management as companies grow, with a lack of a unified information management system leading to information being scattered across various departments or personnel, making centralized management and effective retrieval difficult.

[0018] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0019] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 The first embodiment of the logistics product management method for adaptive decision-making in this invention includes: It integrates heterogeneous data cleaning algorithms and adopts a Transformer-based cross-modal alignment model to extract and standardize features from structured and unstructured data in human resources, finance, and workwear inventory systems. It also embeds an anomaly detection model to identify and correct outliers in applicant information in real time. Specifically, this includes: Acquire data from the human resources system, the financial system, and the uniform inventory system. The data from the human resources system includes basic employee information and attendance and performance information. The data from the financial system includes income and expenditure details and budget accounting information. The data from the uniform inventory system includes uniform procurement information, inventory quantity information, and distribution record information. Data from the human resources system, the financial system, and the uniform inventory system are cleaned using a heterogeneous data cleaning algorithm. By using the Transformer cross-modal alignment model, we can extract key features from the data in the human resources system, the financial system, and the work clothes inventory system. The extracted information is transformed according to a unified data format and standard. Because the data in the human resources system, financial system, and uniform inventory system contains both structured data in tabular form and unstructured data such as documents and emails, data silos are easily formed. To break down these silos, heterogeneous data cleaning algorithms will be integrated, and a Transformer-based cross-modal alignment model will be used to deeply mine key features from data in different systems. For example, information such as education and work experience will be extracted from employee resume documents in the human resources system, integrated with the structured employee basic information table, and transformed according to a unified data format and standard, allowing the originally scattered data to "speak the same language" and achieving efficient data fusion.

[0020] An anomaly detection model is embedded, and the Isolation Forest algorithm is used to monitor applicant information in real time, mark abnormal information, generate anomaly correction instructions, and send them to the application end. The system retrieves and corrects anomaly information from the application process. For example, in the employee onboarding process, if the system detects discrepancies such as an onboarding date earlier than the employment contract signing date, or a department code that does not match the company's existing departmental structure, it will immediately mark the anomaly and prompt relevant personnel to correct the information. This ensures the accuracy and consistency of the data, laying a solid foundation for subsequent business processing.

[0021] An adaptive rule generator based on reinforcement learning is used to construct a reward function and dynamically optimize subsidy rule strategies. Combined with a time series forecasting model, it predicts future demand trends for work uniforms and links with the inventory management system to implement pre-allocation strategies. A graph neural network is used to construct an organizational knowledge graph, dynamically analyze departmental hierarchical relationships, approver workload status and process urgency, and generate the optimal approval path; an approval risk prediction model is introduced to predict approval risks based on the applicant's credit rating and historical violation records, triggering a tiered early warning mechanism. A hybrid recommendation model based on collaborative filtering and knowledge graphs generates a personalized list of recommended work clothes; an integrated computer vision model automatically matches work clothes sizes.

[0022] Please see Figure 1 The second embodiment of the logistics product management method for adaptive decision-making in this invention includes: It integrates heterogeneous data cleaning algorithms and adopts a Transformer-based cross-modal alignment model to extract and standardize features from structured and unstructured data in human resources, finance, and workwear inventory systems. It also embeds an anomaly detection model to identify and correct outliers in applicant information in real time. Specifically, this includes: Acquire data from the human resources system, the financial system, and the uniform inventory system. The data from the human resources system includes basic employee information and attendance and performance information. The data from the financial system includes income and expenditure details and budget accounting information. The data from the uniform inventory system includes uniform procurement information, inventory quantity information, and distribution record information. Data from the human resources system, the financial system, and the uniform inventory system are cleaned using a heterogeneous data cleaning algorithm. By using the Transformer cross-modal alignment model, we can extract key features from the data in the human resources system, the financial system, and the work clothes inventory system. The extracted information is transformed according to a unified data format and standard. Because the data in the human resources system, financial system, and uniform inventory system contains both structured data in tabular form and unstructured data such as documents and emails, data silos are easily formed. To break down these silos, heterogeneous data cleaning algorithms will be integrated, and a Transformer-based cross-modal alignment model will be used to deeply mine key features from data in different systems. For example, information such as education and work experience will be extracted from employee resume documents in the human resources system, integrated with the structured employee basic information table, and transformed according to a unified data format and standard, allowing the originally scattered data to "speak the same language" and achieving efficient data fusion.

[0023] An anomaly detection model is embedded, and the Isolation Forest algorithm is used to monitor applicant information in real time, mark abnormal information, generate anomaly correction instructions, and send them to the application end. The system retrieves and corrects anomaly information from the application process. For example, in the employee onboarding process, if the system detects discrepancies such as an onboarding date earlier than the employment contract signing date, or a department code that does not match the company's existing departmental structure, it will immediately mark the anomaly and prompt relevant personnel to correct the information. This ensures the accuracy and consistency of the data, laying a solid foundation for subsequent business processing.

[0024] An adaptive rule generator based on reinforcement learning is used to construct a reward function and dynamically optimize subsidy rule strategies. Combined with a time series forecasting model, it predicts future demand trends for work uniforms and links with the inventory management system to implement pre-allocation strategies. Specifically, the adaptive rule generator based on reinforcement learning constructs a reward function and dynamically optimizes the subsidy rule strategy, including: An adaptive rule generator based on reinforcement learning generates subsidy rules; The system constructs a reward function based on historical approval records, employee satisfaction feedback, and cost control targets to determine the applicability of subsidy rules. For example, when subsidy rules meet employee needs, improve satisfaction, and effectively control enterprise costs, the system will give a higher reward; otherwise, a lower reward will be given.

[0025] Iterate through the generated subsidy rules, determine the applicability of the subsidy rules, adjust the subsidy rules, and dynamically optimize the subsidy rule strategy.

[0026] The method of combining a time series forecasting model to predict future demand trends for work uniforms and linking it with the inventory management system to implement a pre-allocation strategy includes: Utilizing LSTM networks to build time series forecasting models; Analyze historical work uniform application data and, in conjunction with time factors, predict future trends in work uniform application demand. The inventory management system is linked to implement a pre-allocation strategy.

[0027] If the system predicts a significant increase in demand for short-sleeved work clothes for outdoor workers during the hot summer months, it will proactively link with the inventory management system to pre-allocate the relevant work clothes according to the predicted quantities, thus avoiding inventory backlogs or stockouts and achieving a rational allocation of work clothes resources.

[0028] A graph neural network is used to construct an organizational knowledge graph, dynamically analyze departmental hierarchical relationships, approver workload status and process urgency, and generate the optimal approval path; an approval risk prediction model is introduced to predict approval risks based on the applicant's credit rating and historical violation records, triggering a tiered early warning mechanism. A hybrid recommendation model based on collaborative filtering and knowledge graphs generates a personalized list of recommended work clothes; an integrated computer vision model automatically matches work clothes sizes.

[0029] Please see Figure 1 The third embodiment of the logistics product management method for adaptive decision-making in this invention includes: It integrates heterogeneous data cleaning algorithms and adopts a Transformer-based cross-modal alignment model to extract and standardize features from structured and unstructured data in human resources, finance, and workwear inventory systems. It also embeds an anomaly detection model to identify and correct outliers in applicant information in real time. Specifically, this includes: Acquire data from the human resources system, the financial system, and the uniform inventory system. The data from the human resources system includes basic employee information and attendance and performance information. The data from the financial system includes income and expenditure details and budget accounting information. The data from the uniform inventory system includes uniform procurement information, inventory quantity information, and distribution record information. Data from the human resources system, the financial system, and the uniform inventory system are cleaned using a heterogeneous data cleaning algorithm. By using the Transformer cross-modal alignment model, we can extract key features from the data in the human resources system, the financial system, and the work clothes inventory system. The extracted information is transformed according to a unified data format and standard. Because the data in the human resources system, financial system, and uniform inventory system contains both structured data in tabular form and unstructured data such as documents and emails, data silos are easily formed. To break down these silos, heterogeneous data cleaning algorithms will be integrated, and a Transformer-based cross-modal alignment model will be used to deeply mine key features from data in different systems. For example, information such as education and work experience will be extracted from employee resume documents in the human resources system, integrated with the structured employee basic information table, and transformed according to a unified data format and standard, allowing the originally scattered data to "speak the same language" and achieving efficient data fusion.

[0030] An anomaly detection model is embedded, and the Isolation Forest algorithm is used to monitor applicant information in real time, mark abnormal information, generate anomaly correction instructions, and send them to the application end. The system retrieves and corrects anomaly information from the application process. For example, in the employee onboarding process, if the system detects discrepancies such as an onboarding date earlier than the employment contract signing date, or a department code that does not match the company's existing departmental structure, it will immediately mark the anomaly and prompt relevant personnel to correct the information. This ensures the accuracy and consistency of the data, laying a solid foundation for subsequent business processing.

[0031] An adaptive rule generator based on reinforcement learning is used to construct a reward function and dynamically optimize subsidy rule strategies. Combined with a time series forecasting model, it predicts future demand trends for work uniforms and links with the inventory management system to implement pre-allocation strategies. Specifically, the adaptive rule generator based on reinforcement learning constructs a reward function and dynamically optimizes the subsidy rule strategy, including: An adaptive rule generator based on reinforcement learning generates subsidy rules; The system constructs a reward function based on historical approval records, employee satisfaction feedback, and cost control targets to determine the applicability of subsidy rules. For example, when subsidy rules meet employee needs, improve satisfaction, and effectively control enterprise costs, the system will give a higher reward; otherwise, a lower reward will be given.

[0032] Iterate through the generated subsidy rules, determine the applicability of the subsidy rules, adjust the subsidy rules, and dynamically optimize the subsidy rule strategy.

[0033] The method of combining a time series forecasting model to predict future demand trends for work uniforms and linking it with the inventory management system to implement a pre-allocation strategy includes: Utilizing LSTM networks to build time series forecasting models; Analyze historical work uniform application data and, in conjunction with time factors, predict future trends in work uniform application demand. The inventory management system is linked to implement a pre-allocation strategy.

[0034] If the system predicts a significant increase in demand for short-sleeved work clothes for outdoor workers during the hot summer months, it will proactively link with the inventory management system to pre-allocate the relevant work clothes according to the predicted quantities, thus avoiding inventory backlogs or stockouts and achieving a rational allocation of work clothes resources.

[0035] A graph neural network is used to construct an organizational knowledge graph, dynamically analyze departmental hierarchical relationships, approver workload status and process urgency, and generate the optimal approval path; an approval risk prediction model is introduced to predict approval risks based on the applicant's credit rating and historical violation records, triggering a tiered early warning mechanism. Among these methods, a graph neural network is used to construct an organizational knowledge graph, dynamically analyze departmental hierarchical relationships, approver workload status, and process urgency, and generate the optimal approval path, including: Graph neural networks are used to construct an organizational knowledge graph to obtain departmental hierarchical relationships, approver workload status, and process urgency. The system designs specific steps for approval paths based on departmental hierarchical relationships, and assigns approvers to each step based on process urgency and approver workload, generating the optimal approval path. For example, if an approval task is urgent and the person responsible for it is currently overloaded, the system will automatically assign the task to another suitable approver, improving approval efficiency.

[0036] This includes introducing an approval risk prediction model that predicts approval risks based on the applicant's credit rating and historical violation records, triggering a tiered early warning mechanism that includes: An approval risk prediction model is introduced, which uses the XGBoost classifier to predict approval risks by combining the applicant's credit rating and historical violation records. When the applicant's credit rating is the first preset rating and the number of historical violations is less than the first preset number, the predicted approval risk is low and the approval will proceed normally. When an applicant's credit rating is the second preset rating and the number of historical violations is greater than the first preset number but less than the second preset number, the predicted approval risk is risky, and the review process will be strengthened. When an applicant's credit rating is the third preset rating and the number of historical violations is greater than the second preset number, the predicted approval risk is high, and additional supporting materials are required to assist in the approval process. The credit ratings are ranked from highest to lowest as follows: first preset rating, second preset rating, and third preset rating. This design allows for different handling measures based on the risk level, reducing operational risks for businesses.

[0037] A hybrid recommendation model based on collaborative filtering and knowledge graphs generates a personalized list of recommended work clothes; an integrated computer vision model automatically matches work clothes sizes.

[0038] Specifically, this includes: A hybrid recommendation model based on collaborative filtering and knowledge graph is used to generate a list of recommended work uniforms by obtaining employee job characteristics, seasonal factors and historical wearing preferences. The system integrates the ResNet-50 computer vision model to acquire employee body photos, measures body dimensions based on these photos, and matches these dimensions with uniform size standards.

[0039] The above describes the adaptive decision-making logistics product management method for the logistics industry in embodiments of the present invention. The following describes the adaptive decision-making logistics product management device in embodiments of the present invention. Please refer to [link / reference]. Figure 2 The adaptive decision-making logistics product management device in this embodiment of the invention includes, for the above embodiments: The data cleaning module is used to integrate heterogeneous data cleaning algorithms. It adopts a Transformer-based cross-modal alignment model to extract and standardize features from structured and unstructured data in human resources, finance, and workwear inventory systems. It also embeds an anomaly detection model to identify and correct outliers in applicant information in real time. The subsidy rule generation module is used to construct a reward function and dynamically optimize the subsidy rule strategy based on an adaptive rule generator using reinforcement learning; combined with a time series forecasting model, it predicts the future demand trend of work clothes and links with the inventory management system to realize a pre-allocation strategy. The approval path generation module is used to construct an organizational structure knowledge graph using graph neural networks, dynamically analyze departmental hierarchical relationships, approver workload status and process urgency, and generate the optimal approval path; it also introduces an approval risk prediction model to predict approval risks based on the applicant's credit rating and historical violation records, triggering a tiered early warning mechanism. The work uniform recommendation and matching module is used to generate a personalized work uniform recommendation list based on a hybrid recommendation model of collaborative filtering and knowledge graph; it also integrates a computer vision model to automatically match work uniform sizes.

[0040] above Figure 2The adaptive decision-making logistics product management device in the embodiments of the present invention will be described in detail from the perspective of modular functional entities. The electronic equipment in the embodiments of the present invention will be described in detail from the perspective of hardware processing.

[0041] Figure 3 This is a schematic diagram of the structure of an electronic device 700 provided in an embodiment of the present invention. The electronic device 700 can vary significantly due to differences in configuration or performance. It may include one or more processors 710 (e.g., one or more processors) and a memory 720, and one or more storage media 730 (e.g., one or more storage devices, including RAM, FLASH, etc.) for storing application programs 733 or data 732. The memory 720 and storage media 730 can be temporary or persistent storage. The program stored in the storage media 730 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the electronic device 700. Furthermore, the processor 710 may be configured to communicate with the storage media 730 and execute the series of instruction operations in the storage media 730 on the electronic device 700.

[0042] The electronic device 700 may also include one or more power supplies 740, one or more input / output interfaces 750, and / or one or more operating systems 731, such as FreeRTOS, Android, etc. Those skilled in the art will understand that... Figure 3 The illustrated electronic device structure does not constitute a limitation on electronic devices and may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0043] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of an adaptive decision-making logistics product management method.

[0044] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0045] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, mobile device, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0046] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A logistics product management method with adaptive decision-making, characterized in that, The adaptive decision-making logistics product management method includes: It integrates heterogeneous data cleaning algorithms and adopts a Transformer-based cross-modal alignment model to extract and standardize features from structured and unstructured data in human resources, finance, and workwear inventory systems. It also embeds an anomaly detection model to identify and correct outliers in applicant information in real time. An adaptive rule generator based on reinforcement learning is used to construct a reward function and dynamically optimize subsidy rule strategies. Combined with a time series forecasting model, it predicts future demand trends for work uniforms and links with the inventory management system to implement pre-allocation strategies. A graph neural network is used to construct an organizational knowledge graph, dynamically analyze departmental hierarchical relationships, approver workload status and process urgency, and generate the optimal approval path; an approval risk prediction model is introduced to predict approval risks based on the applicant's credit rating and historical violation records, triggering a tiered early warning mechanism. A hybrid recommendation model based on collaborative filtering and knowledge graphs generates a personalized list of recommended work clothes; an integrated computer vision model automatically matches work clothes sizes.

2. The adaptive decision-making logistics product management method according to claim 1, characterized in that, The integrated heterogeneous data cleaning algorithm employs a Transformer-based cross-modal alignment model to extract and standardize features from structured and unstructured data in human resources, finance, and workwear inventory systems. It also embeds an anomaly detection model to identify and correct outliers in applicant information in real time, including: Acquire data from the human resources system, the financial system, and the uniform inventory system. The data from the human resources system includes basic employee information and attendance and performance information. The data from the financial system includes income and expenditure details and budget accounting information. The data from the uniform inventory system includes uniform procurement information, inventory quantity information, and distribution record information. Data from the human resources system, the financial system, and the uniform inventory system are cleaned using a heterogeneous data cleaning algorithm. By using the Transformer cross-modal alignment model, we can extract key features from the data in the human resources system, the financial system, and the work clothes inventory system. The extracted information is converted according to a unified data format and standard; An anomaly detection model is embedded, and the Isolation Forest algorithm is used to monitor applicant information in real time, mark abnormal information, generate anomaly correction instructions, and send them to the application end. Obtain the anomaly correction information from the application end and correct the marked anomaly information.

3. The adaptive decision-making logistics product management method according to claim 1, characterized in that, The reinforcement learning-based adaptive rule generator constructs a reward function and dynamically optimizes the subsidy rule strategy. An adaptive rule generator based on reinforcement learning generates subsidy rules; A reward function is constructed based on historical approval records, employee satisfaction feedback, and cost control targets to determine the applicability of subsidy rules. Iterate through the generated subsidy rules, determine the applicability of the subsidy rules, adjust the subsidy rules, and dynamically optimize the subsidy rule strategy.

4. The adaptive decision-making logistics product management method according to claim 1, characterized in that, The method of combining time series forecasting models to predict future demand trends for work uniforms and linking them with the inventory management system to implement a pre-allocation strategy includes: Utilizing LSTM networks to build time series forecasting models; Analyze historical work uniform application data and, in conjunction with time factors, predict future trends in work uniform application demand. The inventory management system is linked to implement a pre-allocation strategy.

5. The adaptive decision-making logistics product management method according to claim 1, characterized in that, The method of using graph neural networks to construct an organizational knowledge graph, dynamically analyzing departmental hierarchical relationships, approver workload status, and process urgency to generate the optimal approval path includes: Graph neural networks are used to construct an organizational knowledge graph to obtain departmental hierarchical relationships, approver workload status, and process urgency. The approval path is designed based on the departmental hierarchy, and the approvers for each step are designed based on the urgency of the process and the approver's workload, thus generating the optimal approval path.

6. The adaptive decision-making logistics product management method according to claim 1, characterized in that, The introduction of the approval risk prediction model, which predicts approval risks based on the applicant's credit rating and historical violation records, triggers a tiered early warning mechanism including: An approval risk prediction model is introduced, which uses the XGBoost classifier to predict approval risks by combining the applicant's credit rating and historical violation records. When the applicant's credit rating is the first preset rating and the number of historical violations is less than the first preset number, the predicted approval risk is low and the approval will proceed normally. When an applicant's credit rating is the second preset rating and the number of historical violations is greater than the first preset number but less than the second preset number, the predicted approval risk is risky, and the review process will be strengthened. When an applicant's credit rating is the third preset rating and the number of historical violations is greater than the second preset number, the predicted approval risk is high, and additional supporting materials are required to assist in the approval process. The credit ratings are arranged from highest to lowest as follows: first preset rating, second preset rating, and third preset rating.

7. The adaptive decision-making logistics product management method according to claim 1, characterized in that, The hybrid recommendation model based on collaborative filtering and knowledge graph generates a personalized list of work uniform recommendations. Integrating computer vision models to automatically match work uniform sizes includes: A hybrid recommendation model based on collaborative filtering and knowledge graph is used to generate a list of recommended work uniforms by obtaining employee job characteristics, seasonal factors and historical wearing preferences. The system integrates the ResNet-50 computer vision model to acquire employee body photos, measures body dimensions based on these photos, and matches these dimensions with uniform size standards.

8. An adaptive decision-making logistics product management device, used in the adaptive decision-making logistics product management method as described in any one of claims 1-7, characterized in that, include: The data cleaning module is used to integrate heterogeneous data cleaning algorithms. It adopts a Transformer-based cross-modal alignment model to extract and standardize features from structured and unstructured data in human resources, finance, and workwear inventory systems. It also embeds an anomaly detection model to identify and correct outliers in applicant information in real time. The subsidy rule generation module is used to construct a reward function and dynamically optimize the subsidy rule strategy based on an adaptive rule generator using reinforcement learning; combined with a time series forecasting model, it predicts the future demand trend of work clothes and links with the inventory management system to realize a pre-allocation strategy. The approval path generation module is used to construct an organizational structure knowledge graph using graph neural networks, dynamically analyze departmental hierarchical relationships, approver workload status and process urgency, and generate the optimal approval path; it also introduces an approval risk prediction model to predict approval risks based on the applicant's credit rating and historical violation records, triggering a tiered early warning mechanism. The work uniform recommendation and matching module is used to generate a personalized work uniform recommendation list based on a hybrid recommendation model that combines collaborative filtering and knowledge graphs. Integrating computer vision models, it automatically matches work uniform sizes.

9. An electronic device comprising a memory and at least one processor, wherein the memory stores instructions; characterized in that, The at least one processor invokes the instructions in the memory to cause the electronic device to perform the steps of the adaptive decision-making logistics product management method as described in any one of claims 1-7.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the various steps of the logistics product management method for adaptive decision-making as described in any one of claims 1-7.

Images