Knowledge sharing excitation method and device based on multi-dimensional contribution degree and storage medium

Abstract

The invention discloses a multi-dimensional contribution degree-based knowledge sharing excitation method and device and a storage medium, and relates to the technical field related to the Internet, and the method comprises the following steps: constructing and distributing cross-scene adaptation state metadata; the local node executes local evaluation of the knowledge contribution degree and generates feedback data based on the metadata and the real-time demand data of the local scene, and writes the feedback data into a local adaptation result of the metadata; the aggregation node collects metadata of a plurality of local nodes, and performs federal aggregation on local evaluation related parameters in the metadata; performing full-link adaptation effect simulation verification of knowledge cross-scene circulation based on the updated global optimization parameters, and optimizing metadata according to a verification result; the optimized metadata are distributed to a local node and an aggregation node, the local node updates a knowledge contribution degree evaluation rule, and the aggregation node calibrates local model parameters used for parameter aggregation; the problem of mismatching of excitation resources is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Internet, in particular to a knowledge sharing incentive method and device based on multi-dimensional contribution degree and a storage medium. BACKGROUND

[0002] In the current knowledge sharing incentive method based on multi-dimensional contribution degree, especially in typical application scenarios such as enterprise cross-department knowledge base and industry technology sharing platform, there is a fundamental technical contradiction: the mismatch between static and unified contribution degree evaluation system and dynamic and differentiated knowledge value reality.

[0003] This core contradiction specifically manifests that the system lacks the dynamic modeling capability of the "knowledge type-application scenario" synergistic influence relationship, resulting in two key technical defects: one is that it cannot dynamically adjust the dimension weight distribution according to the value representation characteristics of the knowledge type in a specific scenario, and the other is that it fails to adaptively optimize the evaluation index set based on the scenario demand characteristics.

[0004] The direct consequence is that the core value dimension of cross-scenario knowledge is systematically misjudged, such as the innovative value of technical principle knowledge in the R&D scenario and the timeliness value of emergency disposal knowledge in the customer service scenario cannot be accurately quantified, ultimately causing the mismatch of incentive resources: the cross-scenario knowledge contribution with high application value is underestimated, while the low-value knowledge with adaptive deviation is over-incentivized. SUMMARY

[0005] To solve the defects in the prior art, the present application provides a knowledge sharing incentive method and device based on multi-dimensional contribution degree and a storage medium.

[0006] To solve the above technical problems, the present application provides the following technical solutions: The present application provides a knowledge sharing incentive method based on multi-dimensional contribution degree, comprising the following steps: Construct and distribute cross-scenario adaptive state metadata, which contains knowledge type attributes, scenario demand characteristics, local adaptation results and global optimization parameters; The local node performs local evaluation of knowledge contribution degree based on the metadata and real-time demand data of the local scenario, and generates feedback data, and writes the feedback data into the local adaptation result of the metadata; The aggregation node collects the metadata of multiple local nodes, federates the local evaluation related parameters in the metadata, and updates the global optimization parameters of the metadata; Based on the updated global optimization parameters, perform full-link adaptive effect simulation verification of knowledge cross-scenario flow, and optimize the metadata according to the verification result; The optimized metadata is distributed to the local nodes and the aggregation node, the local nodes update the knowledge contribution degree evaluation rule, and the aggregation node calibrates the local model parameters for parameter aggregation.

[0007] As a preferred technical solution of the present application, the construction and distribution of cross-scene adaptation state metadata comprises: The knowledge type accuracy score data is collected through the user evaluation system, the knowledge update frequency data is extracted through the version management system, and the operability feedback data is obtained through the operation behavior analysis system. The technical depth requirement frequency is counted through the research and development document analysis system, the landing feasibility dependence degree data is collected through the customer demand management system, and the fault correlation degree attention priority is calculated through the fault handling record system. The collected raw data is subjected to text labeling conversion and numerical normalization processing, and data of different dimensions is uniformly mapped to a preset numerical interval.

[0008] As a preferred technical solution of the present application, the local node performs local evaluation of knowledge contribution degree based on metadata and real-time requirement data of the local scene comprises: The fault duration data is obtained in real time through the system monitoring module, the processing personnel skill level data is read through the personnel information database, the fault influence range data is calculated through the influence evaluation algorithm, and the project iteration stage data is extracted through the project management system. The real-time requirement data is subjected to data format unification, data validity verification and data feature extraction processing; After writing the local feedback data into the metadata, a timer is set to automatically trigger metadata synchronization at a fixed period, and a deviation monitoring mechanism is set to trigger metadata synchronization immediately when the evaluation deviation exceeds the threshold.

[0009] As a preferred technical solution of the present application, the local node performs local evaluation of knowledge contribution degree further comprises: The weight distribution coefficient of each evaluation dimension is adjusted according to the analysis result of real-time requirement data, and the evaluation index set is customized based on the scene feature recognition result. The performance of knowledge in each dimension is quantitatively scored according to the set weight coefficient, and the comprehensive contribution degree score is obtained by multiplying each dimension score by the corresponding weight coefficient and then accumulating, wherein the adjustment of the weight coefficient is determined dynamically according to the priority index in the real-time requirement data, and the customization of the evaluation index set is dynamically configured according to the requirement type in the scene feature data.

[0010] As a preferred technical solution of the present application, the trigger condition for the aggregation node to collect metadata of multiple local nodes comprises: The proportion of the number of nodes that have received feedback data is periodically counted, and aggregation is triggered when the proportion exceeds a set threshold. The time interval since the last parameter aggregation is monitored by a timer, and automatic triggering is triggered when the set time length is reached; The adaptation deviation data of each core scene is continuously detected by the monitoring system, and immediate triggering is triggered when the adaptation deviation of any core scene exceeds the set threshold value continuously for multiple times.

[0011] As a preferred technical solution of the present application, the local evaluation related parameters in the metadata are federated and aggregated by the aggregation node, including: The key pair required for homomorphic encryption is created by the key generation system, the received metadata field is transformed by homomorphic encryption using the public key, and the encrypted ciphertext data is transmitted to the parameter aggregation server; After the parameter aggregation server completes the aggregation calculation, the private key is used to decrypt the aggregation result, and the global optimization adaptation parameter in plaintext form is recovered; The parameter average algorithm based on data volume weighting is adopted, wherein the weight of each node is determined by the proportion of the size of its local data set to the size of the total data set.

[0012] As a preferred technical solution of the present application, the full-link adaptation effect simulation verification of knowledge cross-scene flow based on the updated global optimization parameter includes: According to the knowledge flow path, a cross-scene flow model is constructed to simulate the complete flow path of knowledge from the research and development scene to the test scene and then to the operation and maintenance scene; The adaptation accuracy index of each scene is calculated in the simulation process, which is obtained by comparing the difference between the simulation contribution degree and the expected contribution degree; The global balance degree index is also calculated, which is obtained by calculating the standard deviation of the adaptation deviation of each scene; According to the adaptation accuracy index in the simulation result, the scene demand feature weight is adjusted, and according to the global balance degree index, the association strength between the knowledge attribute and the scene is adjusted.

[0013] As a preferred technical solution of the present application, after the optimized metadata is distributed to the local node and the aggregation node, the local node updates the knowledge contribution degree evaluation rule, including: The global parameter field content in the optimized metadata is parsed, and the weight calculation logic and index selection logic in the local adaptation rule are updated according to the parsing result; The aggregation node calibrates the local model parameters for parameter aggregation, including: The global optimization adaptation parameters in the optimized metadata are loaded, and the parameter matrix and weight coefficient of the local model are reinitialized using these parameters; Data synchronization is performed through a secure transmission channel, which uses an encryption protocol to encrypt and encapsulate the metadata packet and transmit it.

[0014] The application further provides a knowledge sharing incentive device based on multidimensional contribution degree, comprising: A metadata construction module is configured to construct and distribute cross-scene adaptive state metadata, wherein the metadata comprises knowledge type attributes, scene requirement characteristics, local adaptation results and global optimization parameters; A local evaluation module is arranged at a local node and is configured to perform local evaluation of knowledge contribution degree based on the metadata and real-time requirement data of a local scene and generate feedback data, and write the feedback data into the local adaptation results of the metadata; A federal aggregation module is arranged at an aggregation node and is configured to collect metadata of multiple local nodes, perform federal aggregation on local evaluation related parameters in the metadata, and update global optimization parameters of the metadata; An emulation verification module is configured to perform full-link adaptive effect emulation verification of knowledge cross-scene circulation based on the updated global optimization parameters, and optimize the metadata according to a verification result; An update calibration module is configured to distribute the optimized metadata to the local nodes and the aggregation node, control the local nodes to update knowledge contribution degree evaluation rules, and control the aggregation node to calibrate local model parameters used for parameter aggregation.

[0015] The application further provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the knowledge sharing incentive method.

[0016] The application has the following beneficial effects: 1. In the application, the local nodes perform local evaluation of knowledge contribution degree based on the metadata and real-time requirement data of local scenes, dynamically adjust weight distribution coefficients of each evaluation dimension according to real-time requirement data analysis results, and customize evaluation index sets based on scene characteristic recognition results, thereby effectively solving the technical defect that a static unified evaluation system cannot dynamically adjust dimension weight distribution according to value representation characteristics of knowledge types in specific scenes.

[0017] 2. In the application, the aggregation nodes perform federal aggregation on local evaluation related parameters in the metadata, perform full-link adaptive effect emulation verification of knowledge cross-scene circulation based on the updated global optimization parameters, and optimize the metadata according to a verification result, thereby solving the problem that the system lacks dynamic modeling capability for the collaborative influence relationship between "knowledge type-application scene".

[0018] 3. In the application, the optimized metadata is distributed to the local nodes and the aggregation node, the local nodes update knowledge contribution degree evaluation rules, the aggregation node calibrates local model parameters used for parameter aggregation, and a closed-loop optimization mechanism is formed, thereby solving the incentive resource mismatch problem, and this process ensures that cross-scene knowledge contribution with high application value is not underestimated, and adaptive biased low-value knowledge is not over-incentivized. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0020] Figure 1 This is a flowchart illustrating the knowledge-sharing incentive method of the present invention; Figure 2 This is a schematic diagram of the knowledge sharing incentive device of the present invention. Detailed Implementation

[0021] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0022] like Figure 1 As shown, the knowledge-sharing incentive method based on multidimensional contribution is characterized by the following steps: Construct and distribute cross-scenario adaptation state metadata, which includes knowledge type attributes, scenario requirement characteristics, local adaptation results, and global optimization parameters; The knowledge type attribute describes the essential classification of knowledge, such as technical principle knowledge or emergency response knowledge; the scenario requirement feature describes the knowledge requirement preference of a specific application scenario, such as R&D scenarios emphasizing innovative value and customer service scenarios emphasizing timeliness value; the local adaptation result records the evaluation feedback of the local node's contribution to knowledge; and the global optimization parameter stores the global adjustment parameters after federated aggregation. During the construction process, metadata integrates raw data from user evaluation systems, version management systems, operational behavior analysis systems, R&D document analysis systems, customer requirement management systems, and fault handling record systems through a multi-source data acquisition system. After text tagging and numerical normalization, it forms a standardized metadata format. The distribution process transmits metadata to local nodes and aggregation nodes through network communication protocols. In enterprise cross-departmental knowledge base scenarios, for example, metadata is distributed from the central server to R&D department nodes, testing department nodes, and operations and maintenance department nodes, ensuring that each node obtains consistent initial parameters.

[0023] Based on the metadata and real-time demand data of the local scenario, the local node performs a local evaluation of knowledge contribution and generates feedback data, and writes the feedback data into the local adaptation result of the metadata. The local node is a computing unit deployed in a specific application scenario, such as a department server in an enterprise cross-department knowledge base or a regional server in an industry technology sharing platform. Real-time demand data is obtained in real time through a system monitoring module, a personnel information database, an impact assessment algorithm, and a project management system, including fault duration data, processing personnel skill level data, fault impact range data, and project iteration stage data. The local evaluation process includes data format unification, data validity verification, and data feature extraction processing on real-time demand data to ensure data quality. Subsequently, the local node adjusts the weight distribution coefficient of each evaluation dimension based on the analysis results of real-time demand data and customizes the evaluation index set based on the scene feature recognition results. For example, in a customer service scenario, for emergency handling type knowledge, the local node dynamically increases the weight distribution coefficient of the timeliness dimension and customizes an evaluation index set with response time as the core. After evaluation is completed, the local node generates feedback data and writes the feedback data into the local adaptation results of the metadata. After writing, the local node sets a timer to automatically trigger metadata synchronization at a fixed period, and sets a deviation monitoring mechanism to trigger metadata synchronization immediately when the evaluation deviation exceeds the threshold.

[0024] The aggregation node collects metadata from multiple local nodes, federated aggregates the local evaluation related parameters in the metadata, and updates the global optimization parameters of the metadata. The aggregation node is a central server responsible for global coordination, such as the main server of an enterprise knowledge management platform or the cloud server of an industry technology sharing platform. The aggregation node collects metadata from multiple local nodes, and the triggering conditions include: periodically counting the proportion of nodes that have received feedback data, and triggering aggregation when the proportion exceeds the set threshold; automatically triggering when the time interval since the last parameter aggregation reaches the set time length through the timer; continuously detecting the adaptation deviation data of each core scenario through the monitoring system, and triggering immediately when the adaptation deviation of any core scenario exceeds the set threshold for multiple times in a row. After collecting the metadata, the aggregation node federated aggregates the local evaluation related parameters in the metadata. The federated aggregation process includes: creating a key pair required for homomorphic encryption through a key generation system, using the public key to perform homomorphic encryption transformation on the received metadata fields, and transmitting the encrypted ciphertext data to the parameter aggregation server; after the parameter aggregation server completes the aggregation calculation, using the private key to decrypt the aggregation result to restore the global optimization adaptation parameters in plaintext form; using a parameter average algorithm based on data volume weighting, where the weight of each node is determined by the proportion of its local data set size in the total data set size; after aggregation is completed, the aggregation node updates the global optimization parameters of the metadata.

[0025] simulate and verify the whole-link adaptation effect of knowledge cross-scene flow based on the updated global optimization parameters, and optimize the metadata according to the verification result; The simulation and verification step builds a virtual environment through a computer, simulates the flow process of knowledge among different scenes, and simulates a cross-scene flow model according to a knowledge flow path based on the updated global optimization parameters. The simulation system simulates the complete flow path of knowledge from the research and development scene to the test scene and then to the operation and maintenance scene. In the simulation process, the adaptation accuracy index of each scene is calculated. The index is obtained by comparing the difference between the simulation contribution and the expected contribution. At the same time, the global balance index is calculated. The index is obtained by calculating the standard deviation of the adaptation deviation of each scene. According to the simulation result, the system optimizes the cross-scene adaptation state metadata: according to the adaptation accuracy index in the simulation result, the weight of the scene demand feature is adjusted, and according to the global balance index, the correlation strength of the knowledge attribute and the scene is adjusted. For example, in an industry technology sharing platform, simulation finds that the innovative value of technical principle knowledge in the test scene is underestimated, so the weight of the innovation dimension in the scene demand feature is correspondingly increased.

[0026] The optimized metadata is distributed to the local node and the aggregation node. The local node updates the knowledge contribution evaluation rule, and the aggregation node calibrates the local model parameters used for parameter aggregation. The optimized cross-scene adaptation state metadata is distributed to the local node and the aggregation node through a secure transmission channel. The secure transmission channel uses an encryption protocol to encrypt and encapsulate the metadata packet and transmit it, ensuring data integrity. After receiving the optimized metadata, the local node updates the knowledge contribution evaluation rule: parses the global parameter field content in the optimized metadata, and updates the weight calculation logic and index selection logic in the local adaptation rule according to the parsing result. For example, the local node adjusts the weight distribution coefficient and the evaluation index set according to the new global optimization parameters. After receiving the optimized metadata, the aggregation node calibrates the local model parameters used for parameter aggregation: loads the global optimization adaptation parameters in the optimized metadata, and uses these parameters to reinitialize the parameter matrix and weight coefficient of the local model. This process ensures the accuracy and consistency of subsequent federated aggregation, forming a closed-loop optimization cycle.

[0027] Through the above steps, the method of the present application realizes the dynamic adaptation of knowledge contribution evaluation and the precise configuration of incentive resources, effectively solving the technical problems of cross-scene knowledge value misjudgment and incentive resource mismatch.

[0028] Further, the method comprises: Collect knowledge type accuracy score data through a user evaluation system integrated in the knowledge management platform for collecting user accuracy evaluation of knowledge content. In the enterprise cross-department knowledge base scenario, for example, R&D department employees score the accuracy of technical documents. These score data reflect the reliability and correctness of knowledge in specific application scenarios. The user evaluation system collects user input through a pre-set scoring interface and stores the score data in a structured format as part of the knowledge type attribute in the metadata.

[0029] Extract knowledge update frequency data through a version management system for tracking knowledge document modification history and version change records. In the industry technology sharing platform scenario, for example, the update frequency data of technical standards or specification documents is extracted, which indicates the maintenance activity and timeliness of knowledge. The version management system calculates the number of updates per unit time by analyzing the version log of the document and includes this data in the metadata to support the evaluation of the dynamic value of knowledge.

[0030] Obtain operability feedback data through an operation behavior analysis system that monitors user interaction behavior with knowledge content, including downloading, referencing, and application practice. In the enterprise cross-department knowledge base scenario, for example, the operation behavior of emergency handling class knowledge (such as the adoption rate and execution efficiency of solutions) of customer service department employees is recorded and analyzed to generate operability feedback data. The system quantifies user operation frequency and effect through behavior log analysis algorithm and uses the result as a reference index for scenario demand characteristics in the metadata.

[0031] Statistical technical depth demand frequency through a research and development document analysis system based on natural language processing technology to analyze technical document content and identify the frequency of occurrence of key terms and complex concepts. In the industry technology sharing platform scenario, for example, the technical depth demand frequency of technical principle class knowledge is calculated by statistical density and reference level of professional terms in the document. The system extracts feature word frequency through text mining algorithm and includes frequency data in metadata to reflect the innovative value of knowledge in the research and development scenario.

[0032] Collect landing feasibility dependency degree data through a customer demand management system that stores customer feedback and demand information for evaluating the feasibility and dependency of knowledge in actual application. In the enterprise cross-department knowledge base scenario, for example, the product development department analyzes the dependency degree of knowledge in the landing process based on the data of the customer demand management system. The system calculates the landing feasibility index through the demand priority algorithm and uses the data as part of the scenario demand characteristics in the metadata.

[0033] The fault correlation degree attention priority is calculated by the fault handling record system, the fault handling record system records fault events and handling history, and is used for analyzing the correlation degree of knowledge and faults, in an industry technology sharing platform scene, for example, in an operation and maintenance scene, the correlation degree of emergency disposal type knowledge and high frequency faults is calculated to determine the emergency degree and value priority of the knowledge, the system analyzes the co-occurrence frequency of faults and knowledge points through an association rule mining algorithm, and generates attention priority data, which is included in the metadata.

[0034] Text labeling conversion and numerical normalization processing are performed on the collected raw data, different dimensional data is uniformly mapped to a preset numerical interval, since the raw data comes from a user evaluation system, a version management system, an operation behavior analysis system, a research and development document analysis system, a customer demand management system and a fault handling record system, the dimensions and formats are different, text labeling conversion converts unstructured data (such as user comments or document content) into a uniform label sequence, numerical normalization processing uniformly maps data of different ranges (such as scores, frequencies and priorities) to a preset numerical interval (for example, 0 to 1) through linear transformation or minimum-maximum scaling method, this processing ensures that each field in the metadata has consistency and comparability, providing standardized input for subsequent local evaluation and federal aggregation.

[0035] Further, the local node performs local evaluation of knowledge contribution degree based on the metadata and real-time demand data of the local scene, including: The fault duration data is obtained in real time by the system monitoring module, the system monitoring module is deployed in the running environment of the local node, and is used for continuously monitoring system state and event log, in an enterprise cross-department knowledge base scene, for example, the local node of the customer service department tracks the time interval from the occurrence to the solution of the fault in real time through the system monitoring module, to generate fault duration data, which reflects the timeliness value of the emergency disposal type knowledge in the customer service scene, the system monitoring module collects data through a timestamp record and an event triggering mechanism, and stores the data in a structured format as a component of the real-time demand data.

[0036] The processing personnel skill level data is read from the personnel information database, the personnel information database stores the ability profile and authentication information of employees in the organization, in an industry technology sharing platform scene, for example, the local node of the operation and maintenance scene queries the skill level data of the personnel processing specific faults through the personnel information database, the data indicates the ability level of the personnel applying knowledge, the personnel information database provides the skill level data through a database query interface, the local node reads the data through SQL query or API calling, and the data is included in the real-time demand data.

[0037] The impact evaluation algorithm is a rule-based or machine learning-based calculation model for quantifying the impact of a fault on a system or business. In the enterprise cross-department knowledge base scenario, for example, the local node of the R&D department analyzes the module dependency interruption range caused by the fault through the impact evaluation algorithm to generate fault impact range data. The impact evaluation algorithm outputs an impact range score by inputting fault type, system topology, and business indicators. This score serves as a key indicator of real-time demand data.

[0038] The project management system is used to manage project lifecycle and milestone information. In the industry technology sharing platform scenario, for example, the local node of the test scenario obtains current project iteration phase data such as development phase, test phase, or release phase through the project management system. The project management system provides project iteration phase data through data export interfaces or event push mechanisms. The local node parses this data to identify scenario demand dynamics.

[0039] Data format unification, data validity verification, and data feature extraction are performed on real-time demand data. Data format unification converts heterogeneous data from system monitoring modules, personnel information databases, impact evaluation algorithms, and project management systems into a standard format. Data validity verification eliminates abnormal data through rule checks. Data feature extraction extracts key features from raw data through feature engineering methods, such as extracting impact severity features from fault impact range data and stage urgency features from project iteration phase data.

[0040] After writing local feedback data into metadata, set a timer to automatically trigger metadata synchronization at fixed intervals. At the same time, set a deviation monitoring mechanism to trigger metadata synchronization immediately when the evaluation deviation exceeds the threshold. Local feedback data includes knowledge contribution scores and adjustment parameters evaluated based on real-time demand data. Write operations are implemented through metadata APIs or database update operations to ensure timely updates to local adaptation results fields. Subsequently, the local node sets a timer to automatically trigger metadata synchronization at fixed intervals. The timer is based on system clock or scheduling tasks to trigger metadata synchronization at fixed time intervals (e.g., every hour or every day). Updated metadata is sent to the aggregation node. At the same time, the local node sets a deviation monitoring mechanism to trigger metadata synchronization immediately when the evaluation deviation exceeds the threshold. The deviation monitoring mechanism compares the differences between current evaluation results and historical evaluation results to calculate the deviation value. When the deviation value exceeds the pre-set threshold (e.g., 10%), metadata synchronization is triggered immediately to ensure timely adjustment of global parameters.

[0041] Through the above steps, the local node can dynamically adapt to changes in scenario requirements, accurately quantify knowledge contribution, and provide a basis for the rational allocation of incentive resources in application scenarios such as enterprise cross-department knowledge bases and industry technology sharing platforms.

[0042] Further, the local node performing local evaluation of knowledge contribution further comprises: adjusting the weight distribution coefficients of each evaluation dimension according to the real-time demand data analysis results, and customizing the evaluation index set based on the scenario feature recognition results; The real-time demand data analysis results are structured outputs obtained by processing real-time demand data, including fault duration data, processing personnel skill level data, fault impact range data, and project iteration stage data, etc. The local node uses a data analysis algorithm (such as a priority sorting algorithm or a machine learning model) to analyze the priority indicators in the real-time demand data, dynamically determines the weight distribution coefficients of each evaluation dimension, and the adjustment of the weight distribution coefficients is realized through the parameter update module of the local node, which dynamically calculates and stores new weight values based on the priority indicators (such as urgency or importance scores) in the real-time demand data;

[0043] The scenario feature recognition results are classification information obtained by analyzing scenario demand characteristics, such as identifying the current scenario as a research and development scenario, a test scenario, or an operation and maintenance scenario. The local node uses a scenario classification algorithm (such as a rule engine or a clustering model) to process scenario demand characteristic data, dynamically configures the evaluation index set, and in the operation and maintenance scenario, the scenario feature recognition results indicate that reliability and response time need to be focused on, and the local node customizes the evaluation index set containing the average fault interval time and the average repair time; in the test scenario, the scenario feature recognition results indicate that coverage and defect detection rate need to be focused on, and the local node customizes the evaluation index set containing the test case coverage rate and the defect detection rate, and the customization of the evaluation index set is realized through the index management module of the local node, which dynamically selects and activates related indicators according to the demand type (such as stability demand or efficiency demand) in the scenario characteristic data.

[0044] The performance of knowledge in each dimension is quantitatively scored according to the set weight coefficients, and the comprehensive contribution score is obtained by multiplying and accumulating the scores of each dimension and the corresponding weight coefficients, wherein the adjustment of the weight coefficients is dynamically determined according to the priority indicators in the real-time demand data, and the customization of the evaluation index set is dynamically configured according to the demand type in the scenario characteristic data; The quantitative scoring process is performed by a scoring engine, which numerically evaluates the performance of knowledge in each evaluation dimension based on predefined scoring rules, for example, for technical principle type knowledge, in the R&D scenario, the score of the innovation dimension is obtained by counting the number of new concepts in the knowledge and the application potential; for emergency disposal type knowledge, in the customer service scenario, the score of the timeliness dimension is obtained by calculating the average resolution time after the application of the knowledge, then the scoring engine multiplies each dimension score by the corresponding weight coefficient, and calculates the comprehensive contribution score by weighted summation algorithm, the whole calculation process is completed in the evaluation calculation module of the local node, which ensures that the adjustment of the weight coefficient is dynamically determined according to the priority index in the real-time demand data, and the customization of the evaluation index set is dynamically configured according to the demand type in the scene characteristic data.

[0045] Further, the trigger conditions for the aggregation node to collect the metadata of the plurality of local nodes include: Periodically, the proportion of the number of nodes that have received feedback data is calculated, and aggregation is triggered when the proportion exceeds a set threshold; The aggregation node maintains a node registration table to record the identifiers and state information of all local nodes, at the beginning of each statistical period, the aggregation node queries the metadata feedback state of each local node through the node communication interface, calculates the number of nodes that have received feedback data, and then calculates the proportion of the number of nodes that have received feedback data to the total number of nodes, when the proportion exceeds a set threshold (e.g. 80%), the aggregation operation is automatically triggered.

[0046] The time interval since the last parameter aggregation is monitored by a timer, and aggregation is automatically triggered when the set time length is reached; The aggregation node has a built-in timer module that records the timestamp of the last parameter aggregation and continuously monitors the time interval between the current time and the last parameter aggregation time, when the time interval reaches a set time length (e.g. 24 hours), the timer module automatically triggers the aggregation operation regardless of the state of the node feedback data reception.

[0047] The adaptation deviation data of each core scenario is continuously detected by a monitoring system, and aggregation is immediately triggered when the adaptation deviation of any core scenario continuously exceeds a set threshold for multiple times; The monitoring system continuously collects the adaptation deviation data of each local node, which is calculated by comparing the difference between the local evaluation result and the expected value, the monitoring system analyzes the adaptation deviation data through a data stream processing engine, and when it detects that the adaptation deviation of any core scenario (such as the R&D scenario or the customer service scenario) continuously exceeds a set threshold (e.g. 15%) for multiple times (e.g. 3 times in a row), the aggregation operation is immediately triggered.

[0048] Through the above trigger condition, the aggregation node can intelligently start federated aggregation in application scenarios such as enterprise cross-department knowledge base and industry technology sharing platform, balance aggregation efficiency and parameter accuracy, and support dynamic optimization of knowledge contribution evaluation system.

[0049] Further, the aggregation node federatedly aggregates the local evaluation related parameters in the metadata, including: The key generation system creates a key pair required for homomorphic encryption, uses the public key to perform homomorphic encryption transformation on the received metadata field, and transmits the encrypted ciphertext data to the parameter aggregation server. The key generation system is deployed in the security environment of the aggregation node, and generates a key pair containing a public key and a private key using an asymmetric encryption algorithm. Optionally, the aggregation node calls the key generation system to create a key pair required for RSA or Paillier encryption algorithm in the initialization stage, where the public key is used for encrypting data and the private key is used for decrypting data. Each local node uses the obtained public key to perform homomorphic encryption transformation on the local evaluation related parameters in its metadata. This encryption method allows mathematical operations to be performed in the ciphertext state.

[0050] After the parameter aggregation server completes the aggregation calculation, the private key is used to decrypt the aggregation result to restore the global optimization adaptation parameter in plaintext form. The parameter aggregation server performs a parameter average algorithm based on data volume weighting in the ciphertext state to calculate the weighted average value. After the aggregation calculation is completed, the parameter aggregation server returns the encrypted aggregation result to the aggregation node. The aggregation node uses the private key to decrypt the aggregation result. The decryption process is implemented through an asymmetric decryption algorithm, ensuring that only the aggregation node holding the private key can restore the plaintext data.

[0051] The parameter average algorithm based on data volume weighting is adopted, where the weight of each node is determined by the proportion of the size of its local data set to the size of the total data set. When the parameter aggregation server performs weighted average calculation, it first counts the size of the local data set of each local node, calculates the proportion of the size of the local data set of each node in the size of the total data set, and takes this proportion as the weight coefficient of the node. In the industry technology sharing platform scenario, for example, the local data set sizes of the research and development scenario node, the test scenario node, and the operation and maintenance scenario node are 1000, 800, and 1200 respectively, and the total data set size is 3000. Therefore, the weight coefficients of the three nodes are 0.33, 0.27, and 0.40 respectively. The parameter aggregation server performs weighted average calculation on the encrypted local evaluation related parameters according to these weight coefficients, ensuring that the node with larger data volume has greater influence on the global parameter.

[0052] Through the above steps, in application scenarios such as cross-departmental knowledge bases and industry technology sharing platforms, aggregation nodes can achieve local evaluation of relevant parameters through federated aggregation while protecting data privacy. This generates globally optimized adaptation parameters that accurately reflect the value of global knowledge, providing a reliable foundation for subsequent simulation verification and rule updates.

[0053] Furthermore, the simulation verification of the end-to-end adaptation effect of knowledge cross-scenario transfer based on the updated global optimization parameters includes: Based on the knowledge flow path, a cross-scenario flow model is constructed to simulate the complete flow path of knowledge from the R&D scenario to the testing scenario and then to the operation and maintenance scenario. The cross-scenario knowledge transfer model is a computational model built on the knowledge lifecycle theory. This model defines the rules and transformation mechanisms for knowledge transfer between different scenarios. The model describes the value changes of knowledge in different scenarios through state transition equations. Among them, the R&D scenario focuses on innovation indicators, the testing scenario focuses on feasibility indicators, and the operation and maintenance scenario focuses on stability indicators. The model construction process includes steps such as defining scenario nodes, setting transfer rules, and configuring evaluation parameters.

[0054] During the simulation, the adaptation accuracy index for each scenario is calculated. This index is obtained by comparing the difference between the simulation contribution and the expected contribution. The simulation contribution is a quantitative result of knowledge value calculated through a cross-scenario flow model. The expected contribution is a benchmark value set based on historical data or expert evaluation. The adaptation accuracy index uses the root mean square error or mean absolute percentage error calculation method to quantify the degree of deviation between the simulation result and the expected value. This index reflects the accuracy of the simulation system in assessing knowledge value.

[0055] Simultaneously, a global balance index is calculated, which is obtained by statistically analyzing the standard deviation of the adaptation bias in each scenario. The global balance index is used to measure the consistency level of knowledge value assessment across different scenarios. It quantifies the balance performance of the system by calculating the standard deviation of the adaptation deviation of each scenario. The adaptation deviation refers to the absolute difference between the simulation contribution and the expected contribution of each scenario. The smaller the index value, the better the balance of knowledge value assessment across different scenarios.

[0056] Adjust the weights of scenario requirement features based on the adaptation accuracy index in the simulation results, and adjust the correlation strength between knowledge attributes and scenarios based on the global balance index. Based on the analysis results of the adaptation accuracy index, the system dynamically adjusts the weight allocation of each dimension in the scenario requirement characteristics. When the accuracy index of the R&D scenario adaptation shows that there is a deviation in the assessment of innovative value, the weight of the innovative dimension in the requirements of the R&D scenario is increased accordingly. Based on the analysis results of the global balance index, the system adjusts the matching relationship between knowledge attributes and scenarios. When the global balance index shows that the correlation strength of the technical principle knowledge in the test scenario is insufficient, the association parameter of the knowledge attribute with the test scenario is enhanced; These adjustments are achieved through parameter optimization algorithms, ensuring continuous improvement of the accuracy and consistency of knowledge value assessment.

[0057] Through the above steps, the simulation verification system can comprehensively evaluate the adaptation effect of knowledge cross-scene transfer in enterprise cross-department knowledge base and industry technology sharing platform application scenarios, and provide quantitative basis for metadata optimization, thereby effectively solving the technical problems of knowledge value misjudgment and incentive resource mismatch.

[0058] Further, after the optimized metadata is distributed to the local nodes and the aggregation node, the local node updates the knowledge contribution assessment rule, comprising: Analyzing the content of the global parameter field in the optimized metadata, and updating the weight calculation logic and index selection logic in the local adaptation rule according to the analysis result; After the optimized metadata is transmitted to each local node through the data distribution system, the rule analysis module of the local node analyzes the content of the global parameter field in the metadata. The global parameter field content includes the global weight coefficient and scene characteristic parameter optimized through federated aggregation and simulation verification. The analysis process includes field identification, data verification and parameter extraction, etc. For example, the local node of the customer service department analyzes the optimized metadata and identifies that the timeliness weight coefficient of emergency disposal type knowledge in the global parameter field has been increased from 0.6 to 0.8. Therefore, the weight calculation logic in the local adaptation rule is updated accordingly, and the weight proportion of the timeliness dimension is increased when calculating the knowledge contribution. At the same time, the index selection logic is updated according to the analysis result, for example, the average response time is added as an evaluation index, and redundant indexes that are not suitable for the current scene are removed. The update of the weight calculation logic is realized by modifying the weight calculation formula in the local evaluation algorithm, and the update of the index selection logic is completed by adjusting the index configuration table of the local evaluation system.

[0059] The aggregation node calibrates the local model parameters used for parameter aggregation, comprising: Load the global optimization adaptation parameters in the optimized metadata, and use these parameters to reinitialize the parameter matrix and weight coefficient of the local model; After the aggregation node receives the optimized metadata, the parameter calibration module loads the global optimization adaptation parameters in the metadata. These parameters include global evaluation parameters optimized through full-link adaptation effect simulation verification; The parameter calibration module uses the global optimization adaptation parameters to reinitialize the parameter matrix of the local model. The parameter matrix stores the correlation and influence coefficient between each evaluation dimension; Meanwhile, the weight coefficients of the local model are reinitialized to ensure that subsequent federated aggregation calculations are based on the latest parameter benchmark.

[0060] The data synchronization is performed through a secure transmission channel that uses an encryption protocol to encrypt and transmit the metadata packets; The secure transmission channel is established on the network connection between the local node and the aggregation node, and uses a transport layer security protocol or an equivalent encryption protocol to ensure data transmission security. During the data synchronization process, the system encrypts and encapsulates the metadata packets, including data encryption, integrity verification, and identity authentication. In the enterprise cross-department knowledge base scenario, the metadata packets are encrypted and encapsulated by the encryption protocol before being sent, and remain in an encrypted state during transmission. Only the target node can recover the original data by using the decryption key. This secure transmission mechanism ensures that the optimized metadata is not tampered with or leaked during distribution, thereby maintaining the data security and operational reliability of the system.

[0061] Through the above steps, the system realizes dynamic updating of the knowledge contribution evaluation rule and continuous optimization of the model parameters in the enterprise cross-department knowledge base and industry technology sharing platform application scenarios, forming a complete closed-loop optimization mechanism, and effectively improving the accuracy of knowledge value evaluation and the rationality of incentive resource allocation.

[0062] Further, as shown in Figure 2 The application also provides a knowledge sharing incentive device based on multi-dimensional contribution, which comprises: A metadata construction module is used to construct and distribute cross-scenario adaptive state metadata, which contains knowledge type attributes, scene demand characteristics, local adaptation results, and global optimization parameters. A local evaluation module is arranged in the local node and is used to perform local evaluation of knowledge contribution based on the metadata and real-time demand data of the local scene, and generate feedback data, which is written into the local adaptation results of the metadata. A federated aggregation module is arranged in the aggregation node and is used to collect metadata of multiple local nodes, perform federated aggregation on local evaluation related parameters in the metadata, and update global optimization parameters of the metadata. A simulation verification module is used to simulate and verify the full-link adaptation effect of knowledge cross-scenario flow based on the updated global optimization parameters, and optimize the metadata according to the verification result. An update calibration module is used to distribute the optimized metadata to the local node and the aggregation node, control the local node to update the knowledge contribution evaluation rule, and control the aggregation node to calibrate the local model parameters used for parameter aggregation.

[0063] Through the cooperation of the above modules, the device realizes dynamic evaluation of knowledge contribution and precise configuration of incentive resources in application scenarios such as enterprise cross-department knowledge base and industry technology sharing platform, and effectively solves the technical problem that the static evaluation system does not match the dynamic knowledge value.

[0064] Further, the application also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the knowledge sharing incentive method.

[0065] Finally, it should be noted that: the above only for the preferred embodiments of the present application, and not for limiting the present application, although the foregoing embodiments of the present application are described in detail, for those skilled in the art, it still can be modified, or part of the technical features of the equivalent replacement. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A knowledge sharing incentive method based on multidimensional contribution degree, characterized by, The method comprises the following steps: Constructing and distributing cross-scene adaptation state metadata, which contains knowledge type attributes, scene requirement characteristics, local adaptation results and global optimization parameters; Based on the metadata and real-time requirement data of the local scene, the local node performs local evaluation of knowledge contribution degree and generates feedback data, and writes the feedback data into the local adaptation results of the metadata; The aggregation node collects metadata of multiple local nodes, performs federated aggregation on local evaluation related parameters in the metadata, and updates global optimization parameters of the metadata; Based on the updated global optimization parameters, perform full-link adaptation effect simulation verification of knowledge cross-scene flow, and optimize the metadata according to the verification result; Distribute the optimized metadata to the local node and the aggregation node, and update the knowledge contribution degree evaluation rule of the local node, and calibrate the local model parameters used for parameter aggregation of the aggregation node.

2. The knowledge sharing motivation method based on multi-dimensional contribution degree according to claim 1, characterized in that, The method of constructing and distributing cross-scene adaptation state metadata comprises: Collect knowledge type accuracy score data through a user evaluation system, extract knowledge update frequency data through a version management system, and obtain operability feedback data through an operation behavior analysis system; Statistical technical depth requirement frequency through a research and development document analysis system, collect landing feasibility dependence degree data through a customer demand management system, and calculate fault correlation degree attention priority through a fault handling record system; Perform text labeling conversion and numerical normalization processing on the collected raw data, and uniformly map data of different dimensions to a preset numerical interval. 3.The multi-dimensional contribution-based knowledge sharing motivation method according to claim 1, wherein, The local node performs local evaluation of knowledge contribution degree based on the metadata and real-time requirement data of the local scene, which comprises: Real-time acquisition of fault duration data through a system monitoring module, reading of processing personnel skill level data through a personnel information database, calculation of fault influence range data through an influence evaluation algorithm, and extraction of project iteration stage data through a project management system; Perform data format unification, data validity verification and data feature extraction processing on real-time requirement data; After writing the local feedback data into the metadata, set a timer to automatically trigger metadata synchronization at a fixed period, and set a deviation monitoring mechanism to trigger metadata synchronization immediately when the evaluation deviation exceeds the threshold. 4.The method of claim 1, wherein, The local node performing local evaluation of knowledge contribution degree further comprises: Adjusting the weight distribution coefficient of each evaluation dimension according to the analysis result of real-time requirement data, and customizing the evaluation index set based on the scene feature recognition result; Quantitative scoring of the performance of knowledge in each dimension according to the set weight coefficient, multiplying each dimension score by the corresponding weight coefficient and then accumulating to obtain the comprehensive contribution degree score, wherein the adjustment of the weight coefficient is determined dynamically according to the priority index in the real-time requirement data, and the customization of the evaluation index set is dynamically configured according to the requirement type in the scene feature data. 5.The method of claim 1, wherein, The triggering conditions for the aggregation node to collect metadata of multiple local nodes comprise: Periodically count the proportion of the number of nodes that have received feedback data, and trigger aggregation when the proportion exceeds the set threshold; Monitor the time interval since the last parameter aggregation through a timer, and automatically trigger when the set time length is reached; The monitoring system continuously detects the adaptation deviation data of each core scene, and triggers immediately when the adaptation deviation of any core scene exceeds the set threshold for continuous multiple times. 6.The method of claim 1, wherein, The federated aggregation of the local evaluation related parameters in the metadata by the aggregation node includes: The key generation system creates a key pair required for homomorphic encryption, uses the public key to perform homomorphic encryption transformation on the received metadata field, and transmits the encrypted ciphertext data to the parameter aggregation server; After the parameter aggregation server completes the aggregation calculation, the private key is used to decrypt the aggregation result to restore the global optimization adaptation parameter in plaintext form; The parameter average algorithm based on data volume weighting is adopted, wherein the weight of each node is determined by the proportion of the size of its local data set to the size of the total data set. 7.The multi-dimensional contribution-based knowledge sharing motivation method according to claim 1, wherein, The full-link adaptation effect simulation verification of knowledge cross-scene flow based on the updated global optimization parameter includes: A cross-scene flow model is constructed according to the knowledge flow path to simulate the complete flow path of knowledge from the research and development scene to the test scene and then to the operation and maintenance scene; An adaptation accuracy index is calculated during the simulation process, which is obtained by comparing the difference between the simulation contribution and the expected contribution; A global balance index is also calculated, which is obtained by calculating the standard deviation of the adaptation deviation of each scene; According to the adaptation accuracy index in the simulation result, the scene demand feature weight is adjusted, and the correlation strength between the knowledge attribute and the scene is adjusted according to the global balance index. 8.The method of claim 1, wherein, After the optimized metadata is distributed to the local node and the aggregation node, the local node updates the knowledge contribution degree evaluation rule, which includes: Parsing the global parameter field content in the optimized metadata, and updating the weight calculation logic and index selection logic in the local adaptation rule according to the parsing result; The aggregation node calibrates the local model parameters for parameter aggregation, which includes: Loading the global optimization adaptation parameters in the optimized metadata, and using these parameters to re-initialize the parameter matrix and weight coefficient of the local model; Data synchronization is performed through a secure transmission channel, which uses an encryption protocol to encrypt and encapsulate the metadata packet and transmit it.

9. A knowledge sharing incentive device based on multidimensional contribution degree, characterized by, It includes: A metadata construction module for constructing and distributing cross-scene adaptation state metadata, which contains knowledge type attributes, scene demand features, local adaptation results, and global optimization parameters; A local evaluation module arranged in the local node for performing local evaluation of knowledge contribution degree based on the metadata and real-time demand data of the local scene, and generating feedback data, and writing the feedback data into the local adaptation result of the metadata; A federated aggregation module arranged in the aggregation node for collecting metadata of multiple local nodes, federated aggregating local evaluation related parameters in the metadata, and updating global optimization parameters of the metadata; A simulation verification module for full-link adaptation effect simulation verification of knowledge cross-scene flow based on the updated global optimization parameter, and optimizing the metadata according to the verification result; An update calibration module for distributing the optimized metadata to the local node and the aggregation node, controlling the local node to update the knowledge contribution degree evaluation rule, and controlling the aggregation node to calibrate the local model parameters for parameter aggregation.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program, when executed by a processor, implements the knowledge sharing incentive method according to any one of claims 1-8.

Images