Empirical knowledge data management method, device and equipment and readable storage medium

Abstract

The invention discloses an empirical knowledge data management method, device and equipment and a computer readable storage medium, and relates to the technical field of knowledge base management, and the method comprises the steps: carrying out the importance grading of empirical knowledge data, and obtaining first-level data, second-level data and third-level data; performing problem description clustering on the first-level data to obtain first clustered data, performing associated component clustering on the second-level data to obtain second clustered data, and performing responsible person clustering on the third-level data to obtain third clustered data; and performing grading arrangement on the first clustering data, the second clustering data and the third clustering data to obtain empirical knowledge arrangement data. According to the method, the experience and knowledge data are graded according to importance, and the corresponding aggregation strategy is set for each grade of experience and knowledge data, so that the experience and knowledge data can be structurally integrated, the experience repetition problem is greatly reduced, and the experience and knowledge data management efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of knowledge base management, and in particular to an experience knowledge data management method, device, equipment and computer readable storage medium. BACKGROUND

[0002] With the increase in the number of new vehicle models and modified vehicle models every year, new configurations, new functions and new applications continue to emerge on vehicles, and the amount of code in vehicle models increases year by year. This further causes the number of problems in the development stage of vehicles to be large in scale, which seriously hinders the listing cycle of vehicle models. Under the dual pressures of industrial upgrading and intensifying competition, quality control in the research and development stage is a key element that determines the core competitiveness of an enterprise. By designing a robust quality control to improve the quality of new product research and development, design risks and development costs can be reduced. In the traditional design prevention process, the establishment of an experience knowledge base is the first step. By integrating and utilizing historical experience knowledge data through the experience knowledge base, data support can be provided for design prevention quality control.

[0003] In related technologies, the process of constructing an experience knowledge base is usually as follows: an enterprise collects historical problem data reported from multiple departments such as research and development, production and after-sales, and then a full-time knowledge management officer processes the problems by experience zeroing, that is, by clearly defining the nature of the problem through root cause analysis, developing standardized solutions and completing verification closure; the experience data after zeroing is stored in a centralized relational database after standardized arrangement. When a research and development personnel faces a similar scenario during the design or problem handling stage, the database is usually queried through simple keyword search or basic condition filtering function to obtain historical cases as reference.

[0004] However, in enterprise knowledge base management, experience data is relatively scattered and lacks uniformity, experience data is distributed in different departments, systems or individuals, and existing experience knowledge bases usually do not consider how to effectively structure and integrate these experience knowledge data, resulting in frequent repeated problems, low efficiency of zeroing processing and standardization arrangement of experience knowledge. SUMMARY

[0005] The present application provides an experience knowledge data management method, device, equipment and computer readable storage medium, which can solve the technical problem that existing experience knowledge bases usually do not consider how to effectively structure and integrate these experience knowledge data, resulting in frequent repeated problems, low efficiency of zeroing processing and standardization arrangement of experience knowledge.

[0006] In a first aspect, an experience knowledge data management method is provided, comprising: grading the experience knowledge data to obtain first-level data, second-level data and third-level data; The first-class data is subjected to problem description clustering to obtain first clustering data, the second-class data is subjected to associated component clustering to obtain second clustering data, and the third-class data is subjected to responsibility person clustering to obtain third clustering data. The first clustering data, the second clustering data and the third clustering data are subjected to hierarchical arrangement to obtain experience knowledge arrangement data.

[0007] In combination with the first aspect, in an implementation, the importance of the experience knowledge data is graded to obtain the first-class data, the second-class data and the third-class data, including: The importance evaluation features of the experience knowledge data are subjected to quantitative processing to obtain quantitative evaluation features; The importance is graded according to the quantitative evaluation features to obtain importance grades; The experience knowledge data is divided into the first-class data, the second-class data and the third-class data according to the importance grades; The importance evaluation features include problem severity description text, time period and processing cost.

[0008] In combination with the first aspect, in an implementation, the importance evaluation features of the experience knowledge data are subjected to quantitative processing to obtain quantitative evaluation features, including: A plurality of grading labels are determined according to the problem severity description text and a preset grading rule, and the problem severity score is obtained by quantifying the grading labels according to a preset weighting scoring model; The time cost grade is determined according to the time period and a preset time threshold, and the time period score is determined according to the time cost grade; The processing cost score is determined according to the cost proportion corresponding to the processing cost; The problem severity score, the time period score and the processing cost score are merged to obtain the quantitative evaluation features.

[0009] In combination with the first aspect, in an implementation, the importance is graded according to the quantitative evaluation features to obtain importance grades, including: The importance grades are obtained by supervised learning importance grading prediction and K-fold cross-validation of the quantitative evaluation features.

[0010] In combination with the first aspect, in an implementation, the first-class data is subjected to problem description clustering to obtain first clustering data, including: The problem description features are obtained by text feature extraction of the problem description text of the first-class data; The first clustering data is obtained by similarity measurement clustering of the problem description features.

[0011] In combination with the first aspect, in an implementation, the second-class data is subjected to associated component clustering to obtain second clustering data, including: performing graph embedding feature extraction on the component multi-dimensional data of the second level to obtain component embedding features; performing hierarchical clustering on the component embedding features to obtain second clustering data; The component multi-dimensional data includes component identification information, component attribute features, and component fault information.

[0012] In combination with the first aspect, in an implementation, performing graph embedding feature extraction on the component multi-dimensional data of the second level to obtain component embedding features includes: constructing a plurality of graph nodes according to the component identification information; determining physical structure relationships and functional dependency relationships according to the component attribute features; determining fault association relationships according to the component fault information; constructing connection edges according to the physical structure relationships, the functional dependency relationships, and the fault association relationships; constructing component graph data according to the graph nodes and the connection edges, and performing graph embedding feature extraction on the component graph data to obtain the component embedding features.

[0013] In combination with the first aspect, in an implementation, performing responsibility person clustering on the third level data to obtain third clustering data includes: performing behavior feature extraction on responsibility person data of the third level data to obtain responsibility person behavior features; performing spectral clustering on the responsibility person behavior features to obtain the third clustering data.

[0014] In combination with the first aspect, in an implementation, performing behavior feature extraction on responsibility person data of the third level data to obtain responsibility person behavior features includes: performing regular expression extraction and named entity recognition on the responsibility person data to obtain responsibility person information; performing structured processing and behavior feature extraction on the responsibility person data to obtain preliminary behavior features, and performing time series analysis on the preliminary behavior features to obtain the behavior features; constructing responsibility person behavior features according to the responsibility person information and the behavior features.

[0015] In combination with the first aspect, in an implementation, performing hierarchical arrangement on the first clustering data, the second clustering data, and the third clustering data to obtain experience knowledge arrangement data includes: performing experience knowledge arrangement review on the first clustering data, the second clustering data, and the third clustering data based on a preset review strategy to obtain the experience knowledge arrangement data.

[0016] In combination with the first aspect, in an implementation, the preset review strategy includes a company-level review strategy, a department-level review strategy, and a division-level review strategy. The company-level review strategy corresponds to the first cluster of data, resulting in company-level experience items; The department-level review strategy corresponds to the second cluster of data, resulting in department-level experience items; The department-level review strategy corresponds to the third cluster data, resulting in department-level experience items.

[0017] In conjunction with the first aspect, in one implementation, company-level experience entries are retrieved using an inverted index and a vector database strategy; Department-level experience entries are retrieved using a distributed caching strategy that prioritizes frequently accessed components. Department-level experience entries are retrieved using a lightweight engine-based responsible person query strategy.

[0018] Secondly, embodiments of this application provide an experience knowledge data management device, including: Importance grading unit, used to grade the importance of experiential knowledge data to obtain first-level data, second-level data and third-level data; The hierarchical clustering unit is used to cluster the first-level data by problem description to obtain the first cluster data, the second-level data by related components to obtain the second cluster data, and the third-level data by responsible persons to obtain the third cluster data. The review and evaluation unit is used to hierarchically organize the first, second, and third cluster data to obtain empirical knowledge organized data.

[0019] Thirdly, embodiments of this application provide an experience knowledge data management device, which includes a processor, a memory, and an experience knowledge data management program stored in the memory and executable by the processor. When the experience knowledge data management program is executed by the processor, it implements the steps of the experience knowledge data management method as described above.

[0020] Fourthly, embodiments of this application provide a computer-readable storage medium storing an experience knowledge data management program, wherein when the experience knowledge data management program is executed by a processor, it implements the steps of the experience knowledge data management method as described above.

[0021] The beneficial effects of the technical solutions provided in this application include at least the following: By classifying experiential knowledge data according to its importance and setting corresponding aggregation strategies for each level of experiential knowledge data, experiential knowledge data can be integrated in a structured manner, greatly reducing the problem of duplicate experience and improving the efficiency of experiential knowledge data management.

[0022] Furthermore, for first-level experiential knowledge, clustering is performed through problem descriptions to focus on common overall issues for knowledge integration; for second-level experiential knowledge, clustering is performed through related components to focus on common issues specific to the business; and for third-level experiential knowledge, clustering is performed through responsible personnel to focus on similar individual operating patterns for knowledge integration. This invention, by setting corresponding clustering strategies based on the focus of different levels of experiential knowledge, can further ensure the effectiveness of experiential knowledge integration. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating an embodiment of the experience knowledge data management method of this application; Figure 2 For this application Figure 1 A detailed flowchart of step S101; Figure 3 For this application Figure 2 A detailed flowchart of step S201; Figure 4 This is a schematic diagram illustrating the detailed process of clustering the problem description in this application; Figure 5 This is a schematic diagram illustrating the detailed process of clustering related components in an embodiment of this application; Figure 6 This is a detailed flowchart of step S501 in this application; Figure 7 This is a flowchart illustrating the clustering of responsible parties in an embodiment of this application. Figure 8 This is a detailed flowchart of step S701 in this application; Figure 9 This is a schematic diagram of the functional modules of an embodiment of the experience knowledge data management device of this application; Figure 10 This is a schematic diagram of the hardware structure of the experience knowledge data management device involved in the embodiments of this application. Detailed Implementation

[0024] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0025] The terms "comprising" and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. The terms "first," "second," and "third," etc., are used to distinguish different objects, etc., and do not indicate a sequence, nor do they limit "first," "second," and "third" to different types.

[0026] In the description of the embodiments of this application, terms such as "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a concrete manner.

[0027] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more.

[0028] In some processes described in the embodiments of this application, multiple operations or steps are included in a specific order. However, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or they may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.

[0029] First, some of the technical terms used in this application will be explained to help those skilled in the art understand this application.

[0030] Zeroing out experience refers to the complete process in R&D and quality management of systematically analyzing the root causes of historical problems, developing standardized solutions, completing the verification loop, and ultimately transforming the problem-solving process into a structured, reusable knowledge asset. To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0031] In a first aspect, embodiments of this application provide a method for managing experiential knowledge data.

[0032] In one embodiment, reference is made to Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the experience knowledge data management method of this application. Figure 1 As shown, the methods for managing experiential knowledge data include: S101. The importance of the experiential knowledge data is classified into first-level data, second-level data, and third-level data. S102. Cluster the first-level data by problem description to obtain the first cluster data, cluster the second-level data by related components to obtain the second cluster data, and cluster the third-level data by responsible persons to obtain the third cluster data. S103. The first cluster data, the second cluster data, and the third cluster data are sorted and hierarchically to obtain the empirical knowledge sorting data.

[0033] Specifically, the implementation plan first classifies the experiential knowledge data according to its importance, dividing it into three levels based on dimensions such as problem severity, resolution timeframe, and resolution cost. Different clustering methods are then set for each level of experiential knowledge, depending on the focus of the analysis.

[0034] For Level 1 experiential knowledge data, this level is relatively important, typically having a wide impact and high processing costs. Considering the presence of some broadly impactful issues within Level 1 experiential knowledge data, which may originate from multiple departments, clustering and integration of these issues is necessary. This embodiment uses problem description as a basis for clustering, focusing on common issues at the overall level to perform knowledge integration and clustering for Level 1 data.

[0035] For the second level of experiential knowledge data, its importance is lower than the previous level, having a certain scope of influence but less than the first level. This level of experiential knowledge data typically corresponds to specific business-level problems, and the common problems often manifest as correlations within components. For example, an unreasonable design of a component may not only cause problems with the component itself but also lead to problems with related components, thus generating related experiential knowledge data. Similarly, a problem in a business process may cause problems with multiple related components within that process, generating multiple corresponding experiential knowledge data. To address this, the embodiment uses a clustering approach with related components to focus on common business problems and perform knowledge integration and clustering at the second level.

[0036] For Level 3 experiential knowledge data, the importance of this level is relatively low, typically corresponding to problems in individual operational behaviors. For Level 3 experiential knowledge data, it can be explored by analyzing the behavioral characteristics of those responsible and identifying similar operational patterns. In this example, the implementation uses a responsible person clustering approach to focus on common problems in individual operations and perform knowledge integration and clustering at Level 3.

[0037] Finally, for the clustered data at the three levels, the implementation example employed corresponding hierarchical organization strategies and performed experience zeroing. The first level corresponds to a company-level review strategy, the second level to a department-level review strategy, and the third level to a section-level review strategy. After completing the hierarchical organization strategy and experience zeroing, the resulting organized experience knowledge data was used to construct an experience knowledge database, effectively improving the efficiency of experience knowledge data management.

[0038] In this embodiment, by classifying experiential knowledge data according to importance and setting corresponding aggregation strategies for each level of experiential knowledge data, experiential knowledge data can be integrated in a structured manner, greatly reducing the problem of experience duplication and improving the efficiency of experiential knowledge data management.

[0039] Furthermore, for first-level experiential knowledge, clustering is performed through problem descriptions to focus on common overall issues for knowledge integration; for second-level experiential knowledge, clustering is performed through related components to focus on common issues specific to the business; and for third-level experiential knowledge, clustering is performed through responsible personnel to focus on similar individual operating patterns for knowledge integration. This invention, by setting corresponding clustering strategies based on the focus of different levels of experiential knowledge, can further ensure the effectiveness of experiential knowledge integration.

[0040] Furthermore, in one embodiment, reference is made to Figure 2 , Figure 2 For this application Figure 1 A detailed flowchart of step S101 is shown below. Figure 2 As shown, the importance of experiential knowledge data is graded into first-level data, second-level data, and third-level data, including: S201. The importance evaluation features of experiential knowledge data are quantified to obtain quantitative evaluation features. S202. Based on the quantitative evaluation characteristics, the importance level is predicted and classified to obtain the importance level; S203. Based on the importance level, experiential knowledge data is divided into first-level data, second-level data, and third-level data; The importance assessment features include the problem severity description text, time frame, and processing cost.

[0041] Specifically, the data collected for the acquisition of experience data and the construction of the knowledge base in the embodiments includes, but is not limited to, problem descriptions, information of responsible persons, data of related components, causes of problems, solutions, relevant experience, severity of problems, time period and / or processing costs.

[0042] The information on responsible parties includes data such as their behavioral characteristics and operational records. Related component data includes the component numbers, names, and fault information related to the problem. Relevant experience refers to historical records of similar problems and their solutions. Problem severity is a quantitative indicator based on assessments of the scope of impact and the degree of loss.

[0043] For collecting experience data, the implementation example utilizes distributed web crawling technology and API interfaces to automatically collect data from internal enterprise systems such as ERP (Enterprise Resource Planning), CRM (Customer Relationship Management), and MES (Manufacturing Execution System). It supports structured (e.g., database records), semi-structured (e.g., log files), and unstructured data (e.g., technical documents and emails). A distributed database is used to store large-scale experience data, ensuring high concurrency access and data scalability. ETL tools are used to perform preliminary cleaning of the raw data, removing duplicates, missing data, or invalid data to ensure data integrity and consistency.

[0044] In the hierarchical classification of empirical data, the implementation example uses importance evaluation features that have been quantified to perform hierarchical classification, and adopts a multi-model automatic hierarchical classification and cross-validation approach.

[0045] The quantitative evaluation features include a description of the problem severity, a time frame, and a processing cost. The problem severity description includes text describing business impact and security risks, and is quantified using a weighted evaluation model. The time frame is determined by setting time thresholds to prioritize the time period from problem discovery to resolution. The processing cost is prioritized based on its cost percentage (e.g., as a percentage of the vehicle model's annual budget).

[0046] The classification process implementation uses multiple supervised learning algorithm models combined with K-fold cross-validation, such as the random forest algorithm and the XGBoost algorithm, to automatically classify experience items.

[0047] In this embodiment, the empirical data is classified according to its importance. This ensures that highly important data is given high priority and also divides the empirical data into groups with different characteristics, providing a basis for setting corresponding aggregation strategies in the future.

[0048] Furthermore, in one embodiment, reference is made to Figure 3 , Figure 3 For this application Figure 2The detailed process diagram of step S201 shows the quantitative evaluation features obtained by quantifying the importance evaluation features of experiential knowledge data, including: S301. Determine several grading labels based on the problem severity description text and preset grading rules, and quantify the grading labels according to the preset weighted scoring model to obtain the problem severity score; S302. Determine the time cost level based on the time period and the preset time threshold, and determine the time period score based on the time cost level. S303. Determine the processing cost score based on the cost percentage corresponding to the processing cost. S304. The combined problem severity score, time cycle score, and processing cost score yield quantitative evaluation characteristics.

[0049] Specifically, in the process of quantifying importance evaluation features, the description text for problem severity includes various types such as time cost levels, business impact (e.g., customer complaints or production line shutdowns) and safety risks (e.g., production safety risks or product safety risks). In this embodiment, the description text is matched with pre-set grading rules to obtain several corresponding grading labels that include both problem type and importance level. Considering that different problem types have different priorities in importance assessment, the embodiment uses a pre-set weighted scoring model to weight and quantify each grading label and then merges them to obtain the problem severity score.

[0050] Regarding time cycles, considering that the actual time from problem discovery to resolution may be affected by the work cycle—for example, whether an event is discovered and raised at night or the next morning may require a working day to process, but the actual time taken will differ—this embodiment determines the time cost level by setting multiple time thresholds to divide the time cycle into corresponding time intervals, and then determines the corresponding time cycle score based on the time cost level.

[0051] For processing costs, the corresponding processing cost score is calculated based on the cost percentage, such as the percentage of the vehicle's annual budget.

[0052] The obtained scores are then combined to obtain the importance evaluation features.

[0053] In this embodiment, different scoring methods are adaptively adopted for different features to improve the accuracy of the scoring, thereby improving the accuracy of subsequent experiential knowledge importance classification.

[0054] Furthermore, in one embodiment, the importance level is obtained by predicting the importance level based on quantitative evaluation characteristics, including: The importance level is obtained by supervised learning to predict the importance level of quantitative evaluation features and K-fold cross-validation.

[0055] Specifically, the classification process in this embodiment employs multiple supervised learning algorithms, such as Random Forest and XGBoost, to classify the experience items. The embodiment uses the model with the highest accuracy after initial training as the default output, then combines the results from other models and evaluates the accuracy and stability of the classification through K-fold cross-validation. Adjustments are made to the classification results to ensure their scientific validity and reliability. The training datasets for each model are derived from historical experience data, and the models are periodically updated to adapt to new problem scenarios.

[0056] Furthermore, in one embodiment, Figure 4 This is a schematic diagram illustrating the detailed process of clustering the problem description in this application, such as... Figure 4 As shown, problem description clustering is performed on the first-level data to obtain the first cluster data, which includes: S401. Extract text features from the problem description text of the first-level data to obtain problem description features; S402. Similarity measurement and clustering of the problem description features are used to obtain the first cluster data.

[0057] Specifically, for the highest priority, first-level experience items, the implementation plan uses cluster analysis based on problem descriptions to focus on common problems. A company-wide cross-functional team is then formed to organize and refine the experience, and a company-wide review process is conducted to accumulate high-quality knowledge.

[0058] During the clustering process, the first step in the implementation example is data cleaning. Natural language processing techniques are used to preprocess the problem description text, such as using the BERT model. This includes word segmentation, stop word removal, and stemming to eliminate noisy data.

[0059] Then, the text is converted into a vector representation using TF-IDF term frequency-inverse document frequency or Word2Vec models to extract semantic features of the problem description. Finally, the implementation uses K-means or DBSCAN algorithms to cluster the problem descriptions, identify similar problem patterns, and combines cosine similarity calculation to achieve intelligent matching of new problems with historical experience. This achieves the first level of clustering of experiential knowledge.

[0060] In this embodiment, clustering the problem descriptions using natural language processing techniques can effectively cluster common problems at the overall level for the first-level experience items.

[0061] Furthermore, in one embodiment, Figure 5 This is a schematic diagram illustrating the detailed process of clustering related components in an embodiment of this application, as shown below. Figure 5 As shown, clustering the second-level data into related components yields the second-cluster data, which includes: S501. Extract component embedding features from the multidimensional data of components at the second level using graph embedding features; S502. Perform hierarchical clustering on the component embedding features to obtain the second clustering data; The multidimensional data of the components includes component identification information, component attribute characteristics, and component fault information.

[0062] Furthermore, in one embodiment, Figure 6 This is a detailed flowchart of step S501 in this application, as shown below. Figure 6 As shown, graph embedding feature extraction is performed on the second-level component multidimensional data to obtain component embedding features, including: S601. Construct several graph nodes based on component identification information; S602. Determine the physical structural relationships and functional dependencies based on the component attribute characteristics; S603. Determine the fault correlation based on component fault information; S604. Construct connection edges based on physical structure relationships, functional dependencies, and fault association relationships; S605. Construct component graph data based on graph nodes and connecting edges, and extract component embedding features from the component graph data.

[0063] Specifically, for the second-level experience items, considering that the experience knowledge data at this level usually corresponds to specific business-level problems, the common problems often manifest as correlations in related components. Therefore, this embodiment performs clustering operations on the second-level data by clustering related components.

[0064] Specifically, for data related to components, the embodiment first uses rule matching and anomaly detection algorithms to remove irrelevant or erroneous data to ensure data quality.

[0065] Then, multi-dimensional features related to the components are extracted, including component identification information, component attribute features, and component fault information. Graph embedding technology is then used to represent the relationships between components. Component identification information includes the component name and number; component attribute features include component parameters and structural and functional connection attributes between components; and component fault information includes fault type, fault frequency, and maintenance cycle.

[0066] It should be noted that the extraction of component fault information in the embodiment takes into account that there may be multiple empirical entries for the same component. In order to avoid mis-clustering of multiple empirical entries for the same component, in addition to component identification information and component attribute features, multidimensional features also need to extract component fault information.

[0067] Based on the extracted multidimensional data features of the components, the embodiment constructs each graph node according to the tagging information. Then, physical structural relationships and functional dependencies are constructed based on the component attribute characteristics, and fault associations are determined based on component fault information. After removing weak connections with excessively low weights, edges with comprehensive weights are merged based on physical structural relationships, functional dependencies, and fault associations. Based on the graph nodes and edges, the component graph data is constructed.

[0068] Then, for the obtained component graph data, to facilitate similarity recognition of graph data, the embodiment extracts component embedding features through graph embedding feature extraction. Next, a hierarchical clustering algorithm is used to cluster related components based on their component embedding features, identifying common issues related to the components. Finally, a graph neural network is used to achieve intelligent matching between components and experience.

[0069] In this embodiment, for the experience items at the second level, graph embedding feature extraction and hierarchical clustering can be used to effectively cluster the experience items at the second level by focusing on common problems in specific businesses from the perspective of component relevance.

[0070] Furthermore, in one embodiment, Figure 7 This is a flowchart illustrating the clustering of responsible parties in an embodiment of this application, as shown below. Figure 7 As shown, the third-level data is clustered by responsible persons to obtain the third cluster data, which includes: S701. Extract behavioral characteristics of the responsible persons from the third-level data; S702. Perform spectral clustering on the behavioral characteristics of the responsible persons to obtain the third cluster data.

[0071] Furthermore, in one embodiment, Figure 8 This is a detailed flowchart of step S701 in this application, as shown below. Figure 8 As shown, behavioral characteristics of the responsible persons are extracted from the third-level data, including: S801. Extract the responsible person information by performing regular expression extraction and named entity recognition on the responsible person data; S802. The data of the responsible persons is structured and behavioral features are extracted to obtain preliminary behavioral features. The preliminary behavioral features are then subjected to time series analysis to obtain the behavioral features. S803. Construct the behavioral characteristics of the responsible person based on the information and behavioral characteristics of the responsible person.

[0072] Specifically, for the experience items at the third level, considering that the experience knowledge at this level usually corresponds to problems in individual operational behavior, the responsible persons for common problems often have similar operational behaviors. Therefore, this embodiment uses a responsible person clustering method to cluster the third-level data.

[0073] Specifically, for data related to responsible persons, such as operation records and reports, responsible person information is extracted using regular expressions and named entities. Then, the embodiment performs structured processing on the responsible person data, identifies and extracts behavioral data, such as operation frequency and problem-solving success rate, and combines this behavioral data with time-series analysis to obtain the behavioral characteristics of each responsible person, incorporating time-series information.

[0074] Meanwhile, considering that identifying the similarity of responsible persons is a non-linear task, and given that spectral clustering is more suitable for identifying complex non-linear similarity relationships, this embodiment applies spectral clustering to the extracted behavioral features to group responsible persons and identify similar operational patterns. Furthermore, a collaborative filtering algorithm is used to match responsible persons with experience.

[0075] In this embodiment, for the experience items at the third level, through behavioral feature extraction and spectral clustering, knowledge integration can be carried out by focusing on similar operation patterns of individuals from the perspective of responsible persons' behavioral patterns, thereby effectively clustering the experience items at the third level.

[0076] Furthermore, in one embodiment, the first cluster data, the second cluster data, and the third cluster data are hierarchically organized to obtain empirical knowledge organized data, including: Based on the preset review strategy, empirical knowledge was organized and reviewed on the first, second, and third cluster data to obtain empirical knowledge organized data.

[0077] Furthermore, in one embodiment, the preset review strategies include company-level review strategies, department-level review strategies, and section-level review strategies. The company-level review strategy corresponds to the first cluster of data, resulting in company-level experience items; The department-level review strategy corresponds to the second cluster of data, resulting in department-level experience items; The department-level review strategy corresponds to the third cluster data, resulting in department-level experience items.

[0078] Specifically, based on the severity of the problem, the timeframe for resolution, and the cost, the experience data is categorized into first, second, and third levels. For each level of experience knowledge, the implementation example sets different review strategies to perform experience reset, optimizing resource allocation and processing efficiency. The review strategies include company-level, department-level, and section-level reviews. For the highest priority Level 1 experience knowledge, a company-wide cross-functional team is formed to conduct company-wide review, responsible for compiling clustering results, analyzing common problems, and developing standardized solutions. For the medium priority Level 2 experience knowledge, a department-level cross-functional team is formed to conduct department-level review, focusing on specific business areas, compiling component-related experience data, and extracting common solutions. For the low priority Level 3 experience knowledge, department heads organize experience compilation, focusing on individual operational optimization.

[0079] In this embodiment, by adopting a hierarchical review strategy corresponding to the hierarchical classification of experience and knowledge, cross-functional teams at the company, department, and section levels are formed to carry out experience sorting and zeroing out. This can ensure the rational allocation of resources, shorten the problem-solving cycle, and improve overall efficiency.

[0080] Furthermore, in one embodiment, company-level experience entries are retrieved using an inverted index and a vector database strategy; Department-level experience entries are retrieved using a distributed caching strategy that prioritizes frequently accessed components. Department-level experience entries are retrieved using a lightweight engine-based responsible person query strategy.

[0081] Specifically, in order to efficiently access experiential knowledge at different levels, the implementation also provides query and search strategies for each level that are adapted to the clustering strategy.

[0082] For first-level experiential knowledge, the implementation example uses an inverted index and a vector database to facilitate fast queries based on problem descriptions, with response times controlled in the millisecond range. The inverted index improves the response speed for recent problem searches, while the vector database enhances retrieval efficiency.

[0083] For the second level of experience knowledge, the embodiment improves retrieval speed by storing frequently accessed component experience data in a distributed cache. Storing frequently accessed component experience data in a distributed cache can effectively improve the query speed of high-frequency experience data.

[0084] For Level 3 experiential knowledge, the implementation example supports quick searches related to responsible parties through a lightweight search engine. By supporting quick searches for responsible parties, user search efficiency can be improved at the retrieval strategy level.

[0085] In this embodiment, by setting a query search strategy adapted to the clustering strategy, the efficiency of retrieving and calling experience data is significantly improved.

[0086] Secondly, embodiments of this application also provide an experience knowledge data management device.

[0087] In one embodiment, reference is made to Figure 9 , Figure 9This is a schematic diagram of the functional modules of an embodiment of the experience knowledge data management device of this application. Figure 9 As shown, the experience knowledge data management device includes: Importance grading unit 901 is used to grade the importance of experiential knowledge data to obtain first-level data, second-level data, and third-level data; The hierarchical clustering unit 902 is used to cluster the first-level data by problem description to obtain the first cluster data, to cluster the second-level data by related components to obtain the second cluster data, and to cluster the third-level data by responsible persons to obtain the third cluster data. The review and evaluation unit 903 is used to hierarchically organize the first, second, and third cluster data to obtain the empirical knowledge organized data.

[0088] The functions of each module in the aforementioned experience knowledge data management device correspond to the steps in the aforementioned experience knowledge data management method embodiment, and their functions and implementation processes will not be described in detail here.

[0089] Thirdly, embodiments of this application provide an experience knowledge data management device, which can be a personal computer (PC), laptop computer, server, or other device with data processing capabilities.

[0090] Reference Figure 10 , Figure 10 This is a schematic diagram of the hardware structure of the experience knowledge data management device involved in the embodiments of this application. In the embodiments of this application, the experience knowledge data management device may include a processor, a memory, a communication interface, and a communication bus.

[0091] The communication bus can be of any type and is used to interconnect the processor, memory, and communication interface.

[0092] In one embodiment, when the processor executes an experiential knowledge data management program stored in memory, it may perform the following steps: The importance of experiential knowledge data is classified into first-level data, second-level data, and third-level data. The first-level data is clustered by problem description to obtain the first cluster data; the second-level data is clustered by related components to obtain the second cluster data; and the third-level data is clustered by responsible persons to obtain the third cluster data. The empirical knowledge data is obtained by hierarchically organizing the first, second, and third cluster data.

[0093] The communication interface includes input / output (I / O) interfaces, physical interfaces, and logical interfaces used for interconnecting devices within the experience knowledge data management device, as well as interfaces used for interconnecting the experience knowledge data management device with other devices (such as other computing devices or user equipment). Physical interfaces can be Ethernet interfaces, fiber optic interfaces, ATM interfaces, etc.; user equipment can be displays, keyboards, etc.

[0094] Memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0095] The processor can be a general-purpose processor, which can call the experience knowledge data management program stored in memory and execute the experience knowledge data management method provided in the embodiments of this application. For example, the general-purpose processor can be a central processing unit (CPU). The method executed when the experience knowledge data management program is called can be referred to in various embodiments of the experience knowledge data management method of this application, and will not be repeated here.

[0096] Those skilled in the art will understand that Figure 10 The hardware structure shown does not constitute a limitation of this application and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0097] Fourthly, embodiments of this application also provide a readable storage medium.

[0098] This application has an experiential knowledge data management program stored on a readable storage medium, wherein when the experiential knowledge data management program is executed by a processor, it implements the following steps of an experiential knowledge data management method: The importance of experiential knowledge data is classified into first-level data, second-level data, and third-level data. The first-level data is clustered by problem description to obtain the first cluster data; the second-level data is clustered by related components to obtain the second cluster data; and the third-level data is clustered by responsible persons to obtain the third cluster data. The empirical knowledge data is obtained by hierarchically organizing the first, second, and third cluster data.

[0099] The method implemented when the experience knowledge data management program is executed can be referred to in various embodiments of the experience knowledge data management method of this application, and will not be repeated here.

[0100] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device to execute the methods of the various embodiments of this application.

[0101] The foregoing has provided a detailed description of the experience knowledge data management method, apparatus, device, and readable storage medium of the present invention. Specific examples have been used to illustrate the principles and implementation schemes of the present invention. The descriptions of the embodiments above are merely to help understand the method and core ideas of the present invention and are not intended to limit the patent scope of this application. Furthermore, those skilled in the art should understand that any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, should be included within the patent protection scope of this application.

Claims

1. An empirical knowledge data management method characterized by, The method comprises the following steps: grading the experience knowledge data to obtain first-level data, second-level data and third-level data; performing problem description clustering on the first-level data to obtain first clustering data, performing associated component clustering on the second-level data to obtain second clustering data, and performing responsibility person clustering on the third-level data to obtain third clustering data; grading and sorting the first clustering data, the second clustering data and the third clustering data to obtain experience knowledge sorting data.

2. The empirical knowledge data management method according to claim 1, characterized by, The method comprises the following steps: grading the experience knowledge data to obtain first-level data, second-level data and third-level data; quantitatively processing the importance evaluation features of the experience knowledge data to obtain quantitatively processed evaluation features; performing importance grading prediction according to the quantitatively processed evaluation features to obtain importance levels; dividing the experience knowledge data into first-level data, second-level data and third-level data according to the importance levels; 3. The empirical knowledge data management method of claim 2, wherein, wherein the importance evaluation features include problem severity description text, time period and processing cost. The method comprises the following steps: determining a plurality of grading labels according to the problem severity description text and a preset grading rule, and quantitatively processing the grading labels to obtain problem severity scores according to a preset weighting scoring model; determining a time cost level according to the time period and a preset time threshold, and determining a time period score according to the time cost level; determining a processing cost score according to the cost proportion corresponding to the processing cost; 4. The empirical knowledge data management method of claim 2, wherein, merging the problem severity scores, the time period score and the processing cost score to obtain quantitatively processed evaluation features. The method comprises the following steps:

5. The empirical knowledge data management method of claim 1, wherein, performing supervised learning importance grading prediction and K-fold cross-validation on the quantitatively processed evaluation features to obtain importance levels. The method comprises the following steps: extracting text features from the problem description text of the first-level data to obtain problem description features; 6. The empirical knowledge data management method of claim 1, wherein, performing similarity measurement clustering on the problem description features to obtain first clustering data. The method comprises the following steps: extracting graph embedding features from the component multi-dimensional data of the second-level data to obtain component embedding features; performing hierarchical clustering on the component embedding features to obtain second clustering data; 7. The empirical knowledge data management method of claim 6, wherein, wherein the component multi-dimensional data includes component identification information, component attribute features and component fault information. The method comprises the following steps: constructing a plurality of graph nodes according to the component identification information; determining physical structure relationships and functional dependency relationships according to the component attribute features; determining fault association relationships according to the component fault information; constructing connection edges according to the physical structure relationships, the functional dependency relationships and the fault association relationships; constructing component graph data according to the graph nodes and the connection edges, and extracting graph embedding features from the component graph data to obtain component embedding features.

8. The empirical knowledge data management method of claim 1, wherein, The responsibility person clustering on the third-level data obtains third clustering data, and the method comprises the following steps of: behavior feature extraction is performed on the responsibility person data of the third-level data to obtain responsibility person behavior features; spectrum clustering is performed on the responsibility person behavior features to obtain third clustering data.

9. The empirical knowledge data management method of claim 8, wherein, The behavior feature extraction is performed on the responsibility person data of the third-level data to obtain responsibility person behavior features, and the method comprises the following steps of: regular expression extraction and named entity recognition are performed on the responsibility person data to obtain responsibility person information; structured processing and behavior feature extraction are performed on the responsibility person data to obtain preliminary behavior features, and time series analysis is performed on the preliminary behavior features to obtain behavior features; responsibility person behavior features are constructed according to the responsibility person information and the behavior features.

10. The empirical knowledge data management method of claim 1, wherein, The hierarchical arrangement of the first clustering data, the second clustering data and the third clustering data obtains experience knowledge arrangement data, and the method comprises the following steps of: experience knowledge arrangement review is performed on the first clustering data, the second clustering data and the third clustering data based on a preset review strategy to obtain experience knowledge arrangement data.

11. The empirical knowledge data management method of claim 10, wherein, The preset review strategy comprises a company-level review strategy, a department-level review strategy and a section-level review strategy; the company-level review strategy corresponds to the first clustering data to obtain company-level experience entries; the department-level review strategy corresponds to the second clustering data to obtain department-level experience entries; the section-level review strategy corresponds to the third clustering data to obtain section-level experience entries.

12. The empirical knowledge data management method of claim 11, wherein, The company-level experience entries are searched through an inverted index and a vector database strategy; the department-level experience entries are searched through a distributed cache high-frequency access component strategy; the section-level experience entries are searched through a lightweight engine responsibility person query strategy.

13. An empirical knowledge data management apparatus characterized by comprising: The method comprises the following steps of: an importance grading unit is configured to grade the experience knowledge data according to importance to obtain first-level data, second-level data and third-level data; a hierarchical clustering unit is configured to perform question description clustering on the first-level data to obtain first clustering data, perform associated component clustering on the second-level data to obtain second clustering data, and perform responsibility person clustering on the third-level data to obtain third clustering data; an arrangement review unit is configured to perform hierarchical arrangement on the first clustering data, the second clustering data and the third clustering data to obtain experience knowledge arrangement data.

14. An experience knowledge data management apparatus characterized by comprising: The experience knowledge data management device comprises a processor, a memory and an experience knowledge data management program stored on the memory and executable by the processor, wherein when the experience knowledge data management program is executed by the processor, the steps of the experience knowledge data management method in any one of claims 1 to 12 are implemented.

15. A computer-readable storage medium, characterized in that, The computer readable storage medium stores an experience knowledge data management program, wherein when the experience knowledge data management program is executed by the processor, the steps of the experience knowledge data management method in any one of claims 1 to 12 are implemented.

Images