Data asset intelligent exploration method and system based on knowledge graph

Abstract

The invention relates to the technical field of data asset exploration, in particular to an intelligent data asset exploration method and system based on a knowledge graph, and the method comprises the steps: obtaining enterprise data assets, carrying out the dynamic annotation, and generating an annotation data set; performing semantic analysis based on the annotation data set to obtain a semantic association graph; constructing a knowledge graph model, calculating node weights, association path lengths and path contribution degrees, and generating a simulation association distribution graph; segmenting the semantic association graph through the node density and the relationship strength to obtain an actual association distribution graph; and adjusting the knowledge graph model based on the actual and simulated association distribution histogram. According to the invention, the intelligent level and accuracy of data asset exploration can be improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of data asset exploration, and specifically relates to a data asset intelligent exploration method and system based on a knowledge graph. BACKGROUND

[0002] With the increasing importance of data assets in enterprise operations and decision-making, how to efficiently manage and explore data assets has become a key issue. Existing data asset management and exploration methods rely heavily on traditional data processing methods, lacking intelligent and automated capabilities, making it difficult to meet the governance needs of complex data environments in modern enterprises. For example, a patent with publication number CN114022270B proposes a data processing method based on an asset knowledge graph, which generates a feature vector for training and prediction of a risk prediction model by constructing an asset knowledge graph and combining the importance data of asset entities. However, this technical solution mainly focuses on asset risk prediction in the financial field and fails to fully cover the needs of data asset exploration. In addition, it does not fully consider the dynamic relationships and contextual information between data assets in the knowledge graph construction process, which may result in insufficient depth and accuracy of the exploration results. At the same time, this solution has low processing efficiency for large-scale heterogeneous data, making it difficult to adapt to real-time exploration needs in complex data environments.

[0003] Another patent with publication number CN118886647B proposes a data asset governance method based on standardized processes, which extracts data governance link descriptions and matches them with standard files to build a standard data governance rule engine to achieve standardized management of data assets. However, this technical solution mainly focuses on the standardization and process of data governance, lacking the ability of intelligent exploration, and unable to automatically identify implicit relationships and potential value between data assets. In addition, this method has limited support for unstructured data, and may have compatibility issues when dealing with diverse data assets. At the same time, this solution has a superficial application of knowledge graphs and fails to fully utilize the semantic reasoning capabilities of knowledge graphs to improve the intelligent level of data asset exploration.

[0004] The above problems show that existing data asset exploration technologies still have certain deficiencies in terms of intelligent level, dynamic relationship mining, heterogeneous data compatibility, and real-time processing capabilities. Especially when facing complex and diverse enterprise data environments, existing technologies are difficult to achieve comprehensive, dynamic, and intelligent exploration of data assets. Therefore, there is an urgent need for a solution that combines advanced knowledge graph technology and large model AI algorithms to improve the efficiency and accuracy of data asset exploration and meet the needs of modern enterprises for efficient data asset management. The present application is based on this background and aims to solve the key problems in existing technologies by introducing intelligent exploration methods and systems, providing enterprises with more comprehensive, dynamic, and intelligent data asset exploration capabilities. Summary of the Invention

[0005] The present invention provides a method and system for intelligent exploration of data assets based on a knowledge graph, the main purpose of which is to improve the intelligence level and accuracy of data asset exploration. To achieve the above purpose, the method and system for intelligent exploration of data assets based on a knowledge graph provided by the present invention include: acquiring enterprise data assets, setting a data type range, a time range, an association strength threshold and a context weight distribution, dynamically annotating data assets based on enterprise data assets, data type range, time range, association strength threshold and context weight distribution to obtain an annotated data set, performing semantic analysis based on the annotated data set to obtain a semantic association graph; constructing a knowledge graph model based on the enterprise data assets, calculating node weights, calculating the association path length using the node weights, and calculating the path contribution; based on the association path length and path contribution Construct a simulated association distribution graph; calculate the node density, set a density dynamic threshold, segment the semantic association graph according to the node density and the density dynamic threshold, and obtain a density segmentation graph; set a relationship dynamic threshold, calculate the relationship strength, segment the density segmentation graph based on the relationship dynamic threshold and the relationship strength, and obtain an actual association distribution graph; mark based on the actual association distribution graph to obtain a marked area, calculate the degree of association of the marked area, and construct an actual association distribution histogram based on the degree of association; adjust the knowledge graph model based on the actual association distribution histogram and the simulated association distribution histogram to obtain an adjusted knowledge graph model, and complete intelligent exploration.

[0006] Optionally, the calculation of node weights and the use of node weights to calculate the associated path length include: setting a feature extraction unit, wherein the feature extraction unit includes: a context parsing unit, a time series unit and a weight allocation unit; parsing context information based on the context parsing unit, and recording context importance and context relevance, and using the time series unit to analyze time span and time weight; using the weight allocation unit to allocate node weights, obtain associated parameters, damping factors and path coefficients, and calculate the associated path length based on the context importance, context relevance, time span, time weight, node weight, damping factor and path coefficient; setting maximum weight and minimum weight, and calculating the associated path length based on the association strength threshold, node weight, context weight distribution, maximum weight, minimum weight and time span.

[0007] Optionally, the calculation of the association path length based on the association strength threshold, node weight, context weight distribution, maximum weight, minimum weight and time span includes: obtaining data asset diversity, data asset complexity, association sensitivity and attenuation factor, and calculating the association path length based on the association strength threshold, node weight, context weight distribution, maximum weight, minimum weight, time span, data asset diversity, data asset complexity, association sensitivity and attenuation factor.

[0008] Optionally, the calculation of path contribution includes: obtaining heterogeneous data compatibility, data flow direction and data dependency, and calculating path contribution based on the node weight, associated path length, context weight distribution, heterogeneous data compatibility, data flow direction and data dependency.

[0009] Optionally, the construction of a simulated association distribution graph based on the association path length and path contribution includes: setting an association identifier and obtaining the total number of nodes, identifying based on the association identifier to obtain identified nodes, obtaining an association path range, dividing the interval based on the association path range to obtain an association path interval; constructing a modified weight parameter based on the path contribution and node weight; constructing a mapping function; calculating the interval association contribution value based on the modified weight parameter, mapping function, association identifier, total number of nodes, association path interval and association path length; constructing an association horizontal axis and an association vertical axis, confirming the association path interval as the association horizontal axis, and the interval association contribution value as the association vertical axis, and constructing an association distribution graph based on the association path interval, the association horizontal axis, the interval association contribution value and the association vertical axis.

[0010] Optionally, constructing a modified weight parameter based on the path contribution and node weight includes: obtaining the maximum weight and maximum contribution, setting a weight factor and a contribution factor, and constructing a modified weight parameter based on the association identifier, time span, path contribution, node weight, maximum weight, maximum contribution, weight factor and contribution factor.

[0011] Optionally, the node density is calculated, a density dynamic threshold is set, and the semantic association graph is segmented according to the node density and the density dynamic threshold to obtain a density segmentation graph, including: setting the node horizontal coordinate and the node vertical coordinate, calculating the horizontal density, vertical density and normal density, obtaining the normal angle, normal direction parameter, horizontal axis sparse parameter and vertical axis sparse parameter, and calculating the node density based on the node horizontal coordinate, the node vertical coordinate, the horizontal density, the vertical density, the normal density, the normal angle, the normal direction parameter, the horizontal axis sparse parameter and the vertical axis sparse parameter; calculating the density mean and the density standard deviation based on the node density, setting dynamic parameters, and setting the density dynamic threshold according to the density mean, the density standard deviation and the dynamic parameters; constructing a segmentation function based on the node density and the density dynamic threshold; and segmenting the semantic association graph based on the segmentation function to obtain a density segmentation graph.

[0012] Optionally, constructing an actual association distribution histogram based on the degree of association includes: summarizing based on the degree of association to obtain an actual association set; constructing an actual association horizontal axis and an actual association vertical axis, confirming the association path interval as the actual association horizontal axis, and confirming the actual association set as the actual association vertical axis; constructing an actual association distribution histogram based on the actual association horizontal axis and the actual association vertical axis.

[0013] Optionally, the knowledge graph model is adjusted based on the actual association distribution histogram and the simulated association distribution histogram to obtain an adjusted knowledge graph model, including: setting a difference threshold and a model state, obtaining a graph resolution, marking based on the association path interval to obtain an association index, and counting the total number of intervals and the actual association contribution value according to the actual association distribution histogram and the simulated association distribution histogram, wherein the model state includes: a normal state and a deviation state; calculating a contribution difference value based on the association index, the total number of intervals, the interval association contribution value and the actual association contribution value; judging whether the contribution difference value is greater than a difference threshold; if the contribution difference value is not greater than the difference threshold, confirming that the model state of the knowledge graph model is a normal state; if the contribution difference value is greater than the difference threshold, confirming that the model state of the knowledge graph model is a deviation state, and increasing the graph resolution by 10% to obtain an adjusted resolution; completing the adjustment based on the adjusted resolution to obtain an adjusted knowledge graph model.

[0014] To achieve the above object, the application further provides a data asset intelligent exploration system based on a knowledge graph, comprising: a semantic analysis module, configured to acquire enterprise data assets, set a data type range, a time range, a correlation strength threshold and a context weight distribution, dynamically label the data assets based on the enterprise data assets, the data type range, the time range, the correlation strength threshold and the context weight distribution, obtain a labeled data set, perform semantic analysis based on the labeled data set, and obtain a semantic correlation graph; a simulated distribution construction module, configured to construct a knowledge graph model based on the enterprise data assets, calculate node weights, calculate correlation path lengths using the node weights, and calculate path contribution degrees; construct a simulated correlation distribution graph based on the correlation path lengths and the path contribution degrees; an actual distribution construction module, configured to calculate node densities, set a density dynamic threshold, segment the semantic correlation graph according to the node densities and the density dynamic threshold, obtain a density segmentation graph, set a relationship dynamic threshold, calculate relationship strengths, segment the density segmentation graph based on the relationship dynamic threshold and the relationship strengths, and obtain an actual correlation distribution graph; mark based on the actual correlation distribution graph, obtain a marked area, calculate a correlation degree of the marked area, and construct an actual correlation distribution histogram based on the correlation degree; and a knowledge graph adjustment module, configured to adjust the knowledge graph model based on the actual correlation distribution histogram and the simulated correlation distribution histogram, obtain an adjusted knowledge graph model, and complete intelligent exploration.

[0015] To solve the above problems, the application further provides an electronic device, comprising: a memory storing at least one instruction; and a processor executing the instruction stored in the memory to implement the above-mentioned data asset intelligent exploration method based on a knowledge graph.

[0016] To solve the above problems, the application further provides a computer-readable storage medium, which stores at least one instruction, and the at least one instruction is executed by a processor in an electronic device to implement the above-mentioned data asset intelligent exploration method based on a knowledge graph.

[0017] In order to solve the problems described in the background technology, the present invention first dynamically annotates data assets based on enterprise data assets, data type range, time range, association strength threshold and context weight distribution to obtain annotated data sets. By setting the data type range, time range, association strength threshold and context weight distribution, data support consistent with actual needs is provided for subsequent operations. By obtaining the annotated data set, the data source is ensured to be reliable, laying the foundation for the verification and optimization of intelligent exploration. Secondly, semantic analysis is performed based on the annotated data set to obtain a semantic association graph. Semantic analysis technology can capture the semantic relationship between data assets with high precision, avoid the limitations of traditional methods, and improve the credibility of data processing. The semantic association graph provides accurate input data for subsequent node analysis, density segmentation and relationship evaluation. Afterwards, a knowledge graph model is constructed based on enterprise data assets to calculate node weights, association path lengths and path contributions. The calculation of node weights, association path lengths and path contributions and the construction of the knowledge graph model provide a basis for the construction of a simulated association distribution graph. By calculating the path contribution, To quantify the contribution of the association between data assets; further, based on the association path length and path contribution, a simulated association distribution graph is constructed, and the simulated association distribution graph can intuitively represent the association contribution value of the data asset in the association path interval; then, the semantic association graph is segmented to obtain a density segmentation graph, and the semantic association graph is segmented by node density and density dynamic threshold, which can accurately segment strong association areas and weak association areas, and effectively remove noise interference; then, based on the relationship dynamic threshold and relationship strength, the density segmentation graph is segmented to obtain an actual association distribution graph, and the density segmentation graph is further segmented by relationship strength and relationship dynamic threshold, which improves the accuracy of the actual association distribution graph and is conducive to the subsequent construction of the actual association distribution histogram; finally, the actual association distribution histogram is constructed, and the knowledge graph model is adjusted according to the actual association distribution histogram and the simulated association distribution histogram. The knowledge graph model is adjusted by comparing the actual and simulated association distribution histograms, so that the output of the knowledge graph model gradually approaches the actual association result, realizing closed-loop optimization and significantly improving the accuracy of data asset intelligent exploration. Therefore, the present invention can improve the intelligence level and accuracy of data asset exploration. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of a process for intelligent data asset discovery based on a knowledge graph according to an embodiment of the present invention;

[0019] Figure 2 This is a functional module diagram of a knowledge graph-based data asset intelligent exploration system provided by one embodiment of the present invention;

[0020] Figure 3A schematic diagram of the structure of an electronic device for implementing the method for intelligent exploration of data assets based on knowledge graphs provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0021] The present invention provides a method and system for intelligent data asset exploration based on a knowledge graph. This system achieves intelligent exploration of data assets by combining semantic analysis, knowledge graph modeling, node weight calculation, path contribution evaluation, and association distribution graph construction. The following describes specific embodiments of the present invention in detail with reference to the accompanying drawings and their reference numbers.

[0022] like Figure 1 As shown, the core process of the present invention includes multiple key steps, which together constitute a complete closed loop from data acquisition to knowledge graph adjustment. First, in step S1, the system obtains enterprise data assets through the semantic analysis module, and sets the data type range, time range, association strength threshold and context weight distribution. The setting of these parameters is the key basis for ensuring that subsequent operations are consistent with actual needs. For example, the data type range can be limited to structured data and unstructured data, the time range can be set to data records within the past year, and the association strength threshold is used to screen out data assets with significant correlation. In addition, the setting of context weight distribution can reflect the importance of different contextual information, thereby providing a basis for dynamic annotation. At this stage, the system generates annotated data sets by dynamically annotating data assets. The annotated data sets not only contain original data, but also have semantic tags attached to clarify the category, time attributes and contextual relationships of each data. This annotation method provides reliable data support for subsequent semantic analysis.

[0023] In step S2, the system performs semantic analysis based on the annotated data set and generates a semantic association graph. The core of semantic analysis is to use natural language processing technology and machine learning algorithms to identify semantic relationships between data assets. For example, for a transaction record describing "Customer A purchased product B", the semantic analysis module will extract the purchase relationship between "Customer A" and "Product B" and convert it into a node connection in the semantic association graph. The construction process of the semantic association graph involves a variety of algorithms. One typical method is to calculate the similarity between words based on a word embedding model (such as Word2Vec or BERT) to determine the strength of the association between nodes. Assuming that the semantic vectors of two nodes are v1 and v2 respectively, their similarity can be calculated using the cosine similarity formula cosθ=(v1·v2) / (|v1|·|v2|). Ultimately, the semantic association graph provides high-precision input data for subsequent knowledge graph modeling.

[0024] Next, in step S3, the simulation distribution construction module constructs a knowledge graph model based on the enterprise data assets, and calculates the node weight, associated path length and path contribution. The construction process of the knowledge graph model includes three main steps: node definition, edge definition and weight assignment. Nodes represent entities in data assets, edges represent relationships between entities, and weights reflect the importance of relationships. In order to calculate node weights, the system sets up a feature extraction unit, which includes a context parsing unit, a time series unit and a weight assignment unit. The context parsing unit is responsible for parsing context information and recording context importance and context relevance, the time series unit is used to analyze time spans and time weights, and the weight assignment unit assigns node weights based on the above results. The calculation formula for node weight can be expressed as W=αC+βT+γP, where W represents node weight, C represents context importance, T represents time weight, P represents path coefficient, and α, β, and γ are corresponding weight factors respectively. On this basis, the system further calculates the associated path length. Assume that the node weights on a certain path are w1, w2,..., w n , then the associated path length L can be calculated by the formula L=Σ(w i ·d i ) calculation, where w1 represents the weight of the i-th node, d i represents the distance from the i-th node to the next node. To improve the accuracy of the calculation, the system also introduces parameters such as data asset diversity, complexity, association sensitivity, and attenuation factor. These parameters affect the final association path length through weighted summation.

[0025] In step S4, the system constructs a simulated association distribution graph based on the association path lengths and path contributions. The construction of the simulated association distribution graph includes the following steps: First, the system sets an association identifier and obtains the total number of nodes. Then, the nodes are labeled based on the association identifier to obtain identified nodes. Next, the system obtains the association path range and divides it into intervals to obtain association path intervals. The purpose of interval division is to map the association path lengths into different intervals to facilitate subsequent statistical analysis. To quantify the association contribution value of each interval, the system constructs a modified weight parameter based on the path contribution and node weight. The calculation formula for the modified weight parameter can be expressed as R = λW + μC, where R represents the modified weight parameter, W represents the node weight, C represents the path contribution, and λ and μ represent the corresponding weight factors. Subsequently, the system constructs a mapping function f(x) and substitutes the modified weight parameter, mapping function, association identifier, total number of nodes, association path interval, and association path length into the formula Q = f(R·L) to calculate the interval association contribution value Q. Finally, the system constructs the association horizontal axis and the association vertical axis, confirms the association path interval as the association horizontal axis, and confirms the interval association contribution value as the association vertical axis, thereby generating a simulated association distribution diagram.

[0026] In step S5, the actual distribution construction module calculates the node density and sets the density dynamic threshold to segment the semantic association graph and obtain the density segmentation graph. The node density calculation process includes the following steps: First, the system sets the node horizontal coordinate and the node vertical coordinate, and calculates the horizontal density, vertical density and normal density respectively. Assume that the node set is N = {n1, n2, ..., n k}, then the horizontal density Dx can be obtained by formula D x =∑(x i -x j )^(-2) calculation, where x i and x j Represents node n i and n j Similarly, the vertical density D y and normal density D z It can also be calculated using the corresponding formula. In order to further optimize the density calculation results, the system introduces auxiliary variables such as normal angle, normal direction parameter, horizontal axis sparse parameter and vertical axis sparse parameter. These variables affect the final node density value through weighted summation. The calculation formula for node density can be expressed as D = δDx + εDy + ζDz, where δ, ε, and ζ are the corresponding weight factors. On this basis, the system calculates the density mean and density standard deviation based on the node density, and sets dynamic parameters to determine the density dynamic threshold. The calculation formula for the density dynamic threshold can be expressed as T = μ + σ·k, where μ represents the density mean, σ represents the density standard deviation, and k is a dynamic parameter. Finally, the system constructs a segmentation function based on the node density and the density dynamic threshold, and segments the semantic association graph through the segmentation function to obtain a density segmentation graph.

[0027] In step S6, the system sets the dynamic threshold of the relationship and calculates the relationship strength to further segment the density segmentation map and obtain the actual association distribution map. The relationship strength calculation process includes the following steps: First, the system obtains parameters such as heterogeneous data compatibility, data flow direction, and data dependency, and calculates the relationship strength based on these parameters. Assuming that the heterogeneous data compatibility of a relationship is H, the data flow direction is F, and the data dependency is D, the relationship strength S can be calculated by the formula Calculate, where ρ, ψ represents the corresponding weighting factor. Based on this, the system sets a dynamic relationship threshold and segments the density segmentation map based on the relationship strength and dynamic relationship threshold. The core of the segmentation process is to determine whether each relationship meets the dynamic relationship threshold requirements. Only relationships that meet the requirements are retained. Finally, the system generates an actual relationship distribution map, which can intuitively demonstrate the actual connections between data assets.

[0028] In step S7, the system marks based on the actual association distribution map to obtain the marked area, and calculates the degree of association of the marked area to construct the actual association distribution histogram. The generation process of the marked area includes the following steps: First, the system marks the strong association area and the weak association area based on the actual association distribution map. The marking standard can be determined according to parameters such as node density and relationship strength. For example, when the node density of a certain area is higher than the density dynamic threshold and the relationship strength is higher than the relationship dynamic threshold, the area will be marked as a strong association area. Then, the system calculates the degree of association of the marked area. Assume that a marked area contains m relationships, and the association contribution values ​​of each relationship are q1, q2, ..., q m , then the correlation degree C of the region can be expressed by the formula C = ∑q i Finally, the system constructs the actual correlation horizontal axis and the actual correlation vertical axis, identifies the correlation path interval as the actual correlation horizontal axis, and identifies the actual correlation set as the actual correlation vertical axis, thereby generating an actual correlation distribution histogram.

[0029] In step S8, the knowledge graph adjustment module adjusts the knowledge graph model based on the actual association distribution histogram and the simulated association distribution histogram to obtain an adjusted knowledge graph model. The adjustment process includes the following steps: First, the system sets the difference threshold and model status, and obtains the graph resolution. Assuming that the total number of intervals in the actual association distribution histogram is M, and the total number of intervals in the simulated association distribution histogram is N, the system marks based on the association path interval to obtain the association index. Next, the system counts the total number of intervals and the actual association contribution value based on the actual association distribution histogram and the simulated association distribution histogram, and calculates the contribution difference value Δ based on the association index, the total number of intervals, the interval association contribution value and the actual association contribution value. The calculation formula of the contribution difference value can be expressed as Δ=∑|Q i -P i |, where Q i Indicates the actual correlation contribution value, P i represents the simulated association contribution value. The system then determines whether the contribution difference value is greater than the difference threshold. If the contribution difference value is not greater than the difference threshold, the knowledge graph model is confirmed to be in a normal state. If the contribution difference value is greater than the difference threshold, the knowledge graph model is confirmed to be in a deviation state and the graph resolution is increased by 10% to obtain an adjusted resolution. Finally, the system completes the adjustment based on the adjusted resolution, obtaining an adjusted knowledge graph model.

[0030] In order to implement the above method, the present invention also provides a data asset intelligent exploration system based on knowledge graph, such as Figure 2As shown in the figure. The system includes a semantic analysis module, a simulated distribution construction module, an actual distribution construction module and a knowledge graph adjustment module. The semantic analysis module is responsible for obtaining enterprise data assets, setting parameters and generating labeled data sets and semantic association graphs; the simulated distribution construction module is responsible for constructing a knowledge graph model, calculating node weights and path contribution, and generating a simulated association distribution graph; the actual distribution construction module is responsible for calculating node density, setting dynamic threshold and generating an actual association distribution graph; the knowledge graph adjustment module is responsible for adjusting the knowledge graph model based on the actual and simulated association distribution histograms. In addition, as shown in the figure, the present application also provides an electronic device, which includes a memory and a processor, the memory is used to store at least one instruction, and the processor is used to execute the instruction stored in the memory to realize the above method. At the same time, the present application also provides a computer readable storage medium, which stores at least one instruction, and the instruction is executed by the processor in the electronic device to realize the above method. Figure 3

[0031] In summary, the present application realizes intelligent exploration of data assets by combining semantic analysis, knowledge graph modeling, node weight calculation, path contribution evaluation and association distribution graph construction. This method not only improves the intelligent level and accuracy of data asset exploration, but also provides strong technical support for enterprise data asset management.​

Claims

1. A method for intelligent data asset exploration based on knowledge graph, characterized in that: The method comprises: Acquire enterprise data assets, set data type range, time range, association strength threshold, and context weight distribution, dynamically annotate data assets based on enterprise data assets, data type range, time range, association strength threshold, and context weight distribution to obtain an annotated data set, perform semantic analysis based on the annotated data set, and obtain a semantic association graph; Building a knowledge graph model based on the enterprise data assets, calculating node weights, calculating the length of associated paths using the node weights, and calculating the path contribution; Constructing a simulated association distribution graph based on the association path length and the path contribution; Calculate the node density, set the density dynamic threshold, segment the semantic association graph according to the node density and density dynamic threshold, and obtain the density segmentation graph; Set the relationship dynamic threshold, calculate the relationship strength, segment the density segmentation map based on the relationship dynamic threshold and relationship strength, and obtain the actual association distribution map; Marking is performed based on the actual correlation distribution map to obtain a marked area, calculating the correlation degree of the marked area, and constructing an actual correlation distribution histogram based on the correlation degree; The knowledge graph model is adjusted based on the actual association distribution histogram and the simulated association distribution histogram to obtain an adjusted knowledge graph model and complete intelligent exploration.

2. The method for intelligent data asset exploration based on knowledge graph according to claim 1, characterized in that: The calculating node weights and calculating the associated path lengths using the node weights include: Setting a feature extraction unit, wherein the feature extraction unit includes a context analysis unit, a time series unit and a weight allocation unit; Parsing context information based on the context parsing unit, recording context importance and context relevance, and analyzing time span and time weight using the time series unit; Assigning node weights using the weight assignment unit, obtaining association parameters, damping factors, and path coefficients, and calculating association path lengths based on the context importance, context relevance, time span, time weight, node weights, damping factors, and path coefficients; The maximum weight and the minimum weight are set, and the association path length is calculated based on the association strength threshold, the node weight, the context weight distribution, the maximum weight, the minimum weight and the time span.

3. The method for intelligent data asset exploration based on knowledge graph according to claim 2, characterized in that: The calculating of the association path length based on the association strength threshold, node weight, context weight distribution, maximum weight, minimum weight and time span includes: Obtain data asset diversity, data asset complexity, association sensitivity and attenuation factor, and calculate the association path length based on the association strength threshold, node weight, context weight distribution, maximum weight, minimum weight, time span, data asset diversity, data asset complexity, association sensitivity and attenuation factor.

4. The method for intelligent data asset exploration based on knowledge graph according to claim 1, characterized in that: The calculating path contribution includes: Obtain heterogeneous data compatibility, data flow direction and data dependency, and calculate path contribution based on the node weight, associated path length, context weight distribution, heterogeneous data compatibility, data flow direction and data dependency.

5. The method for intelligent data asset exploration based on knowledge graph according to claim 1, characterized in that: The constructing of a simulated association distribution graph based on the association path length and the path contribution includes: Set an association identifier and obtain the total number of nodes, identify based on the association identifier to obtain the identified node, obtain an association path range, divide the association path range into intervals based on the association path range, and obtain an association path interval; Constructing a modified weight parameter based on the path contribution and node weight; Construct a mapping function; Calculating an interval association contribution value based on the modified weight parameter, the mapping function, the association identifier, the total number of nodes, the association path interval, and the association path length; Construct an associated horizontal axis and an associated vertical axis, confirm the associated path interval as the associated horizontal axis, confirm the interval associated contribution value as the associated vertical axis, and construct an associated distribution graph based on the associated path interval, the associated horizontal axis, the interval associated contribution value and the associated vertical axis.

6. The method for intelligent data asset exploration based on knowledge graph according to claim 5, characterized in that: The constructing of the modified weight parameter based on the path contribution and the node weight includes: Obtain the maximum weight and maximum contribution, set the weight factor and contribution factor, and construct the modified weight parameter based on the association identifier, time span, path contribution, node weight, maximum weight, maximum contribution, weight factor and contribution factor.

7. The method for intelligent data asset exploration based on knowledge graph according to claim 1, characterized in that: The node density is calculated, a density dynamic threshold is set, and the semantic association graph is segmented according to the node density and the density dynamic threshold to obtain a density segmentation graph, including: Set the node horizontal coordinate and node vertical coordinate, calculate the horizontal density, vertical density and normal density, obtain the normal angle, normal direction parameter, horizontal axis sparse parameter and vertical axis sparse parameter, and calculate the node density based on the node horizontal coordinate, node vertical coordinate, horizontal density, vertical density, normal density, normal angle, normal direction parameter, horizontal axis sparse parameter and vertical axis sparse parameter; Calculating a density mean and a density standard deviation based on the node density, setting dynamic parameters, and setting a density dynamic threshold according to the density mean, the density standard deviation and the dynamic parameters; Constructing a segmentation function based on the node density and the density dynamic threshold; The semantic association graph is segmented based on the segmentation function to obtain a density segmentation graph.

8. The method for intelligent data asset exploration based on knowledge graph according to claim 1, characterized in that: The actual correlation distribution histogram is constructed based on the correlation degree, including: Summarize based on the association degree to obtain an actual association set; Constructing an actual association horizontal axis and an actual association vertical axis, confirming the association path interval as the actual association horizontal axis, and confirming the actual association set as the actual association vertical axis; An actual correlation distribution histogram is constructed based on the actual correlation horizontal axis and the actual correlation vertical axis.

9. The method for intelligent data asset exploration based on knowledge graph according to claim 1, characterized in that: The step of adjusting the knowledge graph model based on the actual association distribution histogram and the simulated association distribution histogram to obtain an adjusted knowledge graph model includes: Setting a difference threshold and a model state, obtaining a map resolution, marking the associated path intervals, obtaining an associated index, and calculating the total number of intervals and the actual associated contribution value based on the actual associated distribution histogram and the simulated associated distribution histogram, wherein the model state includes a normal state and a deviation state; Calculate the contribution difference value based on the association index, the total number of intervals, the interval association contribution value and the actual association contribution value; Determining whether the contribution difference value is greater than a difference threshold; If the contribution difference value is not greater than the difference threshold, the model state of the knowledge graph model is confirmed to be a normal state; If the contribution difference value is greater than the difference threshold, the model state of the knowledge graph model is confirmed to be a deviation state, and the graph resolution is increased by ten percent to obtain an adjusted resolution; The adjustment is completed based on the adjustment resolution to obtain an adjusted knowledge graph model.

10. A data asset intelligent exploration system based on knowledge graph, characterized in that: The system comprises: The semantic analysis module is used to obtain enterprise data assets, set the data type range, time range, association strength threshold and context weight distribution, dynamically annotate the data assets based on the enterprise data assets, data type range, time range, association strength threshold and context weight distribution, obtain annotated data sets, and perform semantic analysis based on the annotated data sets to obtain a semantic association graph; A simulation distribution construction module is used to construct a knowledge graph model based on the enterprise data assets, calculate node weights, calculate the length of the associated path using the node weights, and calculate the path contribution; and construct a simulated association distribution graph based on the associated path length and path contribution; The actual distribution construction module is used to calculate the node density, set the density dynamic threshold, segment the semantic association graph according to the node density and the density dynamic threshold, and obtain a density segmentation graph; set the relationship dynamic threshold, calculate the relationship strength, segment the density segmentation graph based on the relationship dynamic threshold and the relationship strength, and obtain an actual association distribution graph; mark the actual association distribution graph to obtain the marked area, calculate the association degree of the marked area, and construct the actual association distribution histogram based on the association degree; The knowledge graph adjustment module is used to adjust the knowledge graph model based on the actual association distribution histogram and the simulated association distribution histogram to obtain an adjusted knowledge graph model and complete intelligent exploration.

Images