File resource classification method and system based on knowledge graph

Abstract

The invention discloses a file resource classification method and system based on a knowledge graph, and the method comprises the steps: constructing an initial file relation network; calculating a comprehensive weight value of each node, and endowing the comprehensive weight value to the corresponding node; weighting connections in the initial file relation network to obtain a weighted file network; determining a directed weighted time sequence diagram; calculating derivative indexes of the nodes; determining a source node set, a derivative node set and a terminal node set; nodes in the source node set are used as root nodes, a hierarchical evolution tree is constructed based on the connection relation in the directed weighted time sequence diagram, source nodes in the hierarchical evolution tree are used as a first hierarchy, derivative nodes directly connected with the source nodes are used as a second hierarchy, and terminal nodes are used as a final hierarchy. The method and the device are used for describing the time sequence evolution rule and the hierarchical derivative relationship between the file resources, realizing dynamic classification management of the file resources, and providing a more systematic and deeper file association analysis result for a user.

Description

Technical Field

[0001] This application belongs to the field of file resource classification, and in particular relates to a file resource classification method and system based on knowledge graph. Background Art

[0002] With the surge in the number of document resources, especially those involving macro-level guidance and industry standards, their classification and management faces significant challenges. Traditional document resource classification methods rely primarily on manual experience for hierarchical classification and labeling. However, when faced with highly relevant and time-sensitive document resources, it is often difficult to accurately grasp the relationships and evolution of these documents, resulting in suboptimal classification results and hindering the precise delivery and implementation of document resources.

[0003] Related technologies can be used to construct file feature vectors, extract key concepts and entities from files, establish entity relationship networks, and combine them with hierarchical clustering algorithms to classify and organize files. This technology not only considers the subject characteristics of files but also constructs the associations between files, effectively achieving automatic classification of file resources.

[0004] However, the above-mentioned related technologies are difficult to effectively identify and express the derivative relationships and implementation hierarchies between files, making it difficult for the classification results to accurately reflect the dynamic development laws of file resources in actual application scenarios, which is not conducive to users' in-depth understanding and grasp of the evolution path of related files. Summary of the Invention

[0005] This application provides a file resource classification method and system based on knowledge graph, which is used to characterize the temporal evolution rules and hierarchical derivative relationships between file resources, realize dynamic classification management of file resources, and provide users with more systematic and in-depth file association analysis results.

[0006] In the first aspect, the present application provides a file resource classification method based on a knowledge graph, which extracts a reference relationship dataset and a supporting relationship dataset between files, as well as the timeliness parameters and effectiveness level parameters of each file from a preset file resource library; An initial file relationship network is constructed based on the reference relationship dataset and the supporting relationship dataset. Each file in the initial file relationship network is a node, and connections are established between nodes based on reference relationships and supporting relationships. Calculate the comprehensive weight value of each node based on the timeliness parameter and the effectiveness level parameter, and assign the comprehensive weight value to the corresponding node; The connections in the initial file relationship network are weighted using the comprehensive weight value to obtain a weighted file network, where the weights of the connections in the weighted file network are calculated by the difference in the comprehensive weight values of the connected nodes; The weighted file network is converted into a directed time sequence graph. By comparing the time attributes of the connected nodes, the direction of the connection is determined to be from the node with earlier time to the node with later time, thus obtaining a directed weighted time sequence graph. In a directed weighted time series graph, the derived index of each node is calculated based on the sum of its in-degree connection weight and out-degree connection weight. Classify the nodes in the directed weighted time series graph according to the derivative index to obtain the source node set, the derivative node set and the terminal node set; Taking the nodes in the source node set as the root node, a hierarchical evolution tree is constructed based on the connection relationship in the directed weighted time series graph. In the hierarchical evolution tree, the source node is the first level, the derived nodes directly connected to the source node are the second level, and the terminal node is the last level.

[0007] By adopting the above technical solution, by extracting the reference and matching relationships between files, and combining the timeliness parameters and effectiveness level parameters to construct a weighted directed time series graph, the multi-dimensional correlation characteristics between file resources can be accurately reflected. By calculating the node derivative index based on the in-degree and out-degree connection weights, the role positioning of the file in the evolution process can be objectively identified. The nodes are divided into three categories: source, derivative, and terminal, and a hierarchical evolution tree is constructed to make the evolution of file resources clearly visible. The correlation evolution relationship between files is depicted through the graph structure, and the influence of timeliness and effectiveness level on the correlation strength is taken into account. The construction process of the hierarchical evolution tree follows the temporality and correlation of files, which improves the rationality of the classification results. This classification method depicts the temporal evolution law and hierarchical derivative relationship between file resources, realizes the dynamic classification management of file resources, and provides users with more systematic and in-depth file correlation analysis results.

[0008] In conjunction with some embodiments of the first aspect, in some embodiments, extracting a reference relationship dataset and a matching relationship dataset between files from a preset file resource library specifically includes: Perform word segmentation on the file contents in the preset file resource library to obtain a word sequence; Identifying file identification words from a word sequence based on a preset file identification dictionary, the file identification words including file name, file number and release time; Constructing a file identification pair, the file identification pair including a reference file identification and a referenced file identification; Performing contextual scanning in the file content based on the file identifier pair to extract a text segment containing the file identifier pair; Perform semantic analysis on text fragments to identify the relationship types between file identifiers, including reference relationships and supporting relationships; File identification pairs are classified into reference relationship datasets and supporting relationship datasets according to the relationship type.

[0009] By employing this technical solution, file identifiers are identified using a preset file identifier dictionary, improving the accuracy and completeness of file identifier extraction. Semantic analysis of text segments containing file identifier pairs can distinguish between two different types of relationships: reference relationships and supporting relationships. This semantic analysis improves the accuracy of relationship identification. This automated relationship extraction process significantly improves data acquisition efficiency while ensuring data objectivity and reliability.

[0010] In conjunction with some embodiments of the first aspect, in some embodiments, calculating the comprehensive weight value of each node based on the timeliness parameter and the effectiveness level parameter specifically includes: A timeliness rating matrix is constructed based on the timeliness parameter and the effectiveness level parameter. The rows of the timeliness rating matrix represent the document status, which includes currently valid, about to take effect, about to expire, and abolished. The columns of the timeliness rating matrix represent the scoring time period. According to the current time position in the scoring time period, the timeliness score of the corresponding file status is obtained from the timeliness scoring matrix; Constructing an effectiveness level scoring vector, wherein different effectiveness levels in the effectiveness level scoring vector correspond to different basic scores; Obtaining an effectiveness level score from an effectiveness level score vector according to the effectiveness level of the document; The timeliness score and the effectiveness level score are weighted and summed to obtain a comprehensive weight value.

[0011] By adopting the above technical solution, the timeliness score matrix reflects the effectiveness of different file states in different time periods, allowing the weight calculation to dynamically adapt to changes in file states. The effectiveness level score vector reflects the basic importance of files at different effectiveness levels. The timeliness score and the effectiveness level score are weighted and summed to obtain a comprehensive weight value that takes into account both the current timeliness status of the file and the inherent effectiveness level of the file. This weight calculation method makes the evaluation of the importance of the node more comprehensive and reasonable. The calculated comprehensive weight value can objectively reflect the importance of the file in the entire file system, providing a reliable quantitative basis for subsequent network weighting and node classification.

[0012] In conjunction with some embodiments of the first aspect, in some embodiments, after constructing the hierarchical evolution tree based on the connection relationship in the directed weighted time sequence graph, the method further includes: Calculate the hierarchical clustering of nodes at each level in the hierarchical evolution tree. The hierarchical clustering represents the degree of association between nodes at the same level. The hierarchical evolution tree is branch-optimized based on the hierarchical aggregation degree to obtain an optimized hierarchical evolution tree; Calculate the structural importance index of the nodes in the optimized hierarchical evolution tree. The structural importance index is determined based on the node position, number of connections and connection weight; File resources are managed hierarchically according to structural importance indicators.

[0013] By adopting the above technical solution, the hierarchical evolution tree is optimized and hierarchically managed by calculating the hierarchical aggregation and structural importance indicators. The hierarchical aggregation reflects the closeness of the association between nodes at the same level. Branch optimization based on this indicator can make the hierarchical structure more reasonable. The structural importance indicator comprehensively considers the position, number of connections and connection weight of the node, and comprehensively reflects the importance of the node in the network structure. This optimization and grading method makes the structure of the hierarchical evolution tree more consistent with the actual association characteristics of file resources, avoiding the one-sidedness of relying solely on hierarchical position to judge importance. Hierarchical management through structural importance indicators can more accurately reflect the importance of files in the entire file system, providing a scientific quantitative basis for the differentiated management of file resources.

[0014] In conjunction with some embodiments of the first aspect, in some embodiments, calculating the hierarchical aggregation degree of nodes at each level in the hierarchical evolution tree specifically includes: Extract possible connection paths between nodes in the same level; Calculate the path weight of each connection path. The path weight is the cumulative value of the weights of each connection on the connection path. Construct a hierarchical connectivity matrix, where the elements in the hierarchical connectivity matrix are the path weights between corresponding node pairs; Calculate the clustering coefficient of nodes based on the hierarchical connectivity matrix; The mean of the clustering coefficients of all nodes in the same level is taken as the level clustering degree to obtain the level clustering degree of nodes in each level.

[0015] By adopting the above technical solution, by extracting possible connection paths between nodes at the same level and calculating the path weights, a hierarchical connectivity matrix is constructed to quantify the strength of the association between nodes. Then, based on the connectivity matrix, the clustering coefficient of the nodes is calculated and the hierarchical aggregation degree is obtained, so that the system can accurately measure the association density of file resources within each level. A higher hierarchical aggregation degree indicates that there are close business connections and logical associations between file resources at this level. The quantification of this association relationship helps to identify important clustering groups formed during the evolution of file resources. Through the cumulative calculation of path weights, not only the strength of direct connections is taken into account, but also the influence of indirect associations is reflected, making the calculation results of hierarchical aggregation more comprehensive and objective, and able to reveal the organizational characteristics of file resources from the perspective of network topology, thereby improving the scientificity and accuracy of file resource classification.

[0016] In conjunction with some embodiments of the first aspect, in some embodiments, after hierarchically managing the file resources according to the structural importance index, the method further includes: Constructing an evolution feature vector of the file resource, the evolution feature vector includes the hierarchical position of the node in the optimized hierarchical evolution tree, the structural importance index, and the associated node information; Calculating the evolution similarity matrix of file resources based on the evolution feature vector; Use clustering algorithms to analyze the evolution similarity matrix and identify groups of files with similar evolution patterns; Build a separate evolution prediction model for each file group.

[0017] By adopting the above technical solution, by constructing an evolutionary feature vector that includes hierarchical position, structural importance and associated node information, calculating the evolutionary similarity matrix of file resources and analyzing it using a clustering algorithm, the system can identify file groups with similar evolutionary patterns from multiple dimensions. The evolutionary feature vector comprehensively characterizes the characteristic attributes of file resources in the hierarchical evolution tree, so that the similarity calculation takes into account both the structural characteristics of the file and the temporal evolution characteristics. Clustering analysis based on the similarity matrix can automatically discover the evolutionary patterns between file resource groups. This pattern reflects the common characteristics of different types of file resources in the process of business development. By establishing an independent evolutionary prediction model for each file group, the evolutionary characteristics of different types of file resources can be more accurately characterized, thereby improving the adaptability and prediction accuracy of the file resource classification system to business changes.

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, building an independent evolution prediction model for each file group specifically includes: Extract the historical evolution sequence of each file group. The historical evolution sequence includes the file's creation time, expiration time, and association relationship change information. Construct a time window feature matrix and segment the historical evolution sequence according to the preset time window size; The evolution trend predictor is trained based on the time window feature matrix to obtain an independent evolution prediction model corresponding to each file group.

[0019] By adopting the above technical solution, by extracting the historical evolution sequence of the file group, the evolution information including the generation time, expiration time and changes in the association relationship is segmented according to the preset time window, the time window feature matrix is constructed and the evolution trend predictor is trained, so that the system can establish a highly targeted independent evolution prediction model. The time window feature matrix retains the temporal characteristics and dynamic change information of the association relationship during the evolution of file resources. This fine-grained feature extraction method helps to capture the evolution pattern of the file group in different time periods. The predictor trained based on the time window feature matrix can learn the unique evolution rules of the file group. The prediction model fully utilizes the evolution pattern information contained in the historical data, improving the file resource classification system's ability to predict future evolution trends and classification accuracy.

[0020] In the second aspect, an embodiment of the present application provides a file resource classification system based on a knowledge graph, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and one or more processors call the computer instructions to enable the system to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer-readable storage medium comprising instructions, which, when executed on a system, enables the system to execute the method described in the first aspect and any possible implementation of the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer program product, which, when executed on a system, enables the system to execute the method described in any possible implementation manner in the first aspect.

[0023] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. The present application provides a file resource classification method based on a knowledge graph. By extracting the references and matching relationships between files and combining the timeliness parameters and effectiveness level parameters to construct a weighted directed time series graph, it can accurately reflect the multi-dimensional association characteristics between file resources. The node derivative index is calculated based on the in-degree and out-degree connection weights, which can objectively identify the role positioning of the file in the evolution process. The nodes are divided into three categories: source, derivative, and terminal, and a hierarchical evolution tree is constructed to make the evolution of file resources clearly visible. The association evolution relationship between files is depicted through the graph structure, and the influence of timeliness and effectiveness level on the association strength is considered. The construction process of the hierarchical evolution tree follows the temporality and association of files, which improves the rationality of the classification results. This classification method depicts the temporal evolution law and hierarchical derivative relationship between file resources, realizes the dynamic classification management of file resources, and provides users with more systematic and in-depth file association analysis results.

[0024] 2. The present application provides a file resource classification method based on a knowledge graph, which optimizes and hierarchically manages the hierarchical evolution tree by calculating the hierarchical aggregation and structural importance indicators. The hierarchical aggregation reflects the degree of association between nodes at the same level. Branch optimization based on this indicator can make the hierarchical structure more reasonable. The structural importance index comprehensively considers the position, number of connections and connection weight of the node, and comprehensively reflects the importance of the node in the network structure. This optimization and grading method makes the structure of the hierarchical evolution tree more consistent with the actual association characteristics of the file resources, and avoids the one-sidedness of relying solely on the hierarchical position for importance judgment. Hierarchical management through structural importance indicators can more accurately reflect the importance of the file in the entire file system, and provide a scientific quantitative basis for the differentiated management of file resources.

[0025] 3. The present application provides a file resource classification method based on a knowledge graph. By constructing an evolutionary feature vector containing hierarchical position, structural importance, and associated node information, calculating the evolutionary similarity matrix of the file resource and analyzing it using a clustering algorithm, the system can identify file groups with similar evolutionary patterns from multiple dimensions. The evolutionary feature vector comprehensively characterizes the characteristic attributes of the file resource in the hierarchical evolution tree, so that the similarity calculation takes into account both the structural characteristics of the file and the temporal evolution characteristics. Cluster analysis based on the similarity matrix can automatically discover the evolutionary laws between file resource groups. This law reflects the common characteristics of different types of file resources in the process of business development. By establishing an independent evolution prediction model for each file group, the evolutionary characteristics of different types of file resources can be more accurately characterized, thereby improving the adaptability and prediction accuracy of the file resource classification system to business changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1This is a flow chart of a file resource classification method based on knowledge graph in an embodiment of the present application.

[0027] Figure 2 This is another flow chart of a file resource classification method based on knowledge graph in an embodiment of the present application.

[0028] Figure 3 This is a schematic diagram of the physical device structure of a file resource classification system based on knowledge graph provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] The terms used in the following examples of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification and appended claims of this application, the singular expressions "a," "an," "said," "above," "the," and "this" are intended to include plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in this application refers to any or all possible combinations comprising one or more of the listed items.

[0030] In the following, the terms "first" and "second" are used for descriptive purposes only and should not be understood to imply or suggest relative importance or implicitly indicate the number of the technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of this application, unless otherwise specified, "plurality" means two or more.

[0031] The following uses an embodiment and combines Figure 1 , a file resource classification method based on knowledge graph in an embodiment of the present application is described: See also Figure 1 , which is a flow chart of a file resource classification method based on knowledge graph in an embodiment of the present application.

[0032] S101, extracting a reference relationship dataset and a matching relationship dataset between files, as well as a timeliness parameter and effectiveness level parameter of each file from a preset file resource library; The system first extracts a dataset of reference and supporting relationships between files from a pre-set file repository. Specifically, the file contents in the pre-set file repository are segmented to obtain word sequences. File identifiers are identified from these word sequences based on a pre-set file identifier dictionary. These include file names, file numbers, and release dates. File identifier pairs are constructed, consisting of referencing file identifiers and referenced file identifiers. Based on these file identifier pairs, contextual scanning is performed within the file contents to extract text fragments containing these file identifier pairs. Semantic analysis is performed on these text fragments to identify the relationship types between the file identifier pairs, including reference and supporting relationships. File identifier pairs are then classified into reference and supporting relationship datasets based on their relationship types. Timeliness and effectiveness level parameters are also extracted for each file.

[0033] In this step, the system first extracts the reference relationship dataset and the matching relationship dataset between files from a pre-set file resource library. The file resource library can be a pre-established structured database that stores file information, such as a relational database or graph database, or it can be a collection of unstructured files, such as text files or PDF files. The system can use appropriate data extraction and processing techniques based on the actual file storage format to obtain the relationship information between files.

[0034] For structured file repositories, the system can directly extract file metadata information, such as file identifiers, publication dates, and citation relationships, through database query statements. For unstructured file repositories, the system needs to use text analysis technology to process file content and identify file identifiers and semantic relationships between files.

[0035] In specific implementation, the system can follow the steps below: Each file in the document repository undergoes text preprocessing, such as stop word removal, word segmentation, and part-of-speech tagging, to obtain a standardized word sequence. Based on a preset file identifier dictionary, file identifiers, such as file name, file number, and publication date, are identified from the word sequence. This file identifier dictionary can be customized for specific fields and application scenarios, such as patents and papers in scientific literature. The identified file identifiers are combined according to specific rules to construct file identifier pairs, representing potential relationships between two files. Common file identifier pair construction rules include identical file names, identical file numbers, and similar publication dates. Contextual scanning is performed within the original file content, centering on the file identifier pairs, to extract text segments containing the file identifier pairs. Contextual scanning can employ techniques such as sliding windows and syntactic analysis to identify relevant text segments based on the location and frequency of the file identifier pairs. Semantic analysis is then performed on the extracted text segments to identify the specific relationship types between the file identifier pairs. Common relationship types include citation, reference, and matching. Semantic analysis can adopt rule-based methods, such as pattern matching, keyword matching, etc., or machine learning-based methods, such as text classification, relationship extraction, etc. According to the identified relationship type, the file identification pairs are classified into reference relationship data sets and supporting relationship data sets respectively. The data sets can be organized using data structures such as lists and dictionaries to facilitate subsequent storage and query. In addition to extracting the relationship data sets between files, the system also needs to obtain the timeliness parameters and effectiveness level parameters of each file. These parameters can be obtained from file metadata, file content or external knowledge bases. For example, for legal and regulatory documents, the release time information can be extracted from the file name, and timeliness-related keywords such as "currently valid" and "repealed" can be identified from the file content. The legal hierarchy information can be obtained from the legal knowledge graph to determine the effectiveness level.

[0036] During this step, problems such as inaccurate recognition of file identifiers and incomplete relationship extraction may occur. The system can address these issues through algorithm optimization and knowledge base expansion. For example, when constructing file identifier pairs, in addition to considering the file name and number, information such as the issuing agency and publication date can also be incorporated to improve the accuracy of file identification. During contextual analysis, pre-trained language models can be used in conjunction with domain knowledge to better understand the semantic relationships between file identifier pairs.

[0037] S102, constructing an initial file relationship network based on the reference relationship dataset and the supporting relationship dataset; The system constructs an initial file relationship network based on the reference relationship dataset and the supporting relationship dataset. Each file in the initial file relationship network serves as a node, and connections are established between nodes based on the reference relationship and the supporting relationship.

[0038] In this step, the system uses the extracted reference relationship dataset and supporting relationship dataset to build an initial file relationship network model. In this network model, each file is regarded as a node, and the reference relationship and supporting relationship between files are reflected as connections between nodes.

[0039] The process of constructing the initial file relationship network can be directly based on the relationship dataset, mapping files as nodes and relationships as edges. At the same time, to enrich the semantics of the network model, additional information such as node attributes and edge types can also be considered.

[0040] Because file relationship networks in real-world scenarios can be large and complex, performance bottlenecks can occur during construction. To address this, the system can adopt an incremental construction strategy, starting with a small subgraph and then gradually expanding the network. Furthermore, graph database technology can be utilized to improve the efficiency of network construction and operation, leveraging its storage and query capabilities.

[0041] S103. Calculate the comprehensive weight value of each node based on the timeliness parameter and the effectiveness level parameter, and assign the comprehensive weight value to the corresponding node; The system calculates the comprehensive weight of each node based on the timeliness parameters and effectiveness level parameters. Specifically, a timeliness scoring matrix is constructed based on the timeliness parameters and effectiveness level parameters. The rows of the timeliness scoring matrix represent the file status, including currently valid, about to take effect, about to expire, and abolished, and the columns of the timeliness scoring matrix represent the scoring time period. Based on the current time's position in the scoring time period, the timeliness score of the corresponding file status is obtained from the timeliness scoring matrix. A effectiveness level scoring vector is constructed, with different effectiveness levels in the effectiveness level scoring vector corresponding to different basic scores. Based on the effectiveness level of the file, an effectiveness level score is obtained from the effectiveness level scoring vector. The timeliness score and effectiveness level score are weighted and summed to obtain a comprehensive weight value. The comprehensive weight value is then assigned to the corresponding node.

[0042] In this step, the system calculates a comprehensive weight based on the timeliness and effectiveness parameters of each file node and assigns this weight to the corresponding node. This weight reflects the importance of the file in the entire relationship network.

[0043] The process for calculating the comprehensive weight value can be flexibly designed. One possible implementation involves first quantifying the timeliness parameter and the effectiveness level parameter to obtain two scoring vectors, and then performing a weighted summation of these two scoring vectors to obtain the final comprehensive weight value. The scoring and weighted summation strategies can be appropriately designed based on the specific application scenario.

[0044] During this step, problems such as missing parameters and inconsistent scoring criteria may occur, affecting the accuracy of weight calculation. To address these issues, the system can be optimized by introducing missing value processing mechanisms and building a unified scoring system. Furthermore, consideration can be given to introducing other parameters, such as document quality and influence, to enrich the factors considered in weight calculation.

[0045] S104, weighting the connections in the initial file relationship network using the comprehensive weight value to obtain a weighted file network; The system uses the comprehensive weight value to weight the connections in the initial file relationship network to obtain a weighted file network. The weight of the connection in the weighted file network is calculated by the difference between the comprehensive weight values of the connected nodes.

[0046] The goal of this step is to use the node weights calculated in the previous step to weight the edges in the initial file relationship network, resulting in a weighted file network model. This weighted network can more accurately describe the relative importance of nodes compared to the initial network.

[0047] A straightforward weighting approach is to use the difference in the combined weights of connected nodes as the weight of the connecting edge. This method is simple to use, but may not accurately reflect the relative importance of nodes. To optimize weighting, the system can also consider introducing other weighting strategies, such as using the ratio of node weights or the harmonic mean as a method for calculating edge weights.

[0048] During this step, problems may arise, such as excessive weighting of the network or significant variations in edge weights, which can affect subsequent analysis. To address these issues, the system can normalize edge weights, mapping them to a smaller range. Furthermore, network simplification methods can be employed, such as removing edges with low weights and merging edges with similar weights, to appropriately reduce network complexity without compromising key information.

[0049] S105, converting the weighted file network into a directed time sequence graph, and determining the direction of the connection from the node with earlier time to the node with later time by comparing the time attributes of the connected nodes, thereby obtaining a directed weighted time sequence graph; This step requires further converting the weighted document network into a directed weighted time-series graph model. Compared to a typical weighted network, a time-series graph introduces an additional time dimension, reflecting the dynamic evolution of the network structure.

[0050] During the conversion process, the system uses the time attributes of file nodes, such as release time and effective time, to determine the direction of connections between nodes. Generally, it can be assumed that older file nodes influence later file nodes, so the connection direction is defined as from the earlier node to the later node.

[0051] A technical difficulty in this step is how to handle situations where time attributes are missing or time information conflicts. To address this issue, the system can refer to other time-related node attributes, such as the year information contained in the file identifier, to infer the temporal relationship of file nodes. For situations where time information conflicts occur, such as when file A references file B, but the timestamp of file A is earlier than that of file B, the system can set a threshold to ignore conflicting edges with smaller time differences, or introduce a manual inspection mechanism to verify the temporal relationship of key file nodes.

[0052] S106. In the directed weighted time-series graph, calculating a derived index of the node based on the sum of the in-degree connection weight and the out-degree connection weight of each node; In this step, the system needs to calculate a derivative index based on the sum of the in-degree connection weight and the out-degree connection weight of each node in the directed weighted time-series graph to reflect the node's propagation influence in the network.

[0053] Specifically, the sum of a node's in-degree weights reflects the total influence it receives from other nodes, while the sum of its out-degree weights reflects the total influence it exerts on other nodes. The derivative index can be expressed as a function of the sum of the in-degree weights and the sum of the out-degree weights, such as their ratio or difference.

[0054] The construction of this indicator can be an iterative optimization process. The initial indicator design may be too simple and fail to accurately capture the influence of a node. To address this issue, the system can continuously optimize the indicator construction method in practice. For example, a time decay factor can be introduced into the calculation process to assign different weights to influences at different times; or the topological characteristics of the node in the network, such as centrality and community structure, can be taken into account to modify the node's derived index.

[0055] S107, classifying the nodes in the directed weighted time series graph according to the derivative index to obtain a source node set, a derivative node set, and a terminal node set; In this step, the system uses the node derivation index calculated in the previous step to classify all nodes in the directed weighted time series graph into three categories: source node set, derived node set, and terminal node set. Source nodes are nodes with only out-degree connections but no in-degree connections, corresponding to the starting point of information propagation; derived nodes are nodes with both in-degree and out-degree connections, corresponding to the intermediate links of information propagation; and terminal nodes are nodes with only in-degree connections but no out-degree connections, corresponding to the end point of information propagation.

[0056] Node classification can be achieved based on a derivative index threshold. For example, nodes with a derivative index of 0 are classified as source nodes, nodes with a positive derivative index are classified as derivative nodes, and nodes with an infinite derivative index are classified as terminal nodes. The threshold can be adjusted based on actual application requirements and the distribution of node derivative indices.

[0057] During this step, the classification of some nodes may be ambiguous, with the derivative index very close to the threshold. To address this issue, the system can consider introducing fuzzy classification techniques, allowing nodes to have membership in multiple categories. Furthermore, the classification results can be post-processed, for example, by appropriately merging derivative nodes directly connected to the source or terminal nodes to simplify the network structure.

[0058] S108 , taking a node in the source node set as a root node, and constructing a hierarchical evolution tree based on the connection relationship in the directed weighted time sequence graph.

[0059] The system takes the nodes in the source node set as the root node and constructs a hierarchical evolution tree based on the connection relationship in the directed weighted time series graph. In the hierarchical evolution tree, the source node is the first level, the derived nodes directly connected to the source node are the second level, and the terminal node is the last level.

[0060] The goal of this step is to use the source node set and the connection relationship in the directed weighted time series graph to build a hierarchical evolution tree model to intuitively display the information evolution and propagation process starting from the source node.

[0061] When constructing a hierarchical evolutionary tree, the system can use graph traversal algorithms such as breadth-first search or depth-first search, starting from the source node and searching layer by layer for its directly or indirectly connected derivative nodes and terminal nodes. During the search, the node's hierarchical information and connection weights can be recorded to construct the evolutionary tree structure.

[0062] To optimize the readability and analytical value of the evolutionary tree, the system can consider introducing some post-processing and optimization strategies. For example, for overly large evolutionary trees, appropriate pruning can be performed to remove branches with weaker propagation influence; for levels with too sparse nodes, merging them with adjacent levels can be considered; for nodes with complex connections, display priorities can be set based on connection weights to highlight important propagation paths.

[0063] In this step, there may be too many source nodes, resulting in an overly complex evolutionary tree. To address this, the system can consider setting filtering conditions for source nodes, selecting representative nodes as the root nodes of the evolutionary tree. Furthermore, the system can support merging the evolutionary processes of multiple source nodes to generate a comprehensive evolutionary tree, which globally displays the evolutionary context of the entire file relationship network.

[0064] In the above embodiment, by extracting the references and matching relationships between files and combining the timeliness parameters and effectiveness level parameters to construct a weighted directed time series graph, the multi-dimensional association characteristics between file resources can be accurately reflected. By calculating the node derivative index based on the in-degree and out-degree connection weights, the role positioning of the file in the evolution process can be objectively identified. The nodes are divided into three categories: source, derivative, and terminal, and a hierarchical evolution tree is constructed to make the evolution of file resources clearly visible. The association evolution relationship between files is depicted through the graph structure, and the influence of timeliness and effectiveness level on the association strength is taken into account. The construction process of the hierarchical evolution tree follows the timeliness and association of files, which improves the rationality of the classification results. This classification method depicts the temporal evolution law and hierarchical derivative relationship between file resources, realizes the dynamic classification management of file resources, and provides users with more systematic and in-depth file association analysis results.

[0065] The above embodiment realizes the basic classification of file resources by constructing a hierarchical evolution tree. However, in order to further improve the accuracy and practicality of the classification, the hierarchical evolution tree needs to be optimized and deeply analyzed. Figure 2 , describes another file resource classification method based on knowledge graph in the embodiment of this application: please refer to Figure 2 , is another flow chart of a file resource classification method based on knowledge graph in an embodiment of the present application.

[0066] S201, calculating the hierarchical aggregation degree of nodes at each level in the hierarchical evolution tree; The system calculates the hierarchical clustering of nodes at each level in the hierarchical evolution tree. The hierarchical clustering represents the degree of association between nodes at the same level, specifically including: extracting possible connection paths between nodes at the same level; calculating the path weight of each connection path, which is the cumulative value of the connection weights on the connection path; constructing a hierarchical connectivity matrix, in which the elements of the hierarchical connectivity matrix are the path weights between corresponding node pairs; calculating the clustering coefficient of the nodes based on the hierarchical connectivity matrix; and taking the mean of the clustering coefficients of all nodes at the same level as the hierarchical clustering to obtain the hierarchical clustering of nodes at each level.

[0067] In this step, the system quantifies the degree of clustering of nodes at each level of the hierarchical evolution tree and calculates a hierarchical clustering index. This index reflects the close connections between nodes within a hierarchy and serves as an important basis for evaluating the rationality of hierarchical divisions. There are various ways to calculate hierarchical clustering, such as based on node attribute similarity or node connection density.

[0068] In specific implementation, the system can use a connection path-based method to calculate the hierarchical clustering. First, the system extracts all possible connection paths between nodes in the same level. The connection path can be a direct connection or an indirect connection through an intermediate node. Then, the system calculates the weight of each connection path. The path weight can be expressed as the cumulative value of the weights of each connection on the path, reflecting the closeness of the connection between the nodes. Next, the system constructs a hierarchical connectivity matrix. The elements in the matrix are the path weights between the corresponding node pairs. If there is no connection path between the two nodes, the corresponding element is 0. Finally, the system calculates the clustering coefficient of each node based on the hierarchical connectivity matrix. The clustering coefficient reflects the connection density between the node and its neighboring nodes. It can be expressed as the ratio of the number of triangle structures in which the node participates to the number of all possible triangle structures. Taking the average of the clustering coefficients of all nodes in the same level, the hierarchical clustering of that level is obtained.

[0069] S202, performing branch optimization on the hierarchical evolution tree based on the hierarchical aggregation degree to obtain an optimized hierarchical evolution tree; The goal of this step is to use the hierarchical clustering index calculated in the previous step to optimize the structure of the original hierarchical evolutionary tree, resulting in a new evolutionary tree with clearer layers and more rational branches. The core idea of this optimization is to identify layers with lower clustering and adjust their branches, re-dividing less relevant nodes into other layers or forming new independent branches.

[0070] In specific implementation, the system can traverse the hierarchical evolution tree from top to bottom and analyze the hierarchical clustering layer by layer. For the level with a clustering degree lower than the preset threshold, the system first extracts the node with the lowest clustering coefficient in the level, and then migrates it to other levels with the highest relevance based on the content attributes and connection relationships of the node. If the node has a low correlation with other levels, it will be separated into a new branch. During the node migration process, the system needs to simultaneously update the weight information of the node's level and related connections. The above process is recursively executed until the clustering degree of all levels reaches the preset threshold, or no further optimization is possible, and the optimized hierarchical evolution tree is finally obtained.

[0071] S203, calculating the structural importance index of the nodes in the optimized hierarchical evolution tree; The system calculates the structural importance index of the nodes in the optimized hierarchical evolution tree, and the structural importance index is determined based on the position, number of connections and connection weight of the nodes.

[0072] The structural importance index reflects the status and influence of a node in the entire evolutionary tree and can serve as an important reference for subsequent hierarchical management. The calculation of the structural importance index comprehensively considers the topological and semantic properties of the node, such as the node's hierarchical depth, number of connections, connection weight, and text relevance.

[0073] In specific implementation, the system can use a method based on network centrality to calculate node importance. Common centrality indicators include degree centrality, betweenness centrality, closeness centrality, etc. Degree centrality considers the number of direct connections of a node, betweenness centrality considers the frequency of a node's appearance on the shortest path, and closeness centrality considers the average distance between a node and other nodes. When calculating centrality indicators, the system can also combine connection weight information to assign different weight factors to different connection types, highlighting the importance of strong correlation connections. In addition, the system can also combine the semantic attributes of the node, such as text similarity, topic distribution, etc., to make appropriate corrections to the centrality indicators to obtain a more comprehensive and accurate assessment of node importance.

[0074] S204. Perform hierarchical management on file resources according to structural importance indicators.

[0075] The goal of this step is to use the node structure importance index calculated in the previous step to perform hierarchical management of file resources in the hierarchical evolution tree, achieving more refined organization and application. The core idea of hierarchical management is to divide file resources into different levels based on the level of node importance index, and adopt differentiated management strategies for files of different levels, such as storage priority, access rights, and lifecycle management.

[0076] In specific implementation, the system can set multiple importance levels, such as high, medium, and low, based on the distribution of node importance indicators. Nodes with importance indicators above the upper threshold are classified as high-level files; nodes below the lower threshold are classified as low-level files; and nodes between the two thresholds are classified as medium-level files. After the levels are divided, the system can perform differentiated processing on files of different levels according to preset management policies. For example, for high-level files, the system can take measures such as high-availability storage and hot backup to ensure their security and accessibility; for low-level files, the system can use low-cost cold storage and regularly archive and clean them. At the same time, the system can also dynamically adjust the access rights and display priorities of files based on their levels to promote the full utilization of important files.

[0077] In the above embodiment, the hierarchical evolution tree is optimized and hierarchically managed by calculating the hierarchical aggregation and structural importance index. The hierarchical aggregation reflects the degree of association between nodes at the same level. Branch optimization based on this index can make the hierarchical structure more reasonable. The structural importance index comprehensively considers the position, number of connections and connection weight of the node, and comprehensively reflects the importance of the node in the network structure. This optimization and grading method makes the structure of the hierarchical evolution tree more consistent with the actual association characteristics of the file resources, and avoids the one-sidedness of relying solely on the hierarchical position for importance judgment. Hierarchical management through structural importance indicators can more accurately reflect the importance of the file in the entire file system, and provide a scientific quantitative basis for the differentiated management of file resources.

[0078] Furthermore, in another embodiment, after hierarchically managing file resources according to structural importance indicators, it also includes: constructing an evolutionary feature vector of the file resource, the evolutionary feature vector includes the hierarchical position of the node in the optimized hierarchical evolution tree, the structural importance indicator and the associated node information; calculating the evolutionary similarity matrix of the file resource based on the evolutionary feature vector; analyzing the evolutionary similarity matrix using a clustering algorithm to identify file groups with similar evolutionary patterns; constructing an independent evolutionary prediction model for each file group, specifically including: extracting the historical evolutionary sequence of each file group, the historical evolutionary sequence includes the generation time, expiration time and association relationship change information of the file; constructing a time window feature matrix, and segmenting the historical evolutionary sequence according to a preset time window size; training an evolutionary trend predictor based on the time window feature matrix to obtain an independent evolutionary prediction model corresponding to each file group.

[0079] In this embodiment, after completing hierarchical management of file resources, the system further explores and predicts the evolution patterns of these resources. By constructing evolutionary feature vectors for each file resource, calculating the evolutionary similarity between file resources, identifying groups of files with similar evolutionary patterns, and constructing independent evolutionary prediction models for each file group, the system achieves accurate prediction and management of file resource evolution trends.

[0080] In specific implementation, the system first constructs an evolutionary feature vector for each file resource. The evolutionary feature vector is a comprehensive description of the file resource's multi-dimensional attributes, such as its position, importance, and association relationships in the hierarchical evolution tree. The hierarchical position reflects the file's hierarchical depth and branch in the evolutionary tree; the structural importance index reflects the file's influence in the entire evolutionary tree; and the associated node information records the association relationship between the file and other files, such as citations, citations, and the same level. Through the evolutionary feature vector, the system can fully capture the static characteristics and dynamic changes of file resources in the evolutionary tree.

[0081] After obtaining the evolutionary feature vectors, the system calculates the evolutionary similarity matrix between file resources. The evolutionary similarity matrix measures the degree of similarity between different file resources in their evolutionary patterns, namely, their consistency in terms of their position in the hierarchical evolutionary tree, trends in importance changes, and evolution of associated relationships. Evolutionary similarity can be calculated using a variety of metrics, such as Euclidean distance, cosine similarity, and Jaccard similarity. A time decay factor can also be introduced to assign different weights to similarities in different time periods. Through the evolutionary similarity matrix, the system can characterize the correlations and differences in the evolutionary patterns of file resources.

[0082] Based on the evolution similarity matrix, the system uses a clustering algorithm to group file resources and identify groups of files with similar evolution patterns. Common clustering algorithms include K-Means, hierarchical clustering, and DBSCAN. Through cluster analysis, the system can automatically discover file resources with similar evolution patterns and reveal their inherent evolution patterns and trends.

[0083] For each identified file group, the system further constructs an independent evolution prediction model to predict the future evolution trend of the file resources in that group. The process of constructing the evolution prediction model includes the following steps: Extract the historical evolution sequence of the file group. The historical evolution sequence records the creation time, expiration time, and changes in the association relationship of each file in the file group, reflecting the complete evolution trajectory of the file group.

[0084] Construct a time window feature matrix. The system segments the historical evolution sequence according to the preset time window size (such as month, quarter, etc.) and extracts the feature representation of the file group within each time window, such as the number of files, the number of associations, and the importance distribution, to form a time window feature matrix.

[0085] The system trains an evolution trend predictor based on the time window feature matrix. The system selects an appropriate machine learning model, such as a time series prediction model or a regression model, and uses the time window feature matrix as input to train the evolution trend predictor. Through training, the predictor can capture the temporal patterns and trends of file group evolution, enabling prediction and inference of future evolution.

[0086] In the above embodiment, by constructing an evolutionary feature vector that includes hierarchical position, structural importance, and associated node information, calculating the evolutionary similarity matrix of file resources and analyzing it using a clustering algorithm, the system can identify file groups with similar evolutionary patterns from multiple dimensions. The evolutionary feature vector comprehensively characterizes the characteristic attributes of file resources in the hierarchical evolution tree, so that the similarity calculation takes into account both the structural characteristics of the file and the temporal evolution characteristics. Cluster analysis based on the similarity matrix can automatically discover the evolutionary patterns between file resource groups. This pattern reflects the common characteristics of different types of file resources in the process of business development. By establishing an independent evolutionary prediction model for each file group, the evolutionary characteristics of different types of file resources can be more accurately characterized, thereby improving the adaptability and prediction accuracy of the file resource classification system to business changes.

[0087] The following describes the system in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , which is a schematic diagram of the physical device structure of a file resource classification system based on knowledge graph provided in an embodiment of the present application.

[0088] It should be noted that Figure 3 The structure of the system shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0089] like Figure 3 As shown, the system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes, such as the methods described in the above embodiments, based on programs stored in a read-only memory (ROM) 302 or programs loaded from a storage unit 308 into a random access memory (RAM) 303. RAM 303 also stores various programs and data required for system operation. CPU 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to bus 304.

[0090] The following components are connected to the I / O interface 305: an input section 306 including a camera, infrared sensor, and the like; an output section 307 including a liquid crystal display (LCD) and speakers; a storage section 308 including a hard disk and the like; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. Removable media 311, such as a magnetic disk, optical disk, magneto-optical disk, or semiconductor memory, is installed in the drive 310 as needed, so that computer programs read from the media can be installed in the storage section 308 as needed.

[0091] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for executing the methods illustrated in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section 309 and / or installed from removable media 311. When executed by the central processing unit (CPU) 301, the computer program performs the various functions defined in the present invention.

[0092] It should be noted that the computer-readable medium described in the embodiments of the present invention may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium may include a data signal transmitted in baseband or as part of a carrier wave, which carries a computer-readable computer program. Such a propagated data signal may take any of a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof.

[0093] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the boxes can also occur in an order different from that marked in the accompanying drawings. For example, two boxes shown in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0094] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the system described in the above embodiments, or may exist independently and not incorporated into the system. The storage medium carries one or more computer programs, and when executed by a processor of a system, the system implements the methods provided in the above embodiments.

[0095] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0096] As used in the above embodiments, the term “when…” may be interpreted as “if…” or “after…” or “in response to determining…” or “in response to detecting…”, depending on the context. Similarly, the phrases “upon determining…” or “if (stated condition or event) is detected” may be interpreted as “if determining…” or “in response to determining…” or “upon detecting (stated condition or event)” or “in response to detecting (stated condition or event)”, depending on the context.

[0097] In the above embodiments, all or part of the embodiments can be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., a floppy disk, hard disk, tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive).

[0098] Those skilled in the art will appreciate that all or part of the process steps in the above-described method embodiments can be implemented by a computer program instructing the relevant hardware. The program can be stored in a computer-readable storage medium, and when executed, the program can include the process steps in the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A file resource classification method based on knowledge graph, characterized in that: include: Extracting reference relationship datasets and supporting relationship datasets between files, as well as timeliness parameters and effectiveness level parameters of each file from a preset file resource library; Constructing an initial file relationship network based on the reference relationship dataset and the matching relationship dataset, wherein each file in the initial file relationship network serves as a node, and connections are established between the nodes based on the reference relationship and the matching relationship; Calculating a comprehensive weight value of each of the nodes based on the timeliness parameter and the effectiveness level parameter, and assigning the comprehensive weight value to the corresponding node; Using the comprehensive weight value to weight the connections in the initial file relationship network to obtain a weighted file network, wherein the weights of the connections in the weighted file network are calculated by the difference between the comprehensive weight values of the connected nodes; Converting the weighted file network into a directed time sequence graph, and determining the direction of the connection from the node with earlier time to the node with later time by comparing the time attributes of the connected nodes, thereby obtaining a directed weighted time sequence graph; In the directed weighted time series graph, calculating the derived index of the node based on the sum of the in-degree connection weight and the sum of the out-degree connection weight of each node; Classifying the nodes in the directed weighted time series graph according to the derivative index to obtain a source node set, a derivative node set, and a terminal node set; Taking the node in the source node set as the root node, a hierarchical evolution tree is constructed based on the connection relationship in the directed weighted time series graph, in which the source node is used as the first level, the derived nodes directly connected to the source node are used as the second level, and the terminal node is used as the last level.

2. The method according to claim 1, characterized in that The step of extracting the reference relationship dataset and the supporting relationship dataset between files from the preset file resource library specifically includes: Perform word segmentation on the file contents in the preset file resource library to obtain a word sequence; identifying a file identification word from the word sequence based on a preset file identification dictionary, the file identification word including a file name, a file number, and a release time; Constructing a file identification pair, wherein the file identification pair includes a reference file identification and a referenced file identification; Performing contextual scanning in the file content based on the file identification pair to extract a text segment containing the file identification pair; Performing semantic analysis on the text fragment to identify the relationship type between the file identifier pairs, wherein the relationship type includes a reference relationship and a matching relationship; The file identification pairs are classified into a reference relationship dataset and a supporting relationship dataset according to the relationship type.

3. The method according to claim 1, characterized in that The calculating of the comprehensive weight value of each node based on the timeliness parameter and the effectiveness level parameter specifically includes: Constructing a timeliness scoring matrix based on the timeliness parameter and the effectiveness level parameter, wherein the rows of the timeliness scoring matrix represent the file status, including currently valid, about to take effect, about to expire, and revoked, and the columns of the timeliness scoring matrix represent the scoring time period; Obtaining a timeliness score corresponding to the file status from the timeliness scoring matrix according to the position of the current time in the scoring time period; Constructing an effectiveness level scoring vector, wherein different effectiveness levels in the effectiveness level scoring vector correspond to different basic scores; obtaining a effectiveness level score from the effectiveness level scoring vector according to the effectiveness level of the document; The timeliness score and the effectiveness level score are weighted and summed to obtain a comprehensive weight value.

4. The method according to claim 1, wherein After constructing the hierarchical evolution tree based on the connection relationship in the directed weighted time sequence graph, the method further includes: Calculating the hierarchical clustering degree of nodes at each level in the hierarchical evolution tree, wherein the hierarchical clustering degree represents the degree of association between nodes at the same level; Performing branch optimization on the hierarchical evolution tree based on the hierarchical aggregation degree to obtain an optimized hierarchical evolution tree; Calculating a structural importance index of the node in the optimized hierarchical evolution tree, wherein the structural importance index is determined based on the position, number of connections, and connection weight of the node; File resources are managed hierarchically according to the structural importance index.

5. The method according to claim 4, characterized in that The calculating of the hierarchical aggregation degree of nodes at each level in the hierarchical evolution tree specifically includes: Extract possible connection paths between nodes in the same level; Calculating a path weight of each connection path, where the path weight is a cumulative value of the weights of each connection on the connection path; Constructing a hierarchical connectivity matrix, wherein the elements in the hierarchical connectivity matrix are the path weights between corresponding node pairs; Calculating the clustering coefficient of the nodes based on the hierarchical connectivity matrix; The mean of the clustering coefficients of all nodes in the same level is taken as the level clustering degree to obtain the level clustering degree of nodes in each level.

6. The method according to claim 4, characterized in that After performing hierarchical management on the file resources according to the structural importance index, the method further includes: Constructing an evolution feature vector of the file resource, wherein the evolution feature vector includes the hierarchical position, structural importance index, and associated node information of the node in the optimized hierarchical evolution tree; Calculating an evolution similarity matrix of the file resource based on the evolution feature vector; Analyzing the evolution similarity matrix using a clustering algorithm to identify file groups with similar evolution patterns; An independent evolution prediction model is constructed for each of the file groups.

7. The method according to claim 6, characterized in that The step of constructing an independent evolution prediction model for each file group specifically includes: Extracting a historical evolution sequence of each file group, wherein the historical evolution sequence includes file generation time, expiration time, and association relationship change information; Constructing a time window feature matrix and segmenting the historical evolution sequence according to a preset time window size; An evolution trend predictor is trained based on the time window feature matrix to obtain an independent evolution prediction model corresponding to each of the file groups.

8. A file resource classification system based on knowledge graph, characterized in that: The system comprises: One or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the system to execute the method according to any one of claims 1 to 7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a system, the system is caused to perform the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on a system, the system is caused to perform the method according to any one of claims 1 to 7.

Images