Automatic filling review and knowledge base integration optimization method based on a dynamic rule engine

Abstract

The present invention provides an automatic filling-in review and knowledge base integration optimization method based on a dynamic rule engine, which relates to the technical field of software engineering. It includes standardizing electronic form data and generating a dataset to be reviewed based on preset filling-in rules. The constructed dynamic rule engine realizes the temporal tracking and adaptive reorganization of rules, generates a scenario-based rule chain for rule matching, and finally obtains the review result. This method significantly improves the accuracy and efficiency of filling-in review by constructing multi-level knowledge modeling and a knowledge dual network, detecting potential knowledge conflicts and generating repair strategies.

Description

Technical Field

[0001] The present invention relates to software engineering technology, and particularly to an automatic filling-in review and knowledge base integration optimization method based on a dynamic rule engine. Background Art

[0002] The automatic filling-in review and knowledge base integration of electronic forms is an important link in modern information management. With the in-depth development of enterprise digital transformation, more and more business processes rely on the processing and analysis of electronic form data. The traditional form review method mainly relies on manual review, which is inefficient and error-prone. There are still some defects and deficiencies in the existing automatic filling-in review and knowledge base integration technology based on a rule engine:

[0003] Insufficient flexibility in rule configuration: The existing rule engines usually adopt a static rule configuration method, which is difficult to adapt to the changes in complex business scenarios. When the business rules change, it is necessary to manually modify the rule configuration, resulting in high maintenance costs and prone to introducing human errors.

[0004] High maintenance cost of the knowledge base: The construction and maintenance of the knowledge base require a large amount of human and time costs. With the continuous development of the business, the scale of the knowledge base will continue to expand, making the update and maintenance of knowledge more difficult.

[0005] Difficult detection and repair of knowledge conflicts: In practical applications, there may be some contradictory or conflicting knowledge in the knowledge base. These knowledge conflicts will lead to inconsistent execution results of the rule engine and even cause incorrect decisions. The existing technologies lack effective knowledge conflict detection and repair mechanisms, making it difficult to ensure the consistency and integrity of the knowledge base. Summary of the Invention

[0006] The embodiments of the present invention provide an automatic filling-in review and knowledge base integration optimization method based on a dynamic rule engine, which can solve the problems in the existing technologies.

[0007] In the first aspect of the embodiments of the present invention,

[0008] An automatic filling-in review and knowledge base integration optimization method based on a dynamic rule engine is provided, including:

[0009] Performing format standardization processing on the electronic form data to obtain preprocessed data, and performing structured analysis on the preprocessed data according to the preset filling-in rules to generate a dataset to be reviewed;

[0010] Constructing a dynamic rule engine based on a rule time series graph, performing time series tracking on the rules, decomposing the rules into atomic rule units, adaptively reorganizing the atomic rule units according to the business scenario to generate a scenario-based rule chain, and using the scenario-based rule chain to perform rule matching on the dataset to be reviewed to obtain a review result;

[0011] Perform multi-level knowledge modeling on the review results to construct a vertically layered knowledge system; establish a knowledge dual network based on the knowledge system, calculate the cross-entropy time series distribution of the original network and the dual network, fit and extrapolate the trend of the entropy value change, combine the dynamic prediction of network consistency, detect and locate potential knowledge conflicts, generate repair strategies, and update the repaired knowledge to the knowledge base.

[0012] In an alternative implementation,

[0013] The steps of constructing a dynamic rule engine based on a rule time series graph, performing time series tracking on the rules, decomposing the rules into atomic rule units, and adaptively reorganizing the atomic rule units according to the business scenario to generate a scenario-based rule chain include:

[0014] Construct a rule time series graph, represent rule node features based on spatio-temporal feature tensors; perform atomic decomposition of the rules based on semantic integrity; use scenario feature vectors to drive rule reorganization to generate a scenario-based rule chain, specifically including:

[0015] Construct a rule time series graph, use business rules as the nodes of the rule time series graph, use the time series dependence relationship between rules as the directed edges of the rule time series graph, construct a spatio-temporal feature tensor for the nodes, the spatio-temporal feature tensor includes the time series attributes, spatial constraints and business attributes of the rules, construct a rule time series matrix based on the spatio-temporal feature tensor, perform feature enhancement on the rule time series matrix, construct a multi-dimensional association matrix including rule time series features and business attributes, and perform time series tracking on the rules according to the time overlap degree and execution order relationship between rules in the multi-dimensional association matrix;

[0016] Perform atomic decomposition of the rules based on the rule time series graph, decompose the rules into atomic rule units that maintain semantic integrity, calculate the semantic correlation degree between the atomic rule units, determine the separation boundary of the atomic rule units when the semantic correlation degree is less than the preset boundary threshold, and generate a set of atomic rule units;

[0017] Construct a scenario feature vector, calculate the adaptation degree scores of each atomic rule unit in the set of atomic rule units and the scenario feature vector, and adaptively reorganize the atomic rule units according to the adaptation degree scores using a dynamic programming algorithm to generate a scenario-based rule chain.

[0018] In an alternative implementation,

[0019] Construct a spatio-temporal feature tensor for the node. The spatio-temporal feature tensor includes regular temporal attributes, spatial constraints, and business attributes. The steps of constructing a regular temporal matrix based on the spatio-temporal feature tensor and enhancing the features of the regular temporal matrix to construct a multi-dimensional association matrix including regular temporal features and business attributes are as follows:

[0020] Construct a spatio-temporal feature tensor for regular nodes. The spatio-temporal feature tensor includes a temporal feature vector, a spatial feature vector, and a business feature vector. The temporal feature vector includes the effective time, expiration time, maximum execution duration, and minimum execution duration of the rule. The spatial feature vector includes the scope of the rule, geographical location restrictions, and management levels. The business feature vector includes the priority of the rule, rule type, and business domain;

[0021] Perform a temporal dimension projection on the spatio-temporal feature tensor to obtain a temporal feature matrix. Calculate the time overlap degree and execution order relationship between rules to obtain the temporal relationship strength. Construct a regular temporal matrix based on the temporal feature matrix and the temporal relationship strength;

[0022] Calculate the similarity of the spatial feature vector to obtain the spatial association strength, and calculate the correlation of the business feature vector to obtain the business association strength;

[0023] Enhance the features of the regular temporal matrix. Perform multi-dimensional feature fusion on the regular temporal matrix, the spatial association strength, and the business association strength to generate an enhanced association matrix;

[0024] Construct a query matrix, a key-value matrix, and a value matrix. Calculate the attention weights based on the enhanced association matrix, and perform weighted fusion on the attention weights and the enhanced association matrix to generate a multi-dimensional association matrix including the high-order association relationship of rules.

[0025] In an alternative embodiment,

[0026] The steps of performing multi-level knowledge modeling on the review results and constructing a vertically layered knowledge system include:

[0027] Extract business entities, rule entities, result entities, and time entities from the review results. Construct an entity relationship set including trigger relationships, generation relationships, and temporal relationships. Extract an attribute set including business types, rule identifiers, result types, timestamps, and confidence levels;

[0028] Based on the entity relationship set and the attribute set, construct the conceptual hierarchical relationships of business, rules, results, and time. Calculate the semantic similarity between concepts. Construct a concept clustering tree according to the semantic similarity, and extract the semantic association rules between concepts;

[0029] Divide the knowledge system into a data layer, a rule layer, an inference layer, and an application layer. The data layer processes time-series data streams. The rule layer generates context-aware rules based on the time-series data streams. The inference layer performs state-dependent inference according to the context-aware rules. The application layer constructs scenario-based applications based on the state-dependent inference. Construct inter-layer context information using the semantic association rules, perform bottom-up knowledge refinement to obtain inter-layer knowledge mapping relationships, determine global constraint conditions based on the inter-layer knowledge mapping relationships, and perform top-down knowledge guidance.

[0030] Construct a knowledge evolution equation by combining external environmental information and internal consistency requirements, calculate the knowledge update gradient according to the knowledge evolution equation, and determine the adaptive update strategy of knowledge by combining the knowledge update gradient and the knowledge drift amount.

[0031] Calculate the difference degree of upper and lower layer knowledge representations and the degree of violation of structural constraints according to the inter-layer knowledge mapping relationship to obtain the consistency deviation. Statistically analyze the knowledge coverage of the target business scenario and the missing situation of the knowledge link to obtain the integrity index. Analyze the decision accuracy rate and scenario adaptation degree of knowledge application to obtain the effectiveness index. Construct a weighted objective function of the consistency deviation, integrity index, and effectiveness index. Set knowledge structure integrity and semantic consistency constraints based on the weighted objective function. Optimize the indicators that violate the constraints by combining gradient descent and coordinate alternating optimization methods. Dynamically adjust the weights of each indicator during the optimization process, and adaptively adjust the optimization step size according to the application feedback.

[0032] Calculate the accuracy rate, coverage rate, and timeliness of knowledge to obtain the knowledge quality evaluation result, and dynamically adjust the knowledge structure according to the evaluation result.

[0033] In an alternative embodiment,

[0034] The steps of establishing a knowledge dual network based on the knowledge system, calculating the cross-entropy time-series distribution of the original network and the dual network, curve fitting and extrapolating the trend of entropy value changes, combining the dynamic prediction of network consistency, detecting and locating potential knowledge conflicts, generating repair strategies, and updating the repaired knowledge to the knowledge base include:

[0035] Construct a knowledge dual network; calculate the dynamic cross-entropy based on the time decay factor; use the attention mechanism to fuse multi-dimensional prediction features, and evaluate the network consistency through hierarchical calculation from local to global; dynamically adjust the warning threshold based on the historical trigger frequency and activity; locate the conflict source and generate repair strategies according to the betweenness centrality, specifically including:

[0036] Map the association relationships between knowledge nodes into a knowledge graph structure, calculate the association strength between nodes based on the feature vector representation of nodes and the node association set, and construct a dual network by reversing the association direction between nodes in the original network and recalculating the weights;

[0037] Obtain the probability distributions of nodes in the original network and the dual network to calculate the network cross-entropy, introduce a time decay factor to dynamically weight the network cross-entropy, and obtain the cross-entropy sequence of the original network and the dual network within a preset time window; construct a polynomial kernel function to map the cross-entropy sequence to a high-dimensional feature space, extract the time-varying feature vector, and the time-varying feature vector includes the feature mapping components calculated by the kernel function; perform curve fitting on the time series sequence of the cross-entropy based on the time-varying feature vector, and perform extrapolation prediction on the fitted curve;

[0038] Construct multi-dimensional prediction features based on the historical data of the cross-entropy sequence, the network topology features, and the node semantic similarity, use the attention mechanism to dynamically fuse the multi-dimensional prediction features to obtain the fused features, perform hierarchical prediction according to the knowledge domain based on the fused features, perform network consistency evaluation through hierarchical calculation from local to global, calculate the adaptive threshold based on the historical trigger frequency and the knowledge update activity for early warning determination, and trigger the conflict detection at the corresponding level according to the early warning result;

[0039] Calculate the betweenness centrality of nodes in the original network and the dual network, locate the conflict source nodes based on the change amount of the betweenness centrality, and generate a repair strategy based on the graph edit distance for the conflict source nodes; update the repair strategy to the knowledge base.

[0040] In an alternative embodiment,

[0041] The steps of performing hierarchical prediction according to the knowledge domain based on the fused features, performing network consistency evaluation through hierarchical calculation from local to global, and calculating the adaptive threshold based on the historical trigger frequency and the knowledge update activity for early warning determination include:

[0042] Construct a multi-level fractal calculation framework, and use a bidirectional mapping matrix to describe the intra-layer knowledge association; fuse the semantic features and the topological features to construct a fractal recursive function, and quantify the hierarchical knowledge complexity through iterative calculation; perform multi-level joint prediction driven by fractal features; evaluate the global consistency based on the geometric mean of the inter-layer consistency covariance and the fractal features; dynamically adjust the early warning threshold according to the knowledge propagation influence degree, specifically including:

[0043] Construct a multi-level fractal calculation framework, divide it into multiple levels according to the professional degree and knowledge correlation degree of the knowledge field based on the fusion features, and construct a bidirectional mapping matrix at each level to characterize the correlation features between knowledge units within the level; extract multi-dimensional semantic features from the node set of each level, and fuse the multi-dimensional semantic features with the topological features of the knowledge graph to generate a relationship strength matrix of the nodes; construct a fractal recursive function based on the relationship strength matrix, and obtain the fractal features reflecting the knowledge complexity of the level through iterative calculation;

[0044] Perform hierarchical prediction driven by fractal features on the nodes in each level, splice the fusion feature vector of the node, the fractal features of the level to which it belongs, and the prediction information of the adjacent levels, and calculate the prediction features of the node through a parameter matrix and an activation function;

[0045] Calculate the local consistency index based on the similarity of the prediction features of the nodes between adjacent levels, and calculate the global consistency index by combining the weights of each level, the consistency covariance between levels, and the geometric mean of the fractal features;

[0046] Construct an adaptive threshold according to the historical trigger frequency and knowledge update activity, introduce the knowledge verification passing rate, knowledge dissemination coverage, and knowledge conversion rate to calculate the knowledge dissemination influence degree, and dynamically adjust the adaptive threshold based on the change rate of the knowledge dissemination influence degree and the global consistency index; trigger an alarm when the global consistency index is lower than the dynamically adjusted adaptive threshold.

[0047] In an alternative embodiment,

[0048] The steps of constructing a multi-level fractal calculation framework include:

[0049] Collect node semantic vectors and multi-dimensional relationship strengths, perform non-linear mapping on the node semantic vectors and multi-dimensional relationship strengths through a bidirectional mapping matrix to obtain initial node features, construct a fractal recursive function based on the initial node features, and the fractal recursive function includes the topological correlation degree and semantic similarity of the nodes; obtain hierarchical fractal features through iterative recursion, and the recursive calculation includes the propagation and diffusion of fractal features and the aggregation of features between levels, and the hierarchical fractal features optimize the global fractal dimension through dynamic programming methods;

[0050] Construct a multi-scale sliding time window for the hierarchical fractal features, introduce an adaptive time decay factor for different scale windows, and the time decay factor is dynamically adjusted according to the knowledge update frequency and association strength; calculate the dynamic fractal dimensions of nodes and hierarchies within each time window, and the dynamic fractal dimensions fuse historical dimension information through an exponentially weighted method; construct a time-series fractal feature sequence based on the dynamic fractal dimensions, and the time-series fractal feature sequence contains the evolution trajectory and change pattern of the knowledge structure; perform trend decomposition and periodic analysis on the time-series fractal feature sequence, extract the long-term trend and short-term fluctuation features of knowledge evolution, calculate the change rate of adjacent moments of the time-series fractal features, construct a multi-scale stability evaluation function in combination with the variance of the time-series fractal feature sequence, and calculate the stability index of the knowledge structure based on the multi-scale stability evaluation function;

[0051] Construct an update threshold function according to the stability index, trigger the knowledge update mechanism when the change rate of the time-series fractal feature exceeds the update threshold function, and determine the intensity of the knowledge update strategy based on the change rate of the time-series fractal feature and the stability index; feedback the execution result of the knowledge update strategy to the fractal feature calculation link for the dynamic update of the next round of fractal features.

[0052] The present invention innovatively introduces the fractal theory into the field of knowledge modeling and establishes a complete set of knowledge evolution perception and early warning methods. First, through the dual structures of the generative mind map and the dual network, the dynamic representation and verification of knowledge are realized; second, a multi-level fractal calculation framework is creatively constructed, and the knowledge complexity is quantified through fractal recursive functions to achieve an accurate description of the knowledge structure; third, a hierarchical consistency evaluation mechanism from local to global is innovatively designed, and combined with the knowledge propagation influence degree, a dynamic early warning system is established; finally, through multi-scale time window analysis and stability evaluation, the adaptive adjustment of the knowledge update strategy is realized.

[0053] The present invention can more accurately capture the dynamic evolution characteristics of the knowledge structure, more effectively identify and warn of knowledge conflicts, provide more accurate knowledge repair strategies, and realize the adaptive update and optimization of the knowledge base.

[0054] The present invention breaks through the limitations of traditional knowledge management methods, provides a new technical path for the dynamic maintenance and optimization of large-scale knowledge bases, and has important theoretical innovation value and practical application prospects. Brief Description of the Drawings

[0055] Figure 1 It is a schematic flowchart of the automatic filling-in review and knowledge base integration optimization method based on a dynamic rule engine in an embodiment of the present invention. Detailed Embodiments

[0056] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part rather than all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0057] The technical solutions of the present invention will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0058] Figure 1 The figure is a schematic flowchart of an automatic filling review and knowledge base integration optimization method based on a dynamic rule engine according to an embodiment of the present invention, as Figure 1 shown, the method includes:

[0059] S1. Perform format standardization processing on the electronic form data to obtain preprocessed data, perform structured analysis on the preprocessed data according to preset filling rules, and generate a dataset to be reviewed;

[0060] S2. Build a dynamic rule engine based on a rule time series graph, perform time series tracking on the rules, decompose the rules into atomic rule units, adaptively reorganize the atomic rule units according to the business scenario to generate a scenario-based rule chain, and use the scenario-based rule chain to perform rule matching on the dataset to be reviewed to obtain a review result;

[0061] S3. Perform multi-level knowledge modeling on the review result, build a vertically layered knowledge system; establish a knowledge dual network based on the knowledge system, calculate the cross-entropy time series distribution of the original network and the dual network, perform curve fitting and extrapolation on the entropy value change trend, combine the dynamic prediction of network consistency, detect and locate potential knowledge conflicts, generate a repair strategy, and update the repaired knowledge to the knowledge base.

[0062] Exemplarily, when performing format standardization processing on the electronic form data, first clean the original data, including removing special characters, unifying the date format, standardizing the numerical unit, etc. For example, convert dates in different formats to the "YYYY-MM-DD" format uniformly, and convert numerical values with thousands separators to pure numerical formats. For missing values, fill or mark them according to business rules. For example, fill the null values in financial data with "0", and mark the null values in descriptive fields with "not filled".

[0063] In the structured analysis phase, the preset filling rules are parsed to build a rule index tree. The preset filling rules include data format rules (such as mandatory item checks, field length rules, data type rules), logical relationship rules (such as in-table verification, cross-table association, year-on-year change), and business rules (such as range constraints, threshold tests, industry characteristics). For example, in the data format rules, the unified social credit code must be 18 digits; in the logical relationship rules, the total assets equal the sum of current assets and non-current assets; in the business rules, if the inventory turnover rate of manufacturing enterprises is lower than 0.5, it needs to be marked as abnormal.

[0064] In the implementation of the dynamic rule engine based on the rule time-series graph, the complex rules are first decomposed into atomic rule units. Taking the filling of enterprise annual reports as an example, the complex rule of "abnormal year-on-year growth rate of enterprise operating income" can be decomposed into atomic rule units such as validity check of operating income, extraction of historical data, calculation of growth rate, and threshold comparison.

[0065] Perform time-series marking on the atomic rule units to record the effective time, expiration time, and update history of the rules. Store this information through the time-series graph. Each rule node contains a timestamp attribute, and the edges between nodes represent the logical relationship and time-series dependence between rules.

[0066] In different business scenarios, dynamically combine atomic rule units to form scenario-based rule chains. Taking the review of quarterly reports of manufacturing enterprises as an example, the rule chain for production and operation indicators includes: verifying capacity utilization rate, checking inventory turnover, reconciling the relationship between raw material procurement volume and production volume, and analyzing the change in unit product cost. The rule chain for financial indicator association includes: checking the matching degree between sales revenue and accounts receivable, verifying the relationship between raw material procurement and accounts payable, reconciling the relevance between production volume and operating cost, and analyzing the reasonableness of period expenses.

[0067] During the rule matching process, the system automatically executes rule verification and generates hierarchical results. For example, in the reasonableness check of production cost, by calculating the unit product cost and comparing it with the industry average level, it is determined to be reasonable when the difference rate is less than 10%; in the income matching check, by comparing the theoretical income and the actual income, it is determined to be normal when the income realization rate is within the reasonable range.

[0068] When performing multi-level knowledge modeling for the review results, a vertically layered knowledge system is constructed according to business domains, data characteristics, and rule types. For example, knowledge is divided into a data format layer, a business rule layer, a cross-table association layer, etc. Knowledge nodes within each layer are connected through semantic associations and rule dependence relationships.

[0069] When establishing the knowledge dual network, the dual network is constructed by reversing the association direction in the original knowledge network. For example, the edge from the "total assets" node to the "asset - liability ratio" node in the original network becomes the edge from the "asset - liability ratio" node to the "total assets" node in the dual network.

[0070] When calculating the cross - entropy time - series distribution of the two networks, network state data at different time points are collected. For example, the cross - entropy value is calculated once a day to obtain a time series. Curve fitting is performed on this series to predict the evolution trend of the knowledge system. When there are abnormal fluctuations in the prediction curve, it indicates that there may be knowledge conflicts.

[0071] After discovering potential conflicts, the conflict source is located according to the centrality index of the nodes. For example, if the centrality of a certain rule node suddenly increases and new associations are generated with multiple nodes, this node is very likely to be the conflict source. Specific repair suggestions are generated for the identified conflicts, such as adjusting rule parameters, updating rule dependencies, etc.

[0072] The dynamic rule engine of the present invention can flexibly adjust the review strategy according to the business scenario; the knowledge dual network provides a verification mechanism, improving the accuracy of the knowledge base; the hierarchical knowledge modeling and conflict detection mechanism ensure the consistency and timeliness of the knowledge base.

[0073] In an alternative embodiment,

[0074] The steps of constructing a dynamic rule engine based on a rule time - series graph, performing time - series tracking on rules, decomposing the rules into atomic rule units, and adaptively reorganizing the atomic rule units according to the business scenario to generate a scenario - based rule chain include:

[0075] Constructing a rule time - series graph, representing rule node features based on spatio - temporal feature tensors; performing atomization decomposition of rules based on semantic integrity; using scenario feature vectors to drive rule reorganization to generate a scenario - based rule chain, specifically including:

[0076] Constructing a rule time - series graph, taking business rules as the nodes of the rule time - series graph, taking the time - series dependence relationship between rules as the directed edges of the rule time - series graph, constructing a spatio - temporal feature tensor for the nodes, where the spatio - temporal feature tensor includes the time - series attributes, spatial constraints, and business attributes of the rules, constructing a rule time - series matrix based on the spatio - temporal feature tensor, performing feature enhancement on the rule time - series matrix, constructing a multi - dimensional association matrix containing rule time - series features and business attributes, and performing time - series tracking on the rules according to the time overlap degree and execution order relationship between rules in the multi - dimensional association matrix;

[0077] Atomize and decompose the rules based on the rule time-series graph, decompose the rules into atomic rule units that maintain semantic integrity, calculate the semantic correlation degree between the atomic rule units, determine the separation boundary of the atomic rule units when the semantic correlation degree is less than the preset boundary threshold, and generate a set of atomic rule units;

[0078] Construct a scenario feature vector, calculate the fitness scores of each atomic rule unit in the set of atomic rule units and the scenario feature vector, and adaptively reorganize the atomic rule units using the dynamic programming algorithm based on the fitness scores to generate a scenario-based rule chain.

[0079] Exemplarily, first, construct a rule time-series graph. Take the business rules as the nodes of the rule time-series graph, and take the temporal dependence relationship between the rules as the directed edges. The spatio-temporal feature tensor of each node includes the temporal attributes, spatial constraints, and business attributes of the rule. By comprehensively analyzing these features, construct a rule time-series matrix. This matrix not only contains the temporal features of the rules but also combines the business attributes. Next, perform feature enhancement on the rule time-series matrix to generate a multi-dimensional correlation matrix. The multi-dimensional correlation matrix contains the time overlap degree and execution order relationship between the rules, thereby realizing the temporal tracking of the rules.

[0080] Secondly, perform atomization decomposition based on the rule time-series graph. Decompose the rules into atomic rule units that maintain semantic integrity. Determine the separation boundary of the atomic rule units by calculating the semantic correlation degree between the atomic rule units. When the semantic correlation degree is lower than the preset boundary threshold, mark the separation boundary of the atomic rule units, and finally generate a set of atomic rule units.

[0081] Next, construct a scenario feature vector. The scenario feature vector forms a multi-dimensional vector representation reflecting the characteristics of a specific business scenario by extracting the key attributes of the business scenario (such as industry type, reporting period type, enterprise scale, review level, etc.) and quantifying them into numerical features. Calculate the fitness scores of each atomic rule unit in the set of atomic rule units and the scenario feature vector. The fitness score reflects the applicability of the atomic rule unit in a specific scenario. According to the fitness scores, use the dynamic programming algorithm to adaptively reorganize the atomic rule units to generate a scenario-based rule chain. This process ensures that the generated rule chain can effectively adapt to different business scenarios.

[0082] Through the dynamic reorganization of atomic rule units, the present invention can quickly respond to changes in business scenarios and ensure the effectiveness of the rules; through temporal tracking, it can clearly understand the dependence relationship and execution order between the rules, which is convenient for subsequent rule management and optimization; through atomization decomposition and scenario-based reorganization, the rule management process is simplified, making the maintenance and update of the rules more efficient.

[0083] In an alternative embodiment,

[0084] The steps of constructing a spatio-temporal feature tensor for the node, where the spatio-temporal feature tensor includes regular temporal attributes, spatial constraints, and business attributes, constructing a regular temporal matrix based on the spatio-temporal feature tensor, and performing feature enhancement on the regular temporal matrix to construct a multi-dimensional association matrix including regular temporal features and business attributes are as follows:

[0085] Construct a spatio-temporal feature tensor for the regular node, where the spatio-temporal feature tensor includes a temporal feature vector, a spatial feature vector, and a business feature vector. The temporal feature vector includes the effective time, expiration time, maximum execution duration, and minimum execution duration of the rule. The spatial feature vector includes the scope of action, geographical location restriction, and management level of the rule. The business feature vector includes the priority, rule type, and business domain of the rule;

[0086] Perform a temporal dimension projection on the spatio-temporal feature tensor to obtain a temporal feature matrix, calculate the time overlap degree and execution order relationship between rules to obtain the temporal relationship strength, and construct a regular temporal matrix based on the temporal feature matrix and the temporal relationship strength;

[0087] Calculate the similarity of the spatial feature vectors to obtain the spatial association strength, and calculate the correlation of the business feature vectors to obtain the business association strength;

[0088] Perform feature enhancement on the regular temporal matrix, and perform multi-dimensional feature fusion on the regular temporal matrix with the spatial association strength and the business association strength to generate an enhanced association matrix;

[0089] Construct a query matrix, a key matrix, and a value matrix, calculate the attention weights based on the enhanced association matrix, and perform weighted fusion on the attention weights and the enhanced association matrix to generate a multi-dimensional association matrix including the high-order association relationship of the rules.

[0090] Exemplarily, first, construct a spatio-temporal feature tensor for the regular node. This tensor consists of three main feature vectors: a temporal feature vector, a spatial feature vector, and a business feature vector. The temporal feature vector includes the effective time, expiration time, maximum execution duration, and minimum execution duration of the rule, and this information can reflect the time attributes of the rule. The spatial feature vector contains the scope of action, geographical location restriction, and management level of the rule, ensuring that the applicable scope of the rule is clear. The business feature vector includes the priority, rule type, and business domain of the rule, helping to identify the importance of the rule in the business process.

[0091] For example, the effective time of a certain rule is January 1, 2023, the expiration time is December 31, 2023, the maximum execution duration is 48 hours, the minimum execution duration is 1 hour, the scope is the whole country, the geographical location restriction is a specific city, the priority is high, the rule type is compliance check, and the business area is finance.

[0092] Perform a temporal dimension projection on the constructed spatio-temporal feature tensor to obtain a temporal feature matrix. By analyzing the time overlap and execution order relationship between rules, the temporal relationship strength is calculated. The temporal relationship strength reflects the degree of mutual influence between rules and provides a basis for constructing the subsequent rule temporal matrix.

[0093] For example, if there is an overlap between the effective time of rule A and the expiration time of rule B, it can be considered that there is a certain time relationship between these two rules, and then the temporal relationship strength between them is calculated.

[0094] Based on the temporal feature matrix and the temporal relationship strength, construct a rule temporal matrix. This matrix presents the temporal relationship between different rules in matrix form, facilitating subsequent analysis and processing.

[0095] For example, if the temporal relationship strength between rule A and rule B is 0.8, and the temporal relationship strength between rule A and rule C is 0.5, then in the rule temporal matrix, the relationship value between A and B is 0.8, and the relationship value between A and C is 0.5.

[0096] Next, calculate the similarity of the spatial feature vectors to obtain the spatial association strength. At the same time, calculate the correlation of the business feature vectors to obtain the business association strength. These two strength values will be used in the subsequent feature enhancement step.

[0097] For example, if rules A and B have the same scope and similar priorities, their spatial association strength is relatively high; if rules A and C have the same business area, their business association strength is also relatively high.

[0098] Perform feature enhancement on the rule temporal matrix, and perform multi-dimensional feature fusion with the spatial association strength and the business association strength to generate an enhanced association matrix. This step aims to enhance the association between rules and make subsequent analysis more accurate.

[0099] For example, if the value of the temporal matrix of rule A is 0.8, the spatial association strength is 0.7, and the business association strength is 0.9, then in the enhanced association matrix, the final value of rule A will comprehensively consider the influence of these three factors.

[0100] Finally, construct a query matrix, a key matrix, and a value matrix. Calculate the attention weights based on the enhanced association matrix, and perform weighted fusion of the attention weights and the enhanced association matrix to generate a multi-dimensional association matrix containing regular high-order association relationships. This matrix will provide an important basis for subsequent decision-making support.

[0101] For example, if the attention weight of a certain query in the query matrix is 0.6 and the value of the corresponding rule in the enhanced association matrix is 0.8, then the value after weighted fusion is 0.48, reflecting the influence degree of this query on the rule.

[0102] The present invention can comprehensively reflect the relationships between rules by constructing spatio-temporal feature tensors and multi-dimensional association matrices, improving the accuracy of analysis results; the enhanced association matrix provides richer information for decision-making, helping business personnel make more reasonable decisions; through comprehensive analysis of the temporal, spatial, and business characteristics of rules, it can better adapt to changes in the business environment and improve the flexibility of rule management.

[0103] In an optional implementation manner,

[0104] The steps of performing multi-level knowledge modeling on the review results and constructing a vertically layered knowledge system include:

[0105] Extract business entities, rule entities, result entities, and time entities from the review results, construct an entity relationship set including triggering relationships, generation relationships, and temporal relationships, and extract an attribute set including business types, rule identifiers, result types, timestamps, and confidence levels;

[0106] Based on the entity relationship set and the attribute set, construct the conceptual hierarchical relationships of business, rules, results, and time, calculate the semantic similarity between concepts, construct a concept clustering tree according to the semantic similarity, and extract the semantic association rules between concepts;

[0107] Divide the knowledge system into a data layer, a rule layer, an inference layer, and an application layer. The data layer processes temporal data streams, the rule layer generates context-aware rules based on the temporal data streams, the inference layer performs state-dependent reasoning according to the context-aware rules, and the application layer constructs scenario-based applications based on the state-dependent reasoning; use the semantic association rules to construct inter-layer context information, perform bottom-up knowledge refinement to obtain inter-layer knowledge mapping relationships, determine global constraint conditions based on the inter-layer knowledge mapping relationships, and perform top-down knowledge guidance;

[0108] Construct a knowledge evolution equation in combination with external environmental information and internal consistency requirements, calculate the knowledge update gradient according to the knowledge evolution equation, and determine the adaptive update strategy of knowledge in combination with the knowledge update gradient and the knowledge drift amount;

[0109] Calculate the difference degree of knowledge representation between the upper and lower layers and the degree of violation of structural constraints according to the inter-layer knowledge mapping relationship to obtain the consistency deviation, count the knowledge coverage of the target business scenario and the missing situation of the knowledge link to obtain the integrity index, and analyze the decision accuracy rate and scenario adaptability of knowledge application to obtain the effectiveness index; construct a weighted objective function of the consistency deviation, integrity index and effectiveness index, set the knowledge structure integrity and semantic consistency constraints based on the weighted objective function, optimize the indicators that violate the constraints through a combined gradient descent and coordinate alternating optimization method, dynamically adjust the weights of each indicator during the optimization process, and adaptively adjust the optimization step size according to the application feedback;

[0110] Calculate the accuracy rate, coverage rate and timeliness of knowledge to obtain the knowledge quality evaluation result, and dynamically adjust the knowledge structure according to the evaluation result.

[0111] Exemplarily, extract business entities, rule entities, result entities and time entities from the review results. Business entities can be customers, products, etc., rule entities can be compliance standards, operation processes, etc., result entities are the specific manifestations of the review results, and time entities are the time information related to the review. Then, construct an entity relationship set, including trigger relationships (such as the operation of a certain business entity triggers the application of a certain rule entity), generation relationships (such as a certain rule entity generates a certain result entity) and temporal relationships (such as the association between a certain time entity and a certain result entity). At the same time, extract the attribute set, including information such as business type, rule identifier, result type, timestamp and confidence level.

[0112] Based on the aforementioned entity relationship set and attribute set, construct the conceptual hierarchical relationships of business, rule, result and time. By analyzing the relationships between these entities, calculate their semantic similarity, and then construct a conceptual clustering tree. This clustering tree classifies similar concepts into one category, extracts the semantic association rules between concepts for subsequent knowledge reasoning and application.

[0113] Divide the knowledge system into a data layer, a rule layer, an inference layer and an application layer. The data layer is responsible for processing time-series data streams to ensure the real-time and accuracy of data. The rule layer generates context-aware rules based on the time-series data stream to ensure the applicability and effectiveness of the rules. The inference layer performs state-dependent reasoning according to the context-aware rules to deduce new knowledge and conclusions. The application layer constructs scenario-based applications based on state-dependent reasoning to ensure the actual application effect of knowledge.

[0114] Utilize the aforementioned extracted semantic association rules to construct inter-layer context information. Through bottom-up knowledge refinement, obtain the inter-layer knowledge mapping relationship. Based on these mapping relationships, determine the global constraint conditions, and perform top-down knowledge guidance to ensure the overall consistency and effectiveness of the knowledge system.

[0115] Construct a knowledge evolution equation by combining external environmental information and internal consistency requirements. Calculate the gradient of knowledge update through this equation, and combine the knowledge update gradient and knowledge drift amount to determine the adaptive update strategy of knowledge to cope with the changing environment and requirements.

[0116] According to the inter-layer knowledge mapping relationship, calculate the difference degree of knowledge representation between the upper and lower layers and the degree of violation of structural constraints to obtain the consistency deviation. At the same time, count the knowledge coverage of the target business scenario and the missing situation of the knowledge link to obtain the integrity index. Analyze the decision accuracy rate and scenario adaptation degree of knowledge application to obtain the effectiveness index.

[0117] Construct a weighted objective function of the consistency deviation, integrity index, and effectiveness index. Based on this objective function, set the integrity and semantic consistency constraints of the knowledge structure. Optimize the indicators that violate the constraints through a combined gradient descent and coordinate alternating optimization method. During the optimization process, dynamically adjust the weights of each indicator and adaptively adjust the optimization step size according to the application feedback to ensure the continuous optimization of the knowledge structure.

[0118] Calculate the accuracy rate, coverage rate, and timeliness of knowledge to obtain the knowledge quality evaluation result. According to the evaluation result, dynamically adjust the knowledge structure to ensure the high quality and applicability of knowledge.

[0119] The present invention improves the accuracy and consistency of the knowledge system, ensuring that information can be effectively transmitted and applied between different levels; through the adaptive update strategy, enhances the response ability of the knowledge system to external environmental changes, improves the timeliness and applicability of knowledge; through optimizing the knowledge structure, improves the coverage rate and effectiveness of knowledge, ensuring that accurate decision-making support can be provided in practical applications.

[0120] In an alternative embodiment,

[0121] The steps of establishing a knowledge dual network based on the knowledge system, calculating the cross-entropy time series distribution of the original network and the dual network, curve fitting and extrapolating the trend of entropy value change, combining the dynamic prediction of network consistency, detecting and locating potential knowledge conflicts, generating a repair strategy, and updating the repaired knowledge to the knowledge base include:

[0122] Construct a knowledge dual network; calculate the dynamic cross-entropy based on the time decay factor; use the attention mechanism to fuse multi-dimensional prediction features, and evaluate the network consistency through hierarchical calculation from local to global; dynamically adjust the warning threshold based on the historical trigger frequency and activity; locate the conflict source according to the betweenness centrality and generate a repair strategy, specifically including:

[0123] Map the association relationships between knowledge nodes into a knowledge graph structure, calculate the association strength between nodes based on the feature vector representation of nodes and the node association set, and construct a dual network by reversing the association direction between nodes in the original network and recalculating the weights;

[0124] Obtain the probability distributions of nodes in the original network and the dual network to calculate the network cross-entropy, introduce a time decay factor to dynamically weight the network cross-entropy, and obtain the cross-entropy sequence of the original network and the dual network within a preset time window; construct a polynomial kernel function to map the cross-entropy sequence to a high-dimensional feature space, extract the time-varying feature vector, and the time-varying feature vector includes the feature mapping components calculated by the kernel function; perform curve fitting on the time series sequence of the cross-entropy based on the time-varying feature vector, and extrapolate and predict the fitting curve;

[0125] Construct multi-dimensional prediction features based on the historical data of the cross-entropy sequence, network topology features, and node semantic similarity, use the attention mechanism to dynamically fuse the multi-dimensional prediction features to obtain the fused feature, perform hierarchical prediction according to the knowledge domain based on the fused feature, perform network consistency evaluation through hierarchical calculation from local to global, calculate the adaptive threshold based on the historical trigger frequency and knowledge update activity for early warning determination, and trigger conflict detection at the corresponding level according to the early warning result;

[0126] Calculate the betweenness centrality of nodes in the original network and the dual network, locate the conflict source nodes based on the change amount of the betweenness centrality, and generate a repair strategy based on the graph edit distance for the conflict source nodes; update the repair strategy to the knowledge base.

[0127] Exemplarily, the process of establishing a knowledge dual network based on the knowledge system includes multiple steps, aiming to detect and locate potential knowledge conflicts and generate corresponding repair strategies through the calculation of dynamic cross-entropy and the evaluation of network consistency. The following are the detailed steps to implement this technical solution.

[0128] First, construct a knowledge dual network. This step involves mapping the association relationships between knowledge nodes into a knowledge graph structure. By analyzing the feature vector representation of nodes and the association set between nodes, calculate the association strength between nodes. Then, reverse the association direction between nodes in the original network and recalculate the weights to construct a dual network.

[0129] Second, calculate the probability distributions of nodes in the original network and the dual network, and calculate the network cross-entropy based on this. Introduce a time decay factor to dynamically weight the cross-entropy, and obtain the cross-entropy sequence of the original network and the dual network within a preset time window. This sequence provides the basic data for subsequent analysis.

[0130] Next, construct a polynomial kernel function to map the cross-entropy sequence to a high-dimensional feature space. By extracting time-varying feature vectors, obtain the feature vectors containing the feature mapping components calculated by the kernel function. Based on these time-varying feature vectors, perform curve fitting on the time series of cross-entropy and extrapolate the fitted curve for prediction to identify potential trend changes.

[0131] On this basis, construct multi-dimensional prediction features based on the historical data of the cross-entropy sequence, network topology features, and node semantic similarity. Use the attention mechanism to dynamically fuse these multi-dimensional prediction features to obtain the fused features. According to the fused features, perform hierarchical prediction according to knowledge domains, and conduct network consistency evaluation through hierarchical calculations from local to global.

[0132] Subsequently, calculate an adaptive threshold based on the historical trigger frequency and the activity of knowledge update for early warning determination. According to the early warning results, trigger conflict detection at corresponding levels to timely discover potential knowledge conflicts.

[0133] Finally, calculate the betweenness centrality of nodes in the original network and the dual network. Based on the change amount of the betweenness centrality, locate the conflict source nodes, and generate a repair strategy based on the graph edit distance for these conflict source nodes. Update the repair strategy to the knowledge base to ensure the accuracy and consistency of the knowledge base.

[0134] Through the calculation of dynamic cross-entropy and time series analysis, the present invention can timely identify the potential risks of knowledge conflicts, thereby improving the efficiency and accuracy of knowledge management; using the attention mechanism to fuse multi-dimensional prediction features can more comprehensively evaluate network consistency and enhance the stability and reliability of the knowledge network; the generated repair strategy is based on data-driven analysis, which can effectively solve knowledge conflict problems and ensure the continuous update and optimization of the knowledge base.

[0135] In an alternative embodiment,

[0136] The steps of performing hierarchical prediction according to knowledge domains based on the fused features, conducting network consistency evaluation through hierarchical calculations from local to global, and calculating an adaptive threshold based on the historical trigger frequency and the activity of knowledge update for early warning determination include:

[0137] Construct a multi-level fractal calculation framework, and use a bidirectional mapping matrix to describe the intra-layer knowledge association; fuse semantic features and topological features to construct a fractal recursive function, and quantify the hierarchical knowledge complexity through iterative calculation; perform multi-level joint prediction driven by fractal features; evaluate the global consistency based on the geometric mean of the inter-level consistency covariance and fractal features; dynamically adjust the early warning threshold according to the knowledge propagation influence degree, specifically including:

[0138] Construct a multi-level fractal calculation framework, which is divided into multiple levels according to the professional degree and knowledge correlation of the knowledge field based on the fusion features. A bidirectional mapping matrix is constructed at each level to characterize the correlation features between knowledge units within the level; extract multi-dimensional semantic features from the node set of each level, and fuse the multi-dimensional semantic features with the topological features of the knowledge graph to generate a relationship strength matrix of the nodes; construct a fractal recursive function based on the relationship strength matrix, and obtain the fractal features reflecting the complexity of hierarchical knowledge through iterative calculation;

[0139] Perform hierarchical prediction driven by fractal features on the nodes in each level, splice the fusion feature vector of the node, the fractal features of the level to which it belongs, and the prediction information of the adjacent level, and calculate the prediction features of the node through a parameter matrix and an activation function;

[0140] Calculate the local consistency index based on the similarity of the prediction features of the nodes between adjacent levels, and calculate the global consistency index by combining the weights of each level, the consistency covariance between levels, and the geometric mean of the fractal features;

[0141] Construct an adaptive threshold according to the historical trigger frequency and knowledge update activity, introduce the knowledge verification passing rate, knowledge dissemination coverage, and knowledge conversion rate to calculate the knowledge dissemination influence degree, and dynamically adjust the adaptive threshold based on the change rate of the knowledge dissemination influence degree and the global consistency index; trigger an alarm when the global consistency index is lower than the dynamically adjusted adaptive threshold.

[0142] Exemplarily, first, construct a multi-level fractal calculation framework. This framework divides the knowledge field into multiple levels according to the professional degree and knowledge correlation. In each level, a bidirectional mapping matrix is constructed to characterize the correlation features between knowledge units within the level. The bidirectional mapping matrix can effectively reflect the mutual influence and relationship between knowledge units, providing a basis for subsequent feature fusion and prediction.

[0143] Next, extract multi-dimensional semantic features from the node set of each level. These features can be extracted from relevant literature, databases, or knowledge graphs through natural language processing techniques. Fuse the extracted multi-dimensional semantic features with the topological features of the knowledge graph to generate a relationship strength matrix of the nodes. The relationship strength matrix not only reflects the direct relationship between nodes but also can embody the potential paths of knowledge dissemination.

[0144] Then, construct a fractal recursive function based on the relationship strength matrix. Through iterative calculation, the fractal recursive function can quantify the complexity of hierarchical knowledge and reflect the process of knowledge dissemination and evolution between different levels. This process will provide necessary feature support for subsequent hierarchical prediction.

[0145] In each layer, a fractal feature-driven hierarchical prediction is performed. The fused feature vector of the node, the fractal features of the layer it belongs to, and the prediction information of adjacent layers are concatenated as features. Through the calculation of the parameter matrix and the activation function, the predicted features of the node are obtained. This process ensures information sharing and synergy between different layers, thereby improving the accuracy of prediction.

[0146] Next, based on the similarity of the predicted features of nodes between adjacent layers, a local consistency index is calculated. Combining the weights of each layer, the consistency covariance between layers, and the geometric mean of the fractal features, a global consistency index is calculated. The global consistency index can reflect the stability and consistency of the entire knowledge system, providing a basis for the subsequent early warning mechanism.

[0147] Finally, an adaptive threshold is constructed based on the historical trigger frequency and the knowledge update activity. The knowledge verification passing rate, the knowledge dissemination coverage, and the knowledge conversion rate are introduced to calculate the knowledge dissemination influence degree. Based on the change rate of the knowledge dissemination influence degree and the global consistency index, the adaptive threshold is dynamically adjusted. When the global consistency index is lower than the dynamically adjusted adaptive threshold, the early warning mechanism is triggered to respond promptly to potential knowledge dissemination risks.

[0148] The present invention uses a bidirectional mapping matrix to characterize the intra-layer knowledge association, combines semantic features and topological features to construct a fractal recursive function, making the quantification of knowledge complexity more accurate; the fractal feature-driven multi-level joint prediction mechanism improves the prediction accuracy through feature concatenation and parameter matrix calculation; the global consistency is evaluated based on the consistency covariance between layers and the geometric mean of fractal features, making the consistency evaluation more comprehensive; the knowledge dissemination influence degree is calculated by introducing multi-dimensional indicators such as the knowledge verification passing rate, the knowledge dissemination coverage, and the knowledge conversion rate, realizing the precise dynamic adjustment of the early warning threshold.

[0149] In an alternative embodiment,

[0150] The steps of constructing a multi-level fractal calculation framework include:

[0151] Collect the node semantic vectors and multi-dimensional relationship strengths, perform non-linear mapping on the node semantic vectors and multi-dimensional relationship strengths through a bidirectional mapping matrix to obtain the initial node features, construct a fractal recursive function based on the initial node features, and the fractal recursive function includes the topological correlation degree and semantic similarity of the nodes; obtain the hierarchical fractal features through iterative recursive calculation, and the recursive calculation includes the propagation and diffusion of fractal features and the aggregation of features between layers, and the hierarchical fractal features optimize the global fractal dimension through the dynamic programming method;

[0152] Construct a multi-scale sliding time window for the hierarchical fractal features, introduce an adaptive time decay factor for different scale windows, and the time decay factor is dynamically adjusted according to the knowledge update frequency and the association strength; calculate the dynamic fractal dimensions of nodes and levels within each time window, and the dynamic fractal dimensions fuse the historical dimension information through an exponentially weighted method; construct a time-series fractal feature sequence based on the dynamic fractal dimensions, and the time-series fractal feature sequence contains the evolution trajectory and change pattern of the knowledge structure; perform trend decomposition and periodic analysis on the time-series fractal feature sequence, extract the long-term trend and short-term fluctuation features of knowledge evolution, calculate the change rate of adjacent moments of the time-series fractal features, combine the variance of the time-series fractal feature sequence to construct a multi-scale stability evaluation function, and calculate the stability index of the knowledge structure based on the multi-scale stability evaluation function;

[0153] Construct an update threshold function according to the stability index, trigger the knowledge update mechanism when the change rate of the time-series fractal features exceeds the update threshold function, and determine the intensity of the knowledge update strategy based on the change rate of the time-series fractal features and the stability index; feedback the execution result of the knowledge update strategy to the fractal feature calculation link for the dynamic update of the next round of fractal features.

[0154] Exemplarily, first, collecting node semantic vectors and multi-dimensional relationship strengths is the basis for constructing a fractal calculation framework. The node semantic vector is a high-dimensional representation of node features, while the multi-dimensional relationship strength reflects the degree of association between nodes. Through a bi-directional mapping matrix, the node semantic vector and the multi-dimensional relationship strength are non-linearly mapped to obtain the initial features of the nodes. This process can be achieved through machine learning algorithms, such as using neural networks for feature extraction and mapping.

[0155] Next, based on the initial features of the nodes, construct a fractal recursive function. This function not only considers the topological association degree of the nodes but also includes semantic similarity. Through recursive calculation, hierarchical fractal features can be obtained. At this time, the propagation and diffusion of fractal features and the feature aggregation between levels are key links. Propagation and diffusion refer to the transfer of features between nodes, while feature aggregation between levels is to integrate features at different levels. The dynamic programming method is applied here to optimize the global fractal dimension and ensure the efficiency and accuracy of the calculation.

[0156] After obtaining the hierarchical fractal features, construct a multi-scale sliding time window. An adaptive time decay factor is introduced within each time window, and this factor is dynamically adjusted according to the knowledge update frequency and the association strength. In this way, the dynamic changes of nodes and levels can be better captured. Within each time window, calculate the dynamic fractal dimensions of nodes and levels. The calculation of the dynamic fractal dimension uses an exponentially weighted method to fuse historical dimension information to ensure sensitivity to changes.

[0157] Based on the dynamic fractal dimension, a time-series fractal feature sequence is constructed. This sequence not only contains the evolution trajectory of the knowledge structure but also reflects the change pattern. By performing trend decomposition and periodic analysis on the time-series fractal feature sequence, the long-term trend and short-term fluctuation characteristics of knowledge evolution can be extracted. Calculate the change rate at adjacent moments and combine it with the variance of the time-series fractal feature sequence to construct a multi-scale stability evaluation function. This function is used to evaluate the stability index of the knowledge structure.

[0158] According to the stability index, an update threshold function is constructed. When the change rate of the time-series fractal feature exceeds the update threshold, the knowledge update mechanism is triggered. At this time, based on the change rate and the stability index, the intensity of the knowledge update strategy is determined. Finally, the execution result of the knowledge update strategy is fed back to the fractal feature calculation link for the next round of dynamic update.

[0159] The present invention adopts the method of combining a bidirectional mapping matrix with a fractal recursive function, and realizes the multi-level propagation and aggregation of fractal features through iterative recursive calculation, making the feature extraction more accurate; through a multi-scale sliding time window and an adaptive time decay factor, it realizes the fine-grained tracking of knowledge evolution; uses the dynamic fractal dimension and the time-series fractal feature sequence to accurately capture the evolution trajectory of the knowledge structure; based on the dynamic adjustment mechanism of the stability index and the update threshold function, it realizes the adaptive optimization of the knowledge update strategy and improves the real-time response ability of the system.

[0160] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. The automatic filling review and knowledge base integration optimization method based on dynamic rule engine is characterized by: include: Standardize the format of electronic form data to obtain pre-processed data, perform structured analysis on the pre-processed data according to preset reporting rules, and generate a data set to be reviewed; Build a dynamic rule engine based on the rule time sequence graph, track the rules in time sequence, decompose the rules into atomic rule units, adaptively reorganize the atomic rule units according to the business scenario, generate a scenario-based rule chain, use the scenario-based rule chain to match the rules of the data set to be reviewed, and obtain the review result; The construction of the rule time sequence graph includes: taking business rules as nodes of the rule time sequence graph, and taking the time sequence dependency between rules as directed edges of the rule time sequence graph; Perform multi-level knowledge modeling on the review results to construct a vertically hierarchical knowledge system, including: extracting business entities, rule entities, result entities and time entities from the review results, constructing an entity relationship set including trigger relationships, generation relationships and time series relationships, and extracting an attribute set including business type, rule identifier, result type, timestamp and confidence; constructing a conceptual hierarchical relationship between business, rules, results and time based on the entity relationship set and the attribute set, calculating the semantic similarity between concepts, constructing a concept clustering tree based on the semantic similarity, and extracting semantic association rules between concepts; dividing the knowledge system into a data layer, a rule layer, an inference layer and an application layer, the data layer processes time series data streams, the rule layer generates context-aware rules based on the time series data streams, the inference layer performs state-dependent reasoning based on the context-aware rules, and the application layer constructs scenario-based applications based on the state-dependent reasoning; using the semantic association rules to construct inter-layer context information, performing bottom-up knowledge extraction to obtain inter-layer knowledge mapping relationships, determining global constraints based on the inter-layer knowledge mapping relationships, and performing top-down knowledge extraction to obtain global constraints. Based on the knowledge system, a knowledge dual network is established, including: mapping the association relationship between knowledge nodes into a knowledge graph structure as the original network, calculating the association strength between nodes based on the feature vector representation of the nodes and the node association set, and building a dual network by reversing the association direction between nodes in the original network and recalculating the weights; calculating the cross entropy time series distribution of the original network and the dual network, curve fitting and extrapolating the entropy value change trend, and combining the dynamic prediction of network consistency, including: building multidimensional prediction features based on the historical data of the cross entropy sequence of the original network and the dual network in a preset time window, network topology features and node semantic similarity, using the attention mechanism to dynamically fuse the multidimensional prediction features to obtain fusion features, performing hierarchical prediction according to the knowledge field according to the fusion features, evaluating the network consistency through local to global hierarchical calculations, calculating the adaptive threshold based on the historical trigger frequency and knowledge update activity for early warning judgment, triggering the corresponding level of conflict detection according to the early warning results, locating potential knowledge conflicts, generating repair strategies, and updating the repaired knowledge to the knowledge base.

2. The method according to claim 1, characterized in that The steps of building a dynamic rule engine based on the rule time sequence graph, tracking the rules in time sequence, decomposing the rules into atomic rule units, and adaptively reorganizing the atomic rule units according to the business scenarios to generate scenario-based rule chains include: Construct a rule time series graph, represent the rule node features based on the spatiotemporal feature tensor; atomically decompose the rules based on semantic integrity; use the scenario feature vector to drive the rule reorganization and generate a scenario-based rule chain, including: Construct a rule timing graph, use business rules as nodes of the rule timing graph, use the timing dependencies between rules as directed edges of the rule timing graph, construct a spatiotemporal feature tensor for the nodes, the spatiotemporal feature tensor includes the timing attributes, spatial constraints and business attributes of the rules, construct a rule timing matrix based on the spatiotemporal feature tensor, perform feature enhancement on the rule timing matrix, construct a multidimensional association matrix including rule timing features and business attributes, and perform timing tracking on the rules according to the time overlap and execution order relationship between the rules in the multidimensional association matrix; Based on the rule time series graph, the rules are atomically decomposed into atomic rule units that maintain semantic integrity, the semantic relevance between the atomic rule units is calculated, and when the semantic relevance is less than a preset boundary threshold, the separation boundary of the atomic rule unit is determined to generate an atomic rule unit set; A scene feature vector is constructed, and the fitness score between each atomic rule unit in the atomic rule unit set and the scene feature vector is calculated. According to the fitness score, a dynamic programming algorithm is used to adaptively reorganize the atomic rule units to generate a scene-based rule chain.

3. The method according to claim 2, characterized in that A spatiotemporal feature tensor is constructed for the node, wherein the spatiotemporal feature tensor includes a regular time series attribute, a spatial constraint, and a business attribute; a regular time series matrix is ​​constructed based on the spatiotemporal feature tensor; and feature enhancement is performed on the regular time series matrix. The steps of constructing a multidimensional association matrix including regular time series features and business attributes include: Constructing a spatiotemporal feature tensor of a rule node, the spatiotemporal feature tensor comprising a time series feature vector, a space feature vector and a business feature vector, wherein the time series feature vector comprises the effective time, expiration time, maximum execution time and minimum execution time of the rule, the space feature vector comprises the scope, geographical location restriction and management level of the rule, and the business feature vector comprises the priority, rule type and business field of the rule; Performing time series dimension projection on the spatiotemporal feature tensor to obtain a time series feature matrix, calculating the time overlap and execution order relationship between the rules to obtain the time series relationship strength, and constructing a rule time series matrix based on the time series feature matrix and the time series relationship strength; Calculating the similarity of the spatial feature vectors to obtain spatial association strength, and calculating the correlation of the service feature vectors to obtain service association strength; Performing feature enhancement on the regular time series matrix, performing multi-dimensional feature fusion on the regular time series matrix, the spatial association intensity and the business association intensity, and generating an enhanced association matrix; Construct a query matrix, a key-value matrix and a numerical matrix, calculate the attention weight based on the enhanced association matrix, perform weighted fusion on the attention weight and the enhanced association matrix, and generate a multidimensional association matrix containing regular high-order association relationships.

4. The method according to claim 1, characterized in that: The steps of performing multi-level knowledge modeling on the review results and constructing a vertically hierarchical knowledge system include: Extract business entities, rule entities, result entities and time entities from the review results, build an entity relationship set including trigger relationships, generation relationships and timing relationships, and extract an attribute set including business type, rule identifier, result type, timestamp and confidence level; Based on the entity relationship set and the attribute set, a conceptual hierarchical relationship among business, rules, results and time is constructed, the semantic similarity between concepts is calculated, a concept clustering tree is constructed according to the semantic similarity, and semantic association rules between concepts are extracted; The knowledge system is divided into a data layer, a rule layer, a reasoning layer, and an application layer. The data layer processes time series data streams. The rule layer generates context-aware rules based on the time series data streams. The reasoning layer performs state-dependent reasoning according to the context-aware rules. The application layer builds scenario-based applications based on the state-dependent reasoning. The semantic association rules are used to construct inter-layer context information, and bottom-up knowledge extraction is performed to obtain inter-layer knowledge mapping relationships. Global constraints are determined based on the inter-layer knowledge mapping relationships, and top-down knowledge guidance is performed. A knowledge evolution equation is constructed by combining external environment information and internal consistency requirements, a knowledge update gradient is calculated according to the knowledge evolution equation, and an adaptive knowledge update strategy is determined by combining the knowledge update gradient and the knowledge drift amount; According to the inter-layer knowledge mapping relationship, the difference between the upper and lower layer knowledge representations and the degree of violation of the structural constraints are calculated to obtain the consistency deviation, the knowledge coverage of the target business scenario and the missing of the knowledge link are counted to obtain the integrity index, and the decision accuracy of the knowledge application and the degree of adaptability to the scenario are analyzed to obtain the effectiveness index; a weighted objective function of the consistency deviation, integrity index and effectiveness index is constructed, and the knowledge structure integrity and semantic consistency constraints are set based on the weighted objective function. The indicators that violate the constraints are optimized by combining the gradient descent and the coordinate alternating optimization method, and the weights of each indicator are dynamically adjusted during the optimization process, and the optimization step size is adaptively adjusted according to the application feedback; Calculate the accuracy, coverage and timeliness of knowledge to obtain knowledge quality assessment results, and dynamically adjust the knowledge structure based on the assessment results.

5. The method according to claim 1, characterized in that: The steps of establishing a knowledge dual network based on the knowledge system, calculating the time series distribution of the cross entropy between the original network and the dual network, performing curve fitting and extrapolation on the entropy value change trend, combining dynamic prediction of network consistency, detecting and locating potential knowledge conflicts, generating a repair strategy, and updating the repaired knowledge to the knowledge base include: Construct a knowledge dual network; calculate dynamic cross entropy based on the time decay factor; use the attention mechanism to fuse multi-dimensional prediction features, and evaluate network consistency through local to global hierarchical calculations; dynamically adjust the warning threshold based on historical trigger frequency and activity; locate the conflict source and generate a repair strategy based on the betweenness centrality, including: The association relationship between knowledge nodes is mapped into a knowledge graph structure. The association strength between nodes is calculated based on the feature vector representation of the nodes and the node association set. The dual network is constructed by reversing the association direction between nodes in the original network and recalculating the weights. Obtain the probability distribution of nodes in the original network and the dual network to calculate the network cross entropy, introduce a time decay factor to dynamically weight the network cross entropy, and obtain the cross entropy sequence of the original network and the dual network within a preset time window; construct a polynomial kernel function to map the cross entropy sequence to a high-dimensional feature space, extract a time-varying feature vector, and the time-varying feature vector includes a feature mapping component calculated by the kernel function; perform curve fitting on the time series of the cross entropy based on the time-varying feature vector, and perform extrapolation prediction on the fitting curve; Based on the historical data of the cross entropy sequence, network topology features and node semantic similarity, a multidimensional prediction feature is constructed, and the multidimensional prediction feature is dynamically fused by using an attention mechanism to obtain a fusion feature. According to the fusion feature, hierarchical prediction is performed according to the knowledge field, and network consistency is evaluated through local to global hierarchical calculation. Based on the historical trigger frequency and knowledge update activity, an adaptive threshold is calculated for early warning judgment, and conflict detection of the corresponding level is triggered according to the early warning result; Calculate the betweenness centrality of nodes in the original network and the dual network, locate the conflict source node based on the change in the betweenness centrality, generate a repair strategy based on the graph edit distance for the conflict source node; and update the repair strategy to the knowledge base.

6. The method according to claim 5, characterized in that The steps of performing hierarchical prediction according to the knowledge domain based on the fusion features, performing network consistency evaluation through local to global hierarchical calculation, and calculating an adaptive threshold based on historical trigger frequency and knowledge update activity for early warning judgment include: Construct a multi-level fractal computing framework, use a bidirectional mapping matrix to characterize the knowledge association within the layer; integrate semantic features and topological features to construct a fractal recursive function, and quantify the complexity of hierarchical knowledge through iterative calculation; perform multi-level joint prediction driven by fractal features; evaluate global consistency based on the geometric mean of inter-level consistency covariance and fractal features; dynamically adjust the warning threshold according to the influence of knowledge dissemination, including: Construct a multi-level fractal computing framework, divide the knowledge into multiple levels according to the fusion features and the degree of expertise in the knowledge field and the degree of knowledge relevance, and construct a bidirectional mapping matrix at each level to characterize the association features between knowledge units in the layer; extract multi-dimensional semantic features from the node set of each level, and fuse the multi-dimensional semantic features with the topological features of the knowledge graph to generate a node relationship strength matrix; construct a fractal recursive function based on the relationship strength matrix, and obtain fractal features reflecting the complexity of hierarchical knowledge through iterative calculation; Perform fractal feature-driven hierarchical prediction on the nodes in each layer, perform feature splicing on the fused feature vector of the node, the fractal feature of the layer to which it belongs, and the prediction information of the adjacent layers, and calculate the prediction feature of the node through the parameter matrix and activation function; The local consistency index is calculated based on the similarity of the prediction features of nodes between adjacent levels, and the global consistency index is calculated by combining the weights of each level, the consistency covariance between levels, and the geometric mean of the fractal features; An adaptive threshold is constructed based on the historical trigger frequency and knowledge update activity, and the knowledge verification pass rate, knowledge dissemination coverage and knowledge conversion rate are introduced to calculate the knowledge dissemination influence. The adaptive threshold is dynamically adjusted based on the change rate of the knowledge dissemination influence and the global consistency index; when the global consistency index is lower than the dynamically adjusted adaptive threshold, an early warning is triggered.

7. The method according to claim 6, characterized in that The steps to build a multi-level fractal computing framework include: Collecting node semantic vectors and multidimensional relationship strengths, performing nonlinear mapping of the node semantic vectors and the multidimensional relationship strengths through a bidirectional mapping matrix to obtain node initial features, constructing a fractal recursive function based on the node initial features, the fractal recursive function including the topological association and semantic similarity of the nodes; obtaining hierarchical fractal features through iterative recursive calculations, the recursive calculations including the propagation and diffusion of fractal features and the aggregation of features between hierarchies, and optimizing the global fractal dimension through a dynamic programming method. A multi-scale sliding time window is constructed for the hierarchical fractal feature, and an adaptive time decay factor is introduced for windows of different scales, and the time decay factor is dynamically adjusted according to the knowledge update frequency and the correlation strength; the dynamic fractal dimension of the node and the level is calculated in each time window, and the dynamic fractal dimension is integrated with the historical dimension information through an exponential weighting method; a time series fractal feature sequence is constructed based on the dynamic fractal dimension, and the time series fractal feature sequence contains the evolution trajectory and change pattern of the knowledge structure; trend decomposition and periodicity analysis are performed on the time series fractal feature sequence to extract the long-term trend and short-term fluctuation characteristics of knowledge evolution, and the adjacent moment change rate of the time series fractal feature is calculated, and a multi-scale stability evaluation function is constructed in combination with the variance of the time series fractal feature sequence, and the stability index of the knowledge structure is calculated based on the multi-scale stability evaluation function; An update threshold function is constructed according to the stability index. When the change rate of the time series fractal feature exceeds the update threshold function, the knowledge update mechanism is triggered. The strength of the knowledge update strategy is determined based on the change rate of the time series fractal feature and the stability index. The execution result of the knowledge update strategy is fed back to the fractal feature calculation link for the next round of dynamic update of fractal features.

Images