An archive time sequence evolution-oriented knowledge graph construction and visualization method

Abstract

The application discloses a file time sequence evolution-oriented knowledge graph construction and visualization method, relates to the technical field of file intelligent management and knowledge graph, and comprises the following steps: analyzing file original operation logs, marking potential missing file intervals through time sequence sorting and time stamp difference calculation, combining window period logs to complete compliance dormancy, log loss, missing file cause classification in the process of transfer, constructing a file transfer knowledge graph with hierarchical labels and reverse transfer marker bits, extracting business compliance constraints through semantic analysis, generating a structured rule set containing necessary node coverage and parallel transfer mutual exclusion rules, designing a two-dimensional correlation strength quantification model to depict node causal correlation, combining a time, path and resource three-dimensional evaluation model to calculate remediation feasibility weight, generating a traceability chain containing risk levels, and rendering a decision view capable of realizing risk highlighting, path prompting, time sequence tracing and interactive query.

Description

Technical Field

[0001] This application relates to the fields of intelligent archival management and knowledge graph technology, and in particular to a method for constructing and visualizing a knowledge graph oriented towards the temporal evolution of archives. Background Technology

[0002] Records management is a core foundational task for government offices, business record keeping in enterprises and institutions, and the traceability of rights and responsibilities. The entire lifecycle of records, from entry, review, circulation, approval to final archiving, exhibits typical temporal evolution characteristics. Its circulation path, status changes, time constraints, and hierarchical approval relationships directly determine the compliance, authenticity, and traceability of record-related business. With the full implementation of digital archives construction, the amount of record circulation data is growing exponentially, temporally, and in increasing complexity. Traditional static record management models can no longer meet the business needs of dynamic temporal circulation supervision, anomaly tracing, and compliance verification.

[0003] Existing technical archive log analysis can only achieve simple time-series record storage and basic query, and cannot accurately identify time-series gaps in the archive circulation process. It can only make general marks on circulation gaps, and cannot distinguish the three different causes of gaps: compliance dormancy, missing system logs, and risk circulation interruption. This results in low accuracy of abnormal circulation identification, high risk misjudgment rate, and difficulty in supporting refined risk management.

[0004] Existing technologies for constructing traditional archive circulation maps mostly focus on static entity association modeling, without combining the hierarchical structure of archive approval and the time-series circulation logic. They cannot identify abnormal circulation behaviors such as cross-level reverse circulation and illegal jumps, lack hierarchical labels and reverse risk marking mechanisms, and have no ability to quantitatively identify non-compliant topological relationships in archive circulation. The risk representation capability of the map is seriously insufficient. Summary of the Invention

[0005] The purpose of this application is to provide a method for constructing and visualizing knowledge graphs based on the temporal evolution of archives to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, this application provides the following technical solution: a method for constructing and visualizing a knowledge graph oriented towards the temporal evolution of archives, comprising:

[0007] Obtain raw operation log data and business specification documents, perform named entity recognition and timestamp extraction, and generate an initial state sequence containing gap interval markers and cause classification based on preset timestamp interval thresholds.

[0008] Based on the initial state sequence containing gap interval markers and cause classification, the connection relationship and hierarchical labels of state nodes are extracted to construct an archive flow map;

[0009] Dependency parsing is performed on business specification documents to extract necessary nodes, dwell time boundaries, and parallel flow mutual exclusion conditions, forming a structured set of rules;

[0010] The archive flow map is traversed and compared with the structured rule set to generate a second judgment report on the feasibility weight of remedial operations;

[0011] Based on the second judgment report, a responsibility attribution time window chain is constructed by sorting the associated strength of the precursor nodes. For nodes whose remedial operation feasibility weight is lower than the preset feasibility weight threshold, the risk propagation path depth level is marked, and a traceability chain is generated.

[0012] The traceability chain is visualized and rendered, color-coded according to the depth level of the risk propagation path, a dashed connecting box is generated for the compensation mark of missing links, a fork icon is added to the convergence point of the path branch, and an interactive view of the file flow decision is generated.

[0013] In a preferred embodiment of this scheme, generating the initial state sequence containing gap interval markers specifically includes:

[0014] Named entity recognition and timestamp extraction are performed on the original operation log data. Each log record is parsed into a tuple consisting of a status node name and an operation timestamp. All tuples are then sorted in ascending order by timestamp to generate a log entry sequence.

[0015] The difference between the timestamps of every two adjacent pairs in the log entry sequence is calculated sequentially to obtain a timestamp interval sequence;

[0016] Each timestamp interval in the timestamp interval sequence is compared one by one with a preset timestamp interval threshold;

[0017] For each timestamp interval that is greater than the preset timestamp interval threshold, the segment between the two corresponding tuples in the log entry sequence is marked as a potential gap interval.

[0018] From the log entry sequence, extract all the state node names of the binary pairs in the order of the log entry sequence to form a state node name sequence;

[0019] From the log entry sequence, extract the timestamps of all tuples in the log entry sequence order to form a timestamp sequence;

[0020] The state node name sequence, the timestamp sequence, and all marked potential gap intervals are integrated and associated to generate the initial state sequence containing gap interval markings.

[0021] In a preferred embodiment of this scheme, generating the initial state sequence containing the gap interval markers specifically includes:

[0022] Obtain the initial state sequence and locate all marked potential gap intervals from it;

[0023] For each identified potential gap, obtain its start and end timestamps.

[0024] The state node in the last tuple before the start timestamp is determined as the pre-gap node, and the state node in the first tuple after the end timestamp is determined as the post-gap node.

[0025] Retrieve the original operation log data that falls between the start timestamp and the end timestamp from the original operation log data;

[0026] If a user suspension instruction record or authorized hibernation identifier record is retrieved, the cause of the currently processed potential gap interval is classified as a compliant hibernation period;

[0027] If only system-level operation records are retrieved, but no user suspend commands or authorized hibernation identifier records are retrieved, the cause is classified as missing system logs.

[0028] If no valid operation log record is found, the cause is classified as a risk-related interruption.

[0029] The original operation log data is appended to the corresponding potential gap interval markers in the initial state sequence to obtain an initial state sequence containing causal classification labels.

[0030] In a preferred embodiment of this solution, the construction of the archive transfer map specifically includes:

[0031] Obtain the sequence of state node names in the initial state sequence, and instantiate each unique state node name in the sequence of state node names into a node entity;

[0032] Based on the order of the state nodes in the state node name sequence, a directed flow edge is created between every two adjacent node entities, pointing from the previous node to the next node, forming an initial file flow graph.

[0033] For each node entity in the initial archive flow map, its corresponding level is queried from the preset organizational structure, and the queried level is assigned as a level label to the node entity.

[0034] Traverse all directed flow edges in the initial file flow graph, identify reverse flow edges whose terminal node level label is lower than the starting node level label, calculate the absolute value of the level difference between the two nodes of each reverse flow edge, and if the absolute value of the level difference is greater than the preset maximum reverse jump threshold, add a reverse flow marker to the terminal node entity of the reverse flow edge to obtain the file flow graph.

[0035] In a preferred embodiment of this scheme, the formation of a structured set of rules specifically includes:

[0036] Obtain the compliance clause paragraphs from the business specification document, perform dependency parsing on the compliance clause paragraphs, and extract verb-object phrases describing necessary operational steps;

[0037] Identify and extract the object component representing the file processing stage from the verb-object phrase, and use it as the name of the necessary node;

[0038] The time constraint parsing of the compliance clause paragraph is performed to extract the upper limit and lower limit of the node dwell time boundary that limits the name of the required node, which are used as the node dwell time boundary parameters.

[0039] Establish a mapping relationship between the names of the required nodes and their corresponding node dwell time boundary parameters, which serves as the required node coverage requirement;

[0040] If keywords characterizing the parallel approval logic are identified in the compliance clause paragraph, the names of multiple approval nodes participating in the parallel process are extracted.

[0041] Based on the extracted names of multiple approval nodes, parallel flow mutual exclusion constraints are generated, which stipulate that these nodes cannot be direct predecessors or successors of each other in the flow path.

[0042] The structured rule set is generated by combining all the required coverage of necessary nodes with all the parallel flow mutual exclusion constraints.

[0043] In a preferred embodiment of this scheme, the second determination report for generating the feasibility weight of the remedial operation specifically includes:

[0044] The integrity of the state node sequence from the starting node to the ending node in the archive flow map is traversed and compared. If the node sequence in the archive flow map has a missing mandatory node recorded in the rule set or the actual dwell time of a certain node exceeds the node dwell time boundary in the rule set, then the node is recorded as a deviation node, and the node dwell time boundary rule of the target mandatory node violated by the deviation node is obtained.

[0045] The predecessor node of the deviation node in the archive flow map is traced back, the correlation strength between the deviation node and the predecessor node is quantified, the missing link of the deviation node is determined, and a missing link compensation identifier is generated for it.

[0046] Identify all intermediate necessary nodes on the normal flow path from the deviation node to the target necessary node in the archive flow map, and retrieve the upper limit value of the node dwell time boundary of all intermediate necessary nodes from the structured rule set;

[0047] The latest compliance trigger time of the target node is obtained by summing the upper limits of the node dwell time boundaries of all the intermediate necessary nodes at the current time, and the time difference between the latest compliance trigger time and the current time is calculated as the remaining available time.

[0048] Obtain the historical average time taken to perform remedial operations to restore the flow to the target node, calculate the ratio of the remaining available time to the historical average time, and compare them to obtain the time constraint score;

[0049] In the archive flow map, a path search is used to determine whether there is an alternative path that starts from the deviation node, bypasses all missing necessary nodes, and arrives at all subsequent unfinished necessary nodes in sequence. If such a path exists, the path constraint score is set to 1; otherwise, it is set to 0.

[0050] Query the availability status of the authorized resources necessary to perform the remedial operation at the current moment. If available, the resource constraint score is set to 1; otherwise, it is set to 0.

[0051] The time constraint score, the path constraint score, and the resource constraint score are weighted and summed to obtain a second determination report on the feasibility weight of the remedial operation.

[0052] In a preferred embodiment of this scheme, the step of summing up the upper limits of the node dwell time boundaries of all the intermediate necessary nodes at the current time to obtain the latest compliance trigger time of the target necessary node specifically includes:

[0053] The current moment is used as the calculation reference point. In accordance with the flow order from the deviation node to the target necessary node, each intermediate necessary node on the normal flow path is obtained in sequence.

[0054] From the structured set of rules, query and obtain the upper limit value of the dwell time boundary for each of the intermediate necessary nodes;

[0055] Using the current time as the initial cumulative value, the upper limit value of the dwell time boundary of each intermediate necessary node is added one by one in the flow order to complete the cumulative summation. The final result of the cumulative summation is determined as the latest compliance trigger time of the target necessary node.

[0056] The difference between the latest compliance trigger time and the current system time is the remaining available time.

[0057] In a preferred embodiment of this solution, generating the traceability chain specifically includes:

[0058] According to the second determination report, all state node records from the time the deviation node was triggered to the current time are extracted. All state nodes are sorted in descending order of their association strength with the predecessor node of the deviation node to construct a responsibility attribution time window chain. If the feasibility weight of the remedial operation of any state node is lower than the preset feasibility weight threshold, the node is labeled with the risk propagation path depth level to generate a traceability chain containing the responsibility attribution time window and the risk propagation path depth level.

[0059] In a preferred embodiment of this solution, generating the file transfer decision interaction view specifically includes:

[0060] Obtain the archive circulation map and the traceability chain;

[0061] Based on the node entities and directed flow edges in the aforementioned archive flow graph, a low-level static main view is rendered and generated;

[0062] Read all nodes marked with the depth level of the risk propagation path from the traceability chain, and highlight these nodes on the bottom static main view;

[0063] Using a preset color mapping function, the risk propagation path depth level value of each highlighted node is mapped to the corresponding hue and saturation, and the node is then filled with color.

[0064] Read all nodes with missing link compensation identifiers from the traceability chain, generate a dashed auxiliary connection line for each such node that starts from itself and points to the next necessary node it should flow to, and render a dashed connection box around the node.

[0065] For all path branch convergence nodes with multiple input edges or multiple output edges in the underlying static main view, a fork icon that can respond to click commands is superimposed and rendered at its geometric center.

[0066] The chain of responsibility attribution time windows is rendered as a semi-transparent sliding window that can be dragged along the timeline, and superimposed on the timeline area of ​​the underlying static main view.

[0067] When a click command is received on any highlighted node, an information overlay is generated at the top of the view. Within this information overlay, the abnormal flow causal anchor point pattern description and remedial operation feasibility weight value corresponding to the node in the second judgment report are displayed in a structured manner, thus obtaining the file flow decision interaction view.

[0068] Compared with the prior art, the beneficial effects of this application are:

[0069] This application addresses the technical shortcomings of existing archival timeline management, such as ambiguous identification of gaps in record flow, difficulty in quantifying compliance rules, low accuracy in risk tracing, and weak visualization capabilities. It designs a knowledge graph construction and visualization solution oriented towards the timeline evolution of records. This application accurately marks potential gaps in record flow by calculating timeline stamp differences, and automatically distinguishes and traces the causes of gaps—compliance dormancy, missing system logs, and risk-related flow interruptions—by combining window period log data. Simultaneously, it constructs an archival flow knowledge graph with hierarchical labels and reverse jump risk markers, and automatically extracts business specifications through syntactic analysis and semantic parsing to form a machine-recognizable set of structured compliance rules. This achieves standardized automatic verification of essential node coverage, time constraints, and parallel flow mutual exclusion relationships in archival flow, effectively solving the problems of inconsistent standards, missed judgments, misjudgments, and low efficiency in traditional manual review, and significantly improving the precision and automation of archival flow compliance identification.

[0070] This application constructs a node association strength quantification model weighted by two indicators: temporal tightness and business coupling. This model accurately depicts the causal relationship between precursor nodes and flow deviations / anomalies. Combined with three-dimensional constraints of time, path, and resources, a feasibility weighting evaluation system is built to quantify the difficulty of anomaly remediation and the level of risk propagation. Based on the association strength, a traceable responsibility attribution time window chain is generated. Finally, a visualized decision-making view is constructed, featuring risk classification highlighting, remediation path display, temporal drag-and-drop tracing, and interactive detail querying. This enables fully automated processing from anomaly identification, compliance verification, risk quantification, responsibility tracing to visualized decision-making, effectively improving the ability to manage archival temporal risks and the efficiency of intelligent decision-making, adapting to various types of archival full lifecycle compliance management scenarios. Attached Figure Description

[0071] The present application will be further illustrated with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present application. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0072] Figure 1This is a schematic diagram showing the connection of steps in an embodiment of this application. Detailed Implementation

[0073] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0074] Please see Figure 1 This application provides a method for constructing and visualizing a knowledge graph oriented towards the temporal evolution of archives, including:

[0075] Obtain raw operation log data and business specification documents, perform named entity recognition and timestamp extraction, and generate an initial state sequence containing gap interval markers and cause classification based on preset timestamp interval thresholds.

[0076] Furthermore, it should be noted that:

[0077] The raw operation log data comprises all system logs and user operation logs generated throughout the entire lifecycle of the archive. Each log entry includes native fields such as the operation subject, operation behavior, archive status, operation time, and operation permissions. Named entity recognition extracts entities from the log text fields, accurately identifying the status node entities corresponding to the archive's flow, including all business status names such as archive receipt, initial review, secondary review, approval, archiving, rejection, and withdrawal, while filtering out invalid entity information such as operation subject and operation remarks. Timestamp extraction precisely captures standardized time fields from the logs and converts them uniformly to millisecond-level Unix timestamp format, eliminating the problem of inconsistent log time formats (year, month, day; hour, minute, second; time zone differences). After parsing, each log entry uniquely corresponds to a (status node name, operation timestamp) tuple. All tuples are sorted in ascending order based on the value of the operation timestamp, ultimately forming a chronologically ordered sequence of log entries, ensuring the temporal continuity of the archive's flow status.

[0078] Suppose a set of ordered binary tuples representing log entries. , For the first Each status node name For the first A standardized millisecond-level timestamp for each tuple The range of values ​​is The formula for calculating the interval between adjacent timestamps is: ,in For the first The timestamp interval between adjacent pairs of bytes. The range of values ​​is , This represents the total number of log entries. The calculation is performed by iterating through all adjacent pairs, and then... Arrange them in traversal order to generate a one-dimensional ordered array, which is the timestamp interval sequence with a sequence length of n-1.

[0079] The preset timestamp interval threshold is a fixed millisecond-level time parameter pre-configured according to archival business specifications. It is used to determine whether there are abnormal pauses in the archive circulation process. The threshold can be configured differently based on the archive type (ordinary archives, classified archives, important documents). It iterates through each timestamp interval in the sequence. Perform a one-to-one numerical comparison and output the comparison results: > Threshold, The results of the two categories, ≤ and ≤ threshold, provide a basis for subsequent gap marking.

[0080] If the current comparison >If a preset timestamp interval threshold is used, then the first... There is an anomaly in the flow of two adjacent entries. The preceding tuple for this interval is: The post-binary pair is The time segment between the two tuples A potential gap interval is defined, and a unique gap identifier is added to this interval to complete the marking of abnormal intervals. The above marking operation is performed on all intervals that meet the conditions, and finally the location, time range, and corresponding node information of all potential gap intervals are obtained.

[0081] Extract all pairs of data in chronological order according to the log entry sequence, without omission or duplication. (State node name), forming an ordered text sequence This sequence fully restores the order of changes in the status of the archives, preserves the original flow sequence logic, does not perform node deduplication, and truly records the repeated jumps, backtracking and other flow behaviors of the archives.

[0082] The extraction order of the synchronization state node name sequence is as follows: extract the normalized timestamps from all pairs in sequence. To form an ordered numerical sequence Each state node name corresponds one-to-one with the sequence of state node names, achieving precise binding between state and time.

[0083] Using time series as the core correlation benchmark, a one-to-one sequence of state node names is bound to a timestamp sequence to form basic time series state data. Then, all marked potential gaps (including time range, corresponding node position, and gap identifier) ​​are embedded into the corresponding time series positions of the basic time series state data, completing the structured correlation and integration of the three. The resulting initial state sequence is a structured time series dataset containing complete flow states, accurate time series information, and abnormal gap markers, providing raw data support for subsequent gap cause classification.

[0084] Furthermore, it should be noted that the causal classification of potential gaps in the initial state sequence is as follows:

[0085] Read the generated complete initial state sequence, and filter and locate all potential gap intervals in batches by retrieving the unique identifier of the gap in the sequence. Extract the core correlation information of each gap interval, including the gap time sequence position, the preceding and following related nodes, and the original time interval, to form a gap interval to be processed list, ensuring that all abnormal intervals are traversed without omission or repetition.

[0086] For a single potential gap in the pending list, the preceding tuple timestamp of the gap is... Defined as the start timestamp of the gap Post-digit timestamp Defined as the end timestamp of the gap That is, the effective time range of a single gap interval is It accurately pinpoints the time boundary of the gap.

[0087] In the time-series log entry sequence, the retrieval timestamp is less than... The largest time-series tuple corresponds to the state node before the gap. ; The search timestamp is greater than The smallest order pair is the state node after the gap. Accurately pinpoint the archive transfer status nodes before and after the gap occurs, and clarify the corresponding transfer status jump relationships.

[0088] Based on the gap time interval To retrieve and filter data, the original operation log dataset is traversed, and all log operation timestamps that meet the criteria are selected. The system retrieves all valid log data within the gap window period from the log records, which serves as the basis for determining the cause classification. Log retrieval supports millisecond-level accurate matching, eliminating data interference outside the time interval.

[0089] User suspension command logs are user-initiated pause operations for file transfer. Authorized hibernation flag logs are compliance hibernation flags automatically generated by the system based on business approval rules. Both types of logs represent transfer interruptions permitted by business specifications. If either type of log record is found within the gap window, the gap is directly determined to be a compliant scenario and tagged as a compliant hibernation period.

[0090] System-level operation logs include non-user business operation logs such as system backend data synchronization, log backup, system maintenance, and permission refresh. These operations do not involve file transfer business activities. If only this type of log exists during the gap period, without any compliant suspension or hibernation records, it indicates that there was no active pause in file transfer. The gap in transfer logs was caused solely by anomalies in system log collection, storage, and synchronization. This is determined to be a missing system log, and a corresponding category tag should be added.

[0091] Effective operation logs include all traceable log data such as user business operations, system-level operations, and compliance dormancy flags. If no log data is retained during the gap period, it indicates a silent interruption in the file transfer without records or traceability, which constitutes an irregular or abnormal transfer behavior and poses a risk to file transfer. This is classified as a risky transfer interruption and a corresponding category tag is added.

[0092] The raw log details retrieved during the gap window period and their corresponding cause classification labels are bound one-to-one to the corresponding potential gap interval markers in the initial state sequence, completing the data association and mounting. The final updated initial state sequence simultaneously possesses four dimensions of information: time-series flow information, gap location markers, gap log source data, and gap cause classification, achieving a full-dimensional structured record of gap anomalies.

[0093] Based on the initial state sequence containing gap interval markers and cause classification, the connection relationship and hierarchical labels of state nodes are extracted to construct an archive flow map;

[0094] Furthermore, it should be noted that:

[0095] Extract the complete sequence of state node names from the initial state sequence, remove duplicate node names within the sequence, and filter out all unique and non-repeating archive flow state nodes; assign a unique entity ID and a set of node attributes to each unique node. The node attributes include basic information such as node name, node business meaning, and the business link to which the node belongs. Complete the initialization and instantiation of the graph node entity to form the basic dataset of the knowledge graph node.

[0096] Based on the temporal flow order, traverse all adjacent node pairs in the state node name sequence. ,in As a pre-processing node, This is a subsequent flow node. A directed flow edge is constructed between the two instantiated node entities, with the direction of the edge fixed. This is used to represent the temporal flow direction and flow relationship of archives. After all adjacent nodes have completed edge construction, a topological initial archive flow graph is formed, consisting of node entities and directed flow edges, which completely restores the temporal flow path of archives.

[0097] The pre-defined organizational structure is a hierarchical system for document approval pre-entered into the system. This system includes a fixed hierarchy across the entire process, from initial handling at the grassroots level, departmental preliminary review, mid-level review, senior-level final review, and archiving management. Hierarchical levels are quantified numerically, with lower values ​​indicating higher levels (e.g., level 1 is the final review level, level 5 is the handling level). The system iterates through all nodes in the initial graph, matching the organizational structure hierarchy to the node's business function. The corresponding quantified hierarchical value is then retrieved and used as a hierarchical label, binding it to the corresponding node entity to assign hierarchical attributes to all nodes.

[0098] Define the basic attributes of directed flow edges: the starting node of each directed edge is the beginning node, corresponding to the level label L; the ending node is the end node, corresponding to the level label L. e Traverse all directed flow edges and determine the flow type: if This indicates that the file flows from a higher-level approval node to a lower-level node, which violates the normal logic of step-by-step approval. Therefore, this edge is determined to be a reverse flow edge.

[0099] Formula for calculating the absolute value of the difference between hierarchical levels: Where ΔL is the absolute value of the hierarchical jump span. Values ​​at the terminal node level. This represents the value at the starting node level.

[0100] The preset maximum span threshold for reverse jumps is a fixed quantization level difference parameter, used to determine whether a reverse jump is an abnormal risk jump. If... >A preset threshold indicates that the current reverse transfer of files involves an excessively large leap across levels, constituting an irregular or abnormal transfer. Immediately add a reverse transfer flag to the terminal node entity corresponding to this reverse transfer edge to mark the risk node; if If the value is less than or equal to the preset threshold, it is considered a normal compliance reverse jump (such as a regular rejection or return for rectification) and no marking is required. After all edges have been traversed and judged, a complete workflow graph with hierarchical labels and risk markers is finally generated.

[0101] Dependency parsing is performed on business specification documents to extract necessary nodes, dwell time boundaries, and parallel flow mutual exclusion conditions, forming a structured set of rules;

[0102] Furthermore, it should be noted that:

[0103] Define the basic attributes of directed flow edges: the starting node of each directed edge is the terminal node, and the corresponding level label is... The terminating node is the terminal node, corresponding to the hierarchical label. Traverse all directed flow edges and determine the flow type: if This indicates that the file flows from a higher-level approval node to a lower-level node, which violates the normal logic of step-by-step approval. Therefore, this edge is determined to be a reverse flow edge.

[0104] Formula for calculating the absolute value of the difference between hierarchical levels: Where ΔL is the absolute value of the hierarchical jump span. Values ​​at the terminal node level. This represents the value at the starting node level.

[0105] The preset maximum span threshold for reverse jumps is a fixed quantization level difference parameter, used to determine whether a reverse jump is an abnormal risk jump. If... >A preset threshold indicates that the current reverse transfer of files involves an excessively large leap across levels, constituting an irregular or abnormal transfer. Immediately add a reverse transfer flag to the terminal node entity corresponding to this reverse transfer edge to mark the risk node; if If the value is less than or equal to the preset threshold, it is considered a normal compliance reverse jump (such as a regular rejection or return for rectification) and no marking is required. After all edges have been traversed and judged, a complete workflow graph with hierarchical labels and risk markers is finally generated.

[0106] The extracted verb-object phrases are semantically decomposed, the operational verb components are removed, and the core object entity representing the business process of archive circulation is retained. This entity is then standardized and named as the name of the mandatory node. The mandatory node is a business status node that must be completed for compliant archive circulation; its absence is judged as a violation of circulation regulations. All extracted mandatory nodes constitute the archive circulation compliance node library.

[0107] Time constraint parsing involves semantic recognition of the time limit descriptions in compliance clauses to extract the required processing and dwell time for each mandatory node. The lower limit of the dwell time is the minimum compliant dwell time at a node (to prevent excessively rapid processing and perfunctory approvals), while the upper limit of the node dwell time boundary is the maximum compliant dwell time at a node (to prevent file backlog and processing timeouts). Both values ​​are uniformly converted into standardized second-level duration parameters, which together constitute the node dwell time boundary parameters, limiting the time range for compliant processing at each node.

[0108] Using the name of the mandatory node as a unique index, a key-value mapping relationship is constructed by binding the upper and lower limits of the corresponding node dwell time boundary. The mandatory node coverage requirement is defined as follows: the complete file flow path must include all compliant mandatory nodes, and the actual dwell time of each mandatory node must be within the upper and lower limit boundaries. This requirement is the core basic rule for subsequent flow compliance determination.

[0109] The pre-defined parallel approval logic keywords include industry standard expressions such as simultaneous approval, parallel review, multi-party review, and synchronous verification. By matching keywords to retrieve compliant clauses, parallel workflow business rules are identified. For the identified parallel processes, the names of all document nodes participating in synchronous approval and processing are precisely extracted to form a set of parallel nodes.

[0110] The core compliance logic of parallel approval processes is that parallel nodes have no sequential time hierarchy and cannot be processed serially. Based on this, a mutual exclusion constraint rule is generated: any two nodes within the same set of parallel nodes must not have a direct predecessor or successor relationship in the file transfer path. This constraint is used to determine whether parallel processing violates regulations.

[0111] By integrating duration boundary rules, full-process node coverage rules, and parallel flow mutual exclusion rules at all node levels, and removing duplicate and conflicting rules, a unified, standardized, quantifiable, and machine-recognizable set of structured rules is formed, providing a compliance benchmark for subsequent flow deviation judgment and feasibility assessment.

[0112] The archive flow map is traversed and compared with the structured rule set to generate a second judgment report on the feasibility weight of remedial operations;

[0113] Furthermore, it should be noted that:

[0114] The entire sequence of nodes in the archive flow graph, from the starting node (archive entry) to the ending node (archive archiving / rejection completion), is traversed and compared one by one with the mandatory node library and duration boundary parameters in the structured rule set. Compliance judgment criteria are divided into two categories: first, node completeness judgment, where the absence of any mandatory node constitutes a violation; second, duration compliance judgment, where the actual dwell time of a node is either less than the lower limit or greater than the upper limit. Nodes meeting either violation condition are considered deviation nodes, and the corresponding violation rule type, target mandatory node, and corresponding duration boundary parameters are precisely extracted for use as the basis for subsequent evaluation.

[0115] The predecessor node is the upstream flow node in the graph that is earlier in time than the deviation node and directly points to the deviation node. This application constructs a two-dimensional quantifiable association strength calculation model, using flow time sequence tightness and business coupling degree as core quantitative indicators. A association strength score in the range of 0 to 1 is obtained through weighted normalization of the two indicators, which is used to accurately characterize the causal contribution of the predecessor node to the current flow deviation anomaly. The higher the coupling degree and the shorter the flow time interval, the higher the association strength score. The specific calculation process is as follows:

[0116] Temporally compact quantization computation: Let the completion time of the predecessor node be... The time when the deviation node anomaly is triggered is The maximum circulation time for standard compliance is (Taken from the upper limit of the node dwell time boundary of the corresponding stage in the rule set). Time interval difference: The time series tightness S is calculated using negative correlation normalization; the smaller the time interval, the higher the tightness. The calculation formula is: When ΔT ≥ T, S = 0, indicating extremely low temporal correlation; when ΔT approaches 0, S approaches 1, indicating extremely close temporal connection.

[0117] Business coupling quantification calculation: Business coupling degree is used to characterize the closeness of the business affiliation between preceding and following nodes. The system presets three levels of coupling weight thresholds: strong coupling (nodes at different levels within the same approval chain, such as initial review → secondary review), medium coupling (nodes across stages within the same business category), and weak coupling (nodes across different business categories). The corresponding quantification scores are as follows: Strong Coupling =1.0, Medium Coupling =0.6, weak coupling =0.2. The coupling level is automatically obtained based on the preset file transfer business topology association table.

[0118] Association strength weighted fusion calculation: Set fixed fusion weights: time series weights =0.5, business coupling weight =0.5, which satisfies + =1. The final association strength Q is calculated using the following formula:

[0119]

[0120] In the formula: Q is the correlation strength between the predecessor node and the deviation node, and the value range is strictly [0,1]. The closer Q is to 1, the stronger the causal relationship between the predecessor node and the current flow deviation anomaly, and the higher the priority of responsibility tracing. The closer Q is to 0, the weaker the correlation.

[0121] Starting from the deviation node and ending at the target mandatory node corresponding to the violation, the standard normal flow path specified in the business specifications is retrieved from the graph. All compliant mandatory nodes on this path are selected and defined as intermediate mandatory nodes. The upper limit values ​​of the node dwell time boundaries corresponding to each intermediate mandatory node in the rule set are retrieved in batches as the core parameters for subsequent duration calculations.

[0122] The latest compliance trigger time of the target node is obtained by summing the upper limits of the node dwell time boundaries of all intermediate necessary nodes at the current time. The time difference between the latest compliance trigger time and the current time is calculated as the remaining available time.

[0123] Let the parameters be defined as follows: T represents the remaining available time, and Ti represents the historical average time for remedial operations. The specific method for obtaining T is as follows: The system backend accumulates historical remedial operation work order data for all similar files that deviated from the expected path. The actual time taken from the discovery of each anomaly to the completion of compliant remediation is calculated, and the average is calculated using an arithmetic mean algorithm. The formula is: , where n is the total number of valid historical remedial work orders, and T is the actual remedial time for the i-th historical work order. The data of invalid work orders and overdue invalid work orders are filtered to ensure the validity of the parameters.

[0124] Formula for calculating the basic ratio: .in This is the ratio of duration sufficiency, which is dimensionless.

[0125] like ≥1 indicates sufficient remaining time, and the time constraint score = 1; if <1 indicates insufficient remaining time; time constraint score = The final standardized time constraint score is obtained, and the higher the score, the higher the feasibility of remediation in the time dimension.

[0126] A graph topology path traversal algorithm is employed, starting from a deviating node, to retrieve all walkable flow paths and verify their compliance: paths must avoid permanently missing essential nodes, completely cover all subsequent incomplete compliant essential nodes, and comply with parallel flow mutual exclusion constraints. If all compliance conditions are met, a valid alternative path is determined to exist, and the path constraint score is 1; if no compliant alternative path exists, the score is 0, achieving a binary quantitative evaluation of path feasibility.

[0127] The necessary authorized resources for remedial operations include core resources such as document approval permissions, system operation permissions, on-duty personnel for review, and business approval limits. The system access permission ledger and personnel on-duty ledger are retrieved in real time to verify whether all necessary resources are ready at the current moment. If all are available, the resource constraint score is 1; if any core resource is missing or permissions are insufficient, the score is 0, achieving a binary quantitative assessment of resource feasibility.

[0128] Let the weighted calculation formula be: .

[0129] Parameter explanation: W is the feasibility weight of the remedial operation (final evaluation result, value range [0,1]); S is the time constraint score; S is the path constraint score; S is the resource constraint score; The preset weighting coefficients are used, and they satisfy the following conditions: The coefficients are obtained by statically pre-setting the business configuration table based on the archive's security classification, importance level, and urgency of transfer. Classified archives are assumed to be the default. Ordinary document archives by default All weighting coefficients can be manually configured and fixed in the system parameter library, ensuring that the evaluation criteria are traceable and reproducible. After weighted calculation, the feasibility weight values ​​in the 0-1 range are obtained. At the same time, all evaluation basis, parameters, and judgment results are integrated to generate a structured second judgment report.

[0130] Furthermore, it should be noted that the specific execution method for calculating the latest compliance trigger time and the remaining available time is as follows;

[0131] The system uses a real-time standardized millisecond-level timestamp as the current time T, serving as the initial benchmark for all duration calculations. The timestamps are uniformly aligned to UTC standard time to avoid time zone discrepancies. Strictly adhering to the standard workflow sequence defined by the business requirements, all intermediate necessary nodes between the deviation node and the target necessary node are extracted sequentially, forming an ordered node calculation queue to ensure that the duration accumulation order is consistent with the actual workflow sequence.

[0132] The ordered node computation queue is traversed. For each intermediate mandatory node, the upper limit value T of the standardized node dwell time boundary bound to the rule set is precisely retrieved, where k is the sequence number of the intermediate mandatory node, taking values ​​of 1, 2, 3...m (m is the total number of intermediate mandatory nodes). All duration parameters are uniformly standardized by converting milliseconds to seconds to eliminate errors caused by inconsistent time units in the original logs.

[0133] Cumulative summation calculation formula: .

[0134] Parameter definitions: T is the latest compliant trigger time of the target's mandatory nodes; T is the current system baseline time; T is the upper limit of the node dwell time boundary of the kth intermediate mandatory node; m is the total number of intermediate mandatory nodes. This formula calculates the latest operation time that does not violate the time limit rules by adding the maximum compliant dwell time of all subsequent nodes starting from the current time.

[0135] Formula for calculating remaining available time: .

[0136] Parameter Explanation: T represents the remaining available time in seconds, indicating the maximum compliant time for remedial operations. A larger value provides more leeway for remedial action. The calculation result is rounded to two decimal places to provide accurate data for calculating the time constraint score. If the result is negative, the remaining available time is determined to be 0, and the time constraint score is 0.

[0137] Based on the second judgment report, a responsibility attribution time window chain is constructed by sorting the associated strength of the precursor nodes. For nodes whose remedial operation feasibility weight is lower than the preset feasibility weight threshold, the risk propagation path depth level is marked, and a traceability chain is generated.

[0138] Furthermore, it should be noted that:

[0139] Extract the abnormal trigger time T of the deviating node from the second judgment report, and filter all flow state node records in the graph whose time series falls within the interval [T,T] to form an abnormal period node dataset. Reorder the abnormal period node dataset according to the descending order of association strength value, construct a responsibility attribution time window chain with dual matching of time series and association degree, and clarify the order of the core associated nodes responsible for the abnormality.

[0140] The preset feasibility weight threshold is a fixed system configuration parameter with a default value of 0.4. It can be fine-tuned according to business scenarios, and the threshold parameter is permanently stored in the system configuration library, making it searchable, modifiable, and traceable. All status nodes within the time window are traversed, and the feasibility weight W of the corresponding remedial operation is compared. If W < the threshold, it indicates that the node is difficult to remedy and has a high risk of diffusion. The risk propagation path depth level adopts a segmented interval mapping quantification rule: three risk levels are set, with W∈[0,0.2) representing level one (severe risk), W∈[0.2,0.4) representing level two (moderate risk), and W∈[0.4,0.6) representing level three (mild risk). W≥0.6 indicates no risk and no level is marked. The larger the level value, the deeper the risk propagation and the wider the impact. After completing the risk level marking, the responsibility attribution sorting chain and the risk level marking results are integrated to generate a complete structured traceability chain.

[0141] The traceability chain is visualized and rendered, color-coded according to the depth level of the risk propagation path, a dashed connecting box is generated for the compensation mark of missing links, a fork icon is added to the convergence point of the path branch, and an interactive view of the file flow decision is generated.

[0142] Furthermore, it should be noted that:

[0143] Read the complete flow graph (including nodes, directed edges, hierarchical labels, and reverse markers) and risk tracing chain (including responsibility time window, risk level, compensation identifier, and feasibility weight) constructed in the previous section as the entire data source for visualization rendering, ensuring that the view data is completely consistent with the algorithm analysis results.

[0144] Based on the graph topology, a force-directed graph layout algorithm is used to adaptively arrange all node entities. The complete flow topology path is drawn according to the direction and relationship of the directed flow edges. The full path topology structure of normal and reverse flow of the archive is statically restored, forming a non-interactive, full-display underlying static main view, which serves as the basic carrier for all visualization overlay effects.

[0145] The system batch filters out abnormal risk nodes in the traceability chain that have been marked with risk levels. It then performs highlight rendering on the corresponding node positions in the underlying static main view. By enlarging the node size, adding outline highlights, and increasing node transparency, it distinguishes risk nodes from normal flow nodes, enabling rapid location of risk nodes.

[0146] The preset color mapping function employs a standardized linear normalized mapping algorithm. Specifically, it calculates a normalized risk coefficient using the highest risk propagation level as the denominator and the current node level as the numerator. This 0-1 risk coefficient is then linearly mapped to the HSV color space, fixing the brightness and saturation ranges while only changing the hue parameter. The higher the risk propagation path depth level, the closer the hue is to the warning color scheme (red) and the higher the saturation; conversely, the lower the level, the closer the hue is to the safe color scheme (green) and the lower the saturation. This quantitative calculation, rather than qualitative matching, ensures precise color filling of nodes, visually distinguishing the severity of risks through color visualization, and guaranteeing consistent and reproducible view rendering rules.

[0147] Filter out deviation nodes with missing link compensation indicators, and determine the next target node that must be reached if the node is compliant, based on the standard flow path in the structured rule set. In the view, draw a dashed auxiliary connecting line with the current deviation node as the starting point and the target node as the ending point to represent the missing compliant flow path; at the same time, render a dashed border around the deviation node to highlight the abnormal node that needs to be remedied, and intuitively show the remedial operation path.

[0148] Traverse all nodes in the underlying view and determine the branching characteristics of the nodes: nodes with ≥2 input edges or ≥2 output edges are defined as path branch convergence nodes. Overlay a unique fork interaction icon at the geometric center of the node. This icon has a built-in click listener event, providing an entry point for subsequent user interaction operations to display branching details.

[0149] At the bottom of the view, a preset timeline area renders the abnormal time periods corresponding to the responsibility attribution time window chain as semi-transparent floating sliding windows. The window's time range perfectly matches the abnormal flow period. Users can manually drag the window left and right to switch and view the flow abnormalities, responsibility nodes, and risk data at different time stages, realizing interactive traceability along the timeline dimension.

[0150] At the bottom of the view, a preset timeline area renders the abnormal time periods corresponding to the responsibility attribution time window chain as semi-transparent floating sliding windows. The window's time range perfectly matches the abnormal flow period. Users can manually drag the window left and right to switch and view the flow abnormalities, responsibility nodes, and risk data at different time stages, realizing interactive traceability along the timeline dimension.

[0151] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for constructing and visualizing knowledge graphs oriented towards the temporal evolution of archives, characterized in that: include: Obtain raw operation log data and business specification documents, perform named entity recognition and timestamp extraction, and generate an initial state sequence containing gap interval markers and cause classification based on preset timestamp interval thresholds. Based on the initial state sequence containing gap interval markers and cause classification, the connection relationship and hierarchical labels of state nodes are extracted to construct an archive flow map; Dependency parsing is performed on business specification documents to extract necessary nodes, dwell time boundaries, and parallel flow mutual exclusion conditions, forming a structured set of rules; The archive flow map is traversed and compared with the structured rule set to generate a second judgment report on the feasibility weight of remedial operations; Based on the second judgment report, a responsibility attribution time window chain is constructed by sorting the associated strength of the precursor nodes. For nodes whose remedial operation feasibility weight is lower than the preset feasibility weight threshold, the risk propagation path depth level is marked, and a traceability chain is generated. The traceability chain is visualized and rendered, color-coded according to the depth level of the risk propagation path, a dashed connecting box is generated for the compensation mark of missing links, a fork icon is added to the convergence point of the path branch, and an interactive view of the file flow decision is generated.

2. The method for constructing and visualizing a knowledge graph oriented towards the temporal evolution of archives as described in claim 1, characterized in that, The generation of the initial state sequence containing gap interval markers specifically includes: Named entity recognition and timestamp extraction are performed on the original operation log data. Each log record is parsed into a tuple consisting of a status node name and an operation timestamp. All tuples are then sorted in ascending order by timestamp to generate a log entry sequence. The difference between the timestamps of every two adjacent pairs in the log entry sequence is calculated sequentially to obtain a timestamp interval sequence; Each timestamp interval in the timestamp interval sequence is compared one by one with a preset timestamp interval threshold. For each timestamp interval that is greater than the preset timestamp interval threshold, the segment between the two corresponding tuples in the log entry sequence is marked as a potential gap interval. From the log entry sequence, extract all the state node names of the binary pairs in the order of the log entry sequence to form a state node name sequence; From the log entry sequence, extract the timestamps of all tuples in the log entry sequence order to form a timestamp sequence; The state node name sequence, the timestamp sequence, and all marked potential gap intervals are integrated and associated to generate the initial state sequence containing gap interval markings.

3. The method for constructing and visualizing a knowledge graph oriented towards the temporal evolution of archives as described in claim 2, characterized in that, Generate an initial state sequence containing gap interval markers and causal classifications, specifically including: Obtain the initial state sequence and locate all marked potential gap intervals from it; For each identified potential gap, obtain its start and end timestamps. The state node in the last tuple before the start timestamp is determined as the pre-gap node, and the state node in the first tuple after the end timestamp is determined as the post-gap node. Retrieve the original operation log data that falls between the start timestamp and the end timestamp from the original operation log data; If a user suspension instruction record or authorized hibernation identifier record is retrieved, the cause of the currently processed potential gap interval is classified as a compliant hibernation period; If only system-level operation records are retrieved, but no user suspend commands or authorized hibernation identifier records are retrieved, the cause is classified as missing system logs. If no valid operation log record is found, the cause is classified as a risk-related interruption. The original operation log data is appended to the corresponding potential gap interval markers in the initial state sequence to obtain an initial state sequence containing causal classification labels.

4. The method for constructing and visualizing a knowledge graph oriented towards the temporal evolution of archives as described in claim 1, characterized in that, The construction of the archive flow map specifically includes: Obtain the sequence of state node names in the initial state sequence, and instantiate each unique state node name in the sequence of state node names into a node entity; Based on the order of the state nodes in the state node name sequence, a directed flow edge is created between every two adjacent node entities, pointing from the previous node to the next node, forming an initial file flow graph. For each node entity in the initial archive flow map, its corresponding level is queried from the preset organizational structure, and the queried level is assigned as a level label to the node entity. Traverse all directed flow edges in the initial file flow graph, identify reverse flow edges whose terminal node level label is lower than the starting node level label, calculate the absolute value of the level difference between the two nodes of each reverse flow edge, and if the absolute value of the level difference is greater than the preset maximum reverse jump threshold, add a reverse flow marker to the terminal node entity of the reverse flow edge to obtain the file flow graph.

5. The method for constructing and visualizing a knowledge graph oriented towards the temporal evolution of archives as described in claim 1, characterized in that, The set of rules for forming a structured set specifically includes: Obtain the compliance clause paragraphs from the business specification document, perform dependency parsing on the compliance clause paragraphs, and extract verb-object phrases describing necessary operational steps; Identify and extract the object component representing the file processing stage from the verb-object phrase, and use it as the name of the necessary node; The time constraint parsing of the compliance clause paragraph is performed to extract the upper limit and lower limit of the node dwell time boundary that limits the name of the required node, which are used as the node dwell time boundary parameters. Establish a mapping relationship between the names of the required nodes and their corresponding node dwell time boundary parameters, which serves as the required node coverage requirement; If keywords characterizing the parallel approval logic are identified in the compliance clause paragraph, the names of multiple approval nodes participating in the parallel process are extracted. Based on the extracted names of multiple approval nodes, parallel flow mutual exclusion constraints are generated, which stipulate that these nodes cannot be direct predecessors or successors of each other in the flow path. The structured rule set is generated by combining all the required coverage of necessary nodes with all the parallel flow mutual exclusion constraints.

6. The method for constructing and visualizing a knowledge graph oriented towards the temporal evolution of archives as described in claim 1, characterized in that, The second determination report for generating the feasibility weight of the remedial operation specifically includes: The integrity of the state node sequence from the starting node to the ending node in the archive flow map is traversed and compared. If the node sequence in the archive flow map has a missing mandatory node recorded in the rule set or the actual dwell time of a certain node exceeds the node dwell time boundary in the rule set, then the node is recorded as a deviation node, and the node dwell time boundary rule of the target mandatory node violated by the deviation node is obtained. The predecessor node of the deviation node in the archive flow map is traced back, the correlation strength between the deviation node and the predecessor node is quantified, the missing link of the deviation node is determined, and a missing link compensation identifier is generated for it. Identify all intermediate necessary nodes on the normal flow path from the deviation node to the target necessary node in the archive flow map, and retrieve the upper limit value of the node dwell time boundary of all intermediate necessary nodes from the structured rule set; The latest compliance trigger time of the target node is obtained by summing the upper limits of the node dwell time boundaries of all the intermediate necessary nodes at the current time, and the time difference between the latest compliance trigger time and the current time is calculated as the remaining available time. Obtain the historical average time taken to perform remedial operations to restore the flow to the target node, calculate the ratio of the remaining available time to the historical average time, and compare them to obtain the time constraint score; In the archive flow map, a path search is used to determine whether there is an alternative path that starts from the deviation node, bypasses all missing necessary nodes, and arrives at all subsequent unfinished necessary nodes in sequence. If such a path exists, the path constraint score is set to 1; otherwise, it is set to 0. Query the availability status of the authorized resources necessary to perform the remedial operation at the current moment. If available, the resource constraint score is set to 1; otherwise, it is set to 0. The time constraint score, the path constraint score, and the resource constraint score are weighted and summed to obtain a second determination report on the feasibility weight of the remedial operation.

7. The method for constructing and visualizing a knowledge graph oriented towards the temporal evolution of archives as described in claim 6, characterized in that, The step of summing up the upper limits of the current time and the node dwell time boundaries of all the intermediate necessary nodes to obtain the latest compliance trigger time of the target necessary node specifically includes: The current time is used as the calculation reference point. According to the flow order from the deviation node to the target necessary node, each intermediate necessary node on the normal flow path is obtained in turn. From the structured set of rules, query and obtain the upper limit value of the dwell time boundary for each of the intermediate necessary nodes; Using the current time as the initial cumulative value, the upper limit value of the dwell time boundary of each intermediate necessary node is added one by one in the flow order to complete the cumulative summation. The final result of the cumulative summation is determined as the latest compliance trigger time of the target necessary node. The difference between the latest compliance trigger time and the current time is the remaining available time.

8. The method for constructing and visualizing a knowledge graph oriented towards the temporal evolution of archives as described in claim 7, characterized in that, The generation of the traceability chain specifically includes: According to the second determination report, all state node records from the time the deviation node was triggered to the current time are extracted. All state nodes are sorted in descending order of their association strength with the predecessor node of the deviation node to construct a responsibility attribution time window chain. If the feasibility weight of the remedial operation of any state node is lower than the preset feasibility weight threshold, the node is labeled with the risk propagation path depth level to generate a traceability chain containing the responsibility attribution time window and the risk propagation path depth level.

9. The method for constructing and visualizing a knowledge graph oriented towards the temporal evolution of archives as described in claim 8, characterized in that, The generated file transfer decision interaction view specifically includes: Obtain the archive circulation map and the traceability chain; Based on the node entities and directed flow edges in the aforementioned archive flow graph, a low-level static main view is rendered and generated; Read all nodes marked with the depth level of the risk propagation path from the traceability chain, and highlight these nodes on the bottom static main view; Using a preset color mapping function, the risk propagation path depth level value of each highlighted node is mapped to the corresponding hue and saturation, and the corresponding node is filled with color. Read all nodes with missing link compensation identifiers from the traceability chain, generate a dashed auxiliary connection line for each such node that starts from itself and points to the next necessary node it should flow to, and render a dashed connection box around the node. For all path branch convergence nodes with multiple input edges or multiple output edges in the underlying static main view, a fork icon that can respond to click commands is superimposed and rendered at its geometric center. The chain of responsibility attribution time windows is rendered as a semi-transparent sliding window that can be dragged along the timeline, and superimposed on the timeline area of ​​the underlying static main view. When a click command is received on any highlighted node, an information overlay is generated at the top of the view. Within this information overlay, the abnormal flow causal anchor point pattern description and remedial operation feasibility weight value corresponding to the node in the second judgment report are displayed in a structured manner, thus obtaining the file flow decision interaction view.

Images