A media terminal data analysis management method based on a knowledge graph

Abstract

The application discloses a kind of media terminal data analysis management methods based on knowledge graph, comprising the following steps: step one: obtaining event record and standardization alignment;Step two: build the knowledge graph of media terminal containing entity node, relationship edge, time attribute;Step three: generate basic constraint set and candidate constraint set containing time window;Step four: establish module mapping and aggregate to generate abstract constraint set;Step five: interval tree screening and determine shrink to get minimum conflict time window;Step six: minimum conflict time window is executed in, and proof learning type QuickXPlain is constructed conflict hypergraph and obtains conflict module set;Step seven: refine conflict module set and solve minimum conflict constraint set, generate conflict explanation record and management instruction.The application improves conflict positioning efficiency and explanation consistency.

Description

Technical Field

[0001] This invention relates to the field of media terminal data analysis and management technology, and in particular to a media terminal data analysis and management method based on knowledge graphs. Background Technology

[0002] Media terminals continuously generate event logs in scenarios such as content distribution, copyright management, frequency control scheduling, session organization, and event tracking. These event logs are scattered across different acquisition paths, system clocks, and field specifications, resulting in fragmented data. Businesses typically manage and analyze this data using relational tables, log indexes, and rule engines. Existing implementations usually rely on field mapping and offline cleaning to standardize fields, then use timestamp concatenation to establish cross-source associations, followed by statistical analysis and verification of sessions, copyright domains, and frequency control cycles using fixed rules. With increasing business scale and frequent policy iterations, the connection relationships and temporal evolution characteristics of event logs become difficult to express in a unified structure. Cross-session and cross-policy association analysis often requires repeatedly building intermediate tables and multiple rounds of screening, leading to complex analysis processes and unstable consistency verification standards.

[0003] For consistency verification and conflict localization, existing technologies often employ static constraint verification, rule-by-rule comparison, and conflict log backtracking to locate the source of anomalies. Some solutions introduce constraint solvers to output unsatisfiable sets, but most rely on direct solving across the entire time range and the entire constraint set. The full-scale solution method requires frequent calls to consistency checks as the size of candidate constraints increases, and conflict localization depends on repeated reduction and testing, leading to a rapid increase in solution time and the number of candidate constraints. In the time dimension, conflicts are often triggered by specific time windows. Existing solutions lack a structured contraction mechanism oriented towards the time axis, and conflict window localization relies on manually setting the time granularity and repeated trials. This easily results in windows that are too large, increasing the solution burden, or windows that are too small, causing the conflict chain to be truncated and localization to fail.

[0004] Existing technologies for organizing and reusing conflict results primarily output in the form of rule hit records or lists of abnormal sessions, lacking a structured expression for modular management. This results in constraint conflicts under the same combination of session identifier, copyright domain identifier, and frequency control cycle identifier failing to form stable aggregation units. Conflict location results are difficult to reuse in policy iteration, and there is a lack of clear mapping between conflict causes and management actions. Furthermore, existing solution processes often output several rules or constraint items involved in the conflict, lacking a conflict proof mechanism that meets the minimum requirement. Conflict explanation records are difficult to maintain consistent granularity and boundaries, and management instruction generation relies on manual context completion and secondary screening. It is difficult to complete the closed-loop processing from conflict discovery to conflict module location to the determination of the minimum conflict constraint set in a unified process.

[0005] Therefore, how to provide a knowledge graph-based method for media terminal data analysis and management is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] One objective of this invention is to propose a knowledge graph-based method for media terminal data analysis and management. This invention utilizes event records to construct a temporal knowledge graph and introduces module mapping, interval tree contraction, and conflict proof solution mechanisms to achieve conflict time window location, conflict module identification, and minimum conflict constraint set solution under complex constraints. It has the advantages of high conflict location accuracy, stable solution efficiency, high degree of structured conflict interpretation, and strong consistency in management instruction generation.

[0007] A method for media terminal data analysis and management based on knowledge graphs according to an embodiment of the present invention includes the following steps: Step 1: Retrieve event records and perform field normalization and time alignment; Step 2: Construct a media terminal knowledge graph based on event records. The media terminal knowledge graph includes entity nodes, relationship edges, and time attributes. Step 3: Generate a basic constraint set and a candidate constraint set based on the pattern constraints and management rules of the media terminal knowledge graph, and associate the candidate constraint set with a time window; Step 4: Establish module mapping. Module keys include session identifier, copyright domain identifier, and frequency control period identifier. Aggregate candidate constraint sets according to module mapping to form module sets, and generate summary constraint sets corresponding to the module sets. Step 5: Build an interval tree, filter the basic constraint set and the candidate constraint set covering the time window interval to form an interval constraint subset, perform consistency judgment on the interval constraint subset and shrink the interval tree to obtain the minimum conflict time window; Step 6: Perform proof learning QuickXPlain on the summary constraint set within the minimum conflict time window. Consistency determination generates conflict proof subsets. Write the conflict proof subsets into the conflict hypergraph to form proof hyperedges. When the interval constraint subsets cover the proof hyperedges, determine inconsistency and obtain the conflict module set. Step 7: Based on the module mapping, refine the conflicting module set into atomic candidate constraints and update the summary constraint set. Solve to obtain the minimum conflicting constraint set, generate conflict interpretation records, and generate management instructions.

[0008] Optionally, step one specifically includes: Obtain a set of event entries, which includes a terminal identifier field, a session identifier field, an event type field, a timestamp field, and a payload field. Perform field standardization on the event entries, which includes standardizing field names, field values, data types, and filling in missing values ​​to obtain standardized event entries. A baseline timeline is determined, which is a baseline timestamp sequence. Time alignment is performed on the timestamp fields of standardized event entries based on the baseline timeline. Time alignment includes time scale conversion and time offset correction to obtain aligned timestamps. The standardized event entries are sorted based on the aligned timestamps and aggregated by the terminal identifier field and the session identifier field to obtain the event records.

[0009] Optionally, step two specifically includes: Based on the event logs, a set of terminal nodes, a set of session nodes, and a set of event nodes are generated. The set of terminal nodes is determined by the terminal identifier field, the set of session nodes is determined by the session identifier field, and the set of event nodes is determined by the event type field and the aligned timestamp field. The terminal nodes, session nodes, and event nodes are set with node identifiers and node type identifiers. A set of relational edges is generated based on event records. The set of relational edges includes session association edges from terminal nodes to session nodes and session event edges from session nodes to event nodes. Relationship edges are set with edge identifiers and edge type identifiers. The alignment timestamp field is written into the event node time attribute and the session event edge time attribute, the event type field is written into the event node attribute, and the terminal identifier field and session identifier field are written into the corresponding node attribute, forming a media terminal knowledge graph containing entity nodes, relationship edges and time attributes.

[0010] Optionally, step three specifically includes: Based on the knowledge graph of media terminals, a pattern constraint rule set is established using node type identifiers, edge type identifiers, node attribute sets, and edge attribute sets. The pattern constraint rule set limits the connection relationship between node type identifiers and edge type identifiers, limits the value range and missing condition of node attribute sets and edge attribute sets, and limits the consistency condition between edge time attributes and node time attributes. A basic constraint set is generated based on the pattern constraint rule set, and the basic constraints of the basic constraint set set the constraint identifier, constraint object identifier, and constraint judgment condition. A candidate constraint generation rule set is established based on management rules. The candidate constraint generation rule set limits the rule fields and threshold fields associated with session identifier, copyright domain identifier, and frequency control cycle identifier. Based on the candidate constraint generation rule set, the associated node identifier and associated edge identifier are extracted from the media terminal knowledge graph to generate a candidate constraint set. The candidate constraint set sets the constraint identifier, associated node identifier set, associated edge identifier set, and constraint judgment conditions. Configure a time window for the candidate constraint set. The time window sets a start timestamp and an end timestamp. The start timestamp is the minimum value of the time attribute of the associated edge, and the end timestamp is the maximum value of the time attribute of the associated edge. Write the time window into the candidate constraints of the candidate constraint set.

[0011] Optionally, step four specifically includes: Based on the candidate constraint set, a module key is generated for the candidate constraint. The module key includes a session identifier, a copyright domain identifier, and a frequency control cycle identifier. The session identifier is determined by locating the session node based on the associated node identifier set of the candidate constraint and reading the session identifier field. The copyright domain identifier is determined by assigning a value to the candidate constraint according to the management rules. The frequency control cycle identifier is determined by periodically mapping the start timestamp of the time window of the candidate constraint according to the period division rules set by the management rules. A module mapping is established, and the module mapping records the constraint identifier and the module key. Candidate constraint sets are grouped by module key to form module sets, and the module sets record the module key and constraint identifier sequence. A summary constraint set is generated based on the module set. The summary constraint record of the summary constraint set includes the module key, constraint identifier sequence, and summary time window. The start timestamp of the summary time window is the minimum value of the start timestamp of the time window corresponding to the constraint identifier sequence, and the end timestamp of the summary time window is the maximum value of the end timestamp of the time window corresponding to the constraint identifier sequence.

[0012] Optionally, step five specifically includes: The starting timestamp of the time axis range is determined based on the minimum starting timestamp of the time window of the candidate constraints in the candidate constraint set, and the ending timestamp of the time axis range is determined based on the maximum ending timestamp of the time window of the candidate constraints in the candidate constraint set. An interval tree is built on the time axis range, the interval of the root node of the interval tree is the time axis range, and the interval tree node records the starting timestamp and ending timestamp of the node interval. For each interval tree node, an interval constraint subset is constructed. The interval constraint subset contains the basic constraint set. The interval constraint subset contains candidate constraints within the candidate constraint set that satisfy the condition that the start timestamp of the time window is not greater than the start timestamp of the node interval and the end timestamp of the time window is not less than the end timestamp of the node interval. A consistency determination is performed on the interval constraint subset and the consistency status is written to the corresponding interval tree node. For interval tree nodes with inconsistent consistency, interval shrinkage is performed. Interval shrinkage uses the median of the start and end timestamps of the interval as the split timestamp to generate left and right sub-intervals. Left and right child nodes are constructed separately, and consistency is determined separately. When both the left and right child nodes are consistent, the node interval is determined as the minimum conflict time window. When both the left and right child nodes are inconsistent, the left child node is selected as the node to be shrunk. When both the left and right child nodes are consistent, the right child node is selected as the node to be shrunk. When both the left and right child nodes are inconsistent, the child node with the smaller interval length is selected as the node to be shrunk. When the interval length of the node to be shrunk is less than or equal to a preset time granularity threshold, the node interval to be shrunk is determined as the minimum conflict time window.

[0013] Optionally, step six specifically includes: Based on the start and end timestamps of the minimum conflict time window, the summary constraints with a start timestamp of the summary time window not greater than the start timestamp of the minimum conflict time window and an end timestamp of the summary time window not less than the end timestamp of the minimum conflict time window are selected from the summary constraint set to form a subset of time window summary constraints. A consistency determination is performed on the basic constraint set and the time window summary constraint subset. The consistency determination calculates and summarizes the constraint determination conditions of the basic constraint set and the constraint determination conditions of the time window summary constraint subset one by one. When all the summary results are satisfied, the consistency state is determined to be consistent. When there are any unsatisfied summary results, the consistency state is determined to be inconsistent. Under the condition of inconsistent consistency, a proof-learning QuickXPlain solution is performed on the time window summary constraint subset. The proof-learning QuickXPlain sets the summary constraint subset to be reduced as the time window summary constraint subset and divides the summary constraint subset to be reduced into a left partition subset and a right partition subset according to the median index of the summary constraint identifier sequence. A consistency determination is performed on the left partition subset based on the basic constraint set. If the consistency status of the left partition subset is inconsistent, the summary constraint subset to be reduced is updated to the left partition subset and the division continues. If the consistency status of the left partition subset is consistent, the right partition subset is used as the summary constraint subset to be reduced and the left partition subset is added to the basic constraint set before the division continues. If the number of elements in the summary constraint subset to be reduced is equal to 1, the summary constraint subset to be reduced is added to the conflict proof subset. Perform a minimality test on the conflict proof subset. The minimality test satisfies that the joint consistency state of the basic constraint set and the conflict proof subset is inconsistent, and the joint consistency state of the basic constraint set and the conflict proof subset after deleting any summary constraint is consistent. Write the conflict proof subset into the conflict hypergraph to form a proof hyperedge. The proof hyperedge records the summary constraint identifier sequence of the conflict proof subset. The proof hyperedge records the module key corresponding to the conflict proof subset. The proof hyperedge records the constraint identifier sequence corresponding to the conflict proof subset. The coverage determination is performed based on the interval constraint subset corresponding to the minimum conflict time window. If the coverage determination satisfies the condition that the interval constraint subset contains the constraint identifier sequence of the proof superedge record, the determination is inconsistent, and the module key of the proof superedge record is written into the conflict module set.

[0014] Optionally, step seven specifically includes: Based on the module mapping, locate the constraint identifier sequence corresponding to the module key recorded in the conflict module set. Based on the constraint identifier sequence, locate the candidate constraint in the candidate constraint set. Write the candidate constraint into the atomic candidate constraint set. The atomic candidate constraint retains the constraint identifier, associated node identifier set, associated edge identifier set, constraint decision condition, and time window. The summary constraint set is updated based on the atomic candidate constraint set. The summary constraint record module key and constraint identifier sequence are used. The start timestamp of the summary time window is the minimum value of the start timestamp of the time window corresponding to the constraint identifier sequence. The end timestamp of the summary time window is the maximum value of the end timestamp of the time window corresponding to the constraint identifier sequence. Based on the basic constraint set, minimum conflict time window, and summary constraint set, the proof-learning QuickXPlain is used to solve and map the minimum conflict constraint set. The minimum conflict constraint set satisfies the condition that removing any atomic candidate constraint in the minimum conflict constraint set determines the consistency state as consistent. Write the constraint identifier sequence, module key, minimum conflict time window start timestamp, and minimum conflict time window end timestamp of the minimum conflict constraint set into the conflict interpretation record. Generate management instructions based on the conflict interpretation record. The management instructions include the module key, constraint identifier sequence, minimum conflict time window start timestamp, and minimum conflict time window end timestamp.

[0015] The beneficial effects of this invention are: This invention standardizes and aligns event records by field and time, constructing a media terminal knowledge graph containing entity nodes, relation edges, and time attributes on a unified timeline. This allows session relationships, event evolution, and management rules to be expressed and constrained within the same structure. Based on the knowledge graph, pattern constraints and management rules are introduced to generate basic constraint sets and candidate constraint sets. Through module mapping, candidate constraints are aggregated according to session identifiers, copyright domain identifiers, and frequency control cycle identifiers, effectively reducing the granular complexity of constraint solving. This allows subsequent analysis to be carried out at the structured module level, thereby avoiding the scale expansion problem caused by all constraints directly participating in the calculation.

[0016] In the conflict analysis process, this invention recursively shrinks the time axis using an interval tree to accurately locate the minimum conflict time window that causes inconsistencies. Within this time window, a proof-learning QuickXPlain is introduced to generate a subset of conflict proofs that meet the minimumity requirement. The proof results are then organized in the form of a conflict hypergraph to achieve stable identification and reuse of conflict modules. On this basis, the conflict modules are further refined into atomic candidate constraints and the minimum conflict constraint set is solved, so that the conflict interpretation record has clear boundaries and consistent granularity. Management instructions can be directly generated based on the minimum conflict constraint set, thereby improving the accuracy of conflict location, the controllability of the solution process, and the certainty and executability of management decisions. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1This is a flowchart of a media terminal data analysis and management method based on knowledge graphs proposed in this invention; Figure 2 This is a schematic diagram illustrating the conflict hypergraph construction and coverage determination of a knowledge graph-based media terminal data analysis and management method proposed in this invention. Detailed Implementation

[0018] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0019] refer to Figure 1-2 A knowledge graph-based method for media terminal data analysis and management includes the following steps: Step 1: Retrieve event records and perform field normalization and time alignment; Step 2: Construct a media terminal knowledge graph based on event records. The media terminal knowledge graph includes entity nodes, relationship edges, and time attributes. Step 3: Generate a basic constraint set and a candidate constraint set based on the pattern constraints and management rules of the media terminal knowledge graph, and associate the candidate constraint set with a time window; Step 4: Establish module mapping. Module keys include session identifier, copyright domain identifier, and frequency control period identifier. Aggregate candidate constraint sets according to module mapping to form module sets, and generate summary constraint sets corresponding to the module sets. Step 5: Build an interval tree, filter the basic constraint set and the candidate constraint set covering the time window interval to form an interval constraint subset, perform consistency judgment on the interval constraint subset and shrink the interval tree to obtain the minimum conflict time window; Step 6: Perform proof learning QuickXPlain on the summary constraint set within the minimum conflict time window. Consistency determination generates conflict proof subsets. Write the conflict proof subsets into the conflict hypergraph to form proof hyperedges. When the interval constraint subsets cover the proof hyperedges, determine inconsistency and obtain the conflict module set. Step 7: Based on the module mapping, refine the conflicting module set into atomic candidate constraints and update the summary constraint set. Solve to obtain the minimum conflicting constraint set, generate conflict interpretation records, and generate management instructions.

[0020] In this embodiment, step one specifically includes: Obtain a set of event entries. Each event entry contains a terminal identifier field, a session identifier field, an event type field, a timestamp field, and a payload field. Field standardization includes unifying field names, field values, data types, and filling missing values. Unifying field names means mapping the event entry field names to a preset set of standard field names. Unifying field values ​​means converting the event type field value and the terminal identifier field value to standard values ​​according to a preset mapping table. Unifying data types means converting the terminal identifier field and the session identifier field to string types, the timestamp field to a numeric type with a uniform time unit, and the payload field to a numeric type. Filling missing values ​​means writing preset default values ​​to missing fields and marking the missing status, thus obtaining standardized event entries. A baseline timeline is determined, which is a baseline timestamp sequence. Time alignment is performed on the timestamp fields of standardized event entries based on the baseline timeline. Time alignment includes time scale conversion and time offset correction. Time scale conversion refers to converting the units and rates of the timestamp fields according to the time scale parameter. Time offset correction refers to correcting the starting zero point of the timestamp fields according to the time offset parameter. The time scale parameter represents the time unit conversion ratio, and the time offset parameter represents the difference of the time starting point, thus obtaining the aligned timestamps. Standardized event entries are sorted based on aligned timestamps, with the sorting rule being ascending alignment timestamps. Aggregation is performed by terminal identifier field and session identifier field, with the aggregation rule being that if the terminal identifier field and the session identifier field are the same, they are grouped into the same aggregation group. The sorting result is maintained within the aggregation group to obtain the event record.

[0021] In this embodiment, step two specifically includes: Based on event records, a terminal node set, a session node set, and an event node set are generated. The terminal node set is formed by deduplicating the terminal identifier field in the event record to obtain a terminal identifier set, and assigning a node identifier to each terminal identifier according to the terminal identifier set. The node type identifier is set to the terminal node type. The session node set is formed by deduplicating the session identifier field in the event record to obtain a session identifier set, and assigning a node identifier to each session identifier according to the session identifier set. The node type identifier is set to the session node type. The event node set is formed by using the event type field and the alignment timestamp field in the event record to construct the event key. The event key is formed by concatenating the values ​​of the event type field and the alignment timestamp field. The event key set is obtained by deduplicating the event keys to obtain an event key set, and assigning a node identifier to each event key. The node type identifier is set to the event node type. A set of relational edges is generated based on event records. The set of relational edges includes session-related edges and session event edges. Session-related edges are formed by the terminal identifier field and the session identifier field as edge keys. The edge keys are formed by concatenating the values ​​of the terminal identifier field and the session identifier field. The edges are deduplicated according to the edge keys and each edge key is assigned an edge identifier to form a directed edge from the terminal node to the session node. The edge type identifier is set to the session-related edge type. Session event edges are formed by the session identifier field and the event key as edge keys. The edge keys are formed by concatenating the values ​​of the session identifier field and the event key. The edges are deduplicated according to the edge keys and each edge key is assigned an edge identifier to form a directed edge from the session node to the event node. The edge type identifier is set to the session event edge type. The alignment timestamp field is written to the event node time attribute and the session event edge time attribute. The event node time attribute is written to the alignment timestamp field value in the corresponding event key. The session event edge time attribute is written to the alignment timestamp field value in the corresponding event key. The event type field is written to the event node attribute. The event node attribute is written to the event type field value in the corresponding event key. The terminal identifier field and the session identifier field are written to the corresponding node attribute. The terminal node attribute is written to the terminal identifier field value. The session node attribute is written to the session identifier field value, forming a media terminal knowledge graph containing entity nodes, relation edges, and time attributes.

[0022] In this embodiment, step three specifically includes: A pattern constraint rule set is established based on the knowledge graph of media terminals, including node type identifiers, edge type identifiers, node attribute sets, and edge attribute sets. Node type identifiers are derived from terminal node type, session node type, and event node type. Edge type identifiers are derived from session-related edge type and session-event edge type. The node attribute set includes the terminal identifier field, session identifier field, event type field, and event node time attribute. The edge attribute set includes the session-event edge time attribute. The connection relationships in the pattern constraint rule set are used to restrict terminal node types to connect only to session node types via session-related edge types, and session node types to connect only to event node types via session-event edge types. The values ​​of the pattern constraint rule set are... The scope is used to limit the terminal identifier field and session identifier field to string values, the event type field to a preset set of event type values, the missing condition of the pattern constraint rule set is used to limit the node attribute set field to null values ​​as missing, the time consistency condition of the pattern constraint rule set is used to limit the event node time attribute to be equal to the corresponding session event edge time attribute, the basic constraint set is generated based on the pattern constraint rule set, the basic constraints of the basic constraint set correspond to the rule items of the pattern constraint rule set, the basic constraints set the constraint identifier, the constraint action object identifier is written into the node identifier and edge identifier, and the constraint judgment condition is written into the connection relationship condition, the value range condition, the missing condition, and the time consistency condition. A candidate constraint generation rule set is established based on management rules. The candidate constraint generation rule set includes session identifier association rule items, copyright domain identifier association rule items, and frequency control cycle identifier association rule items. The session identifier association rule items are used to specify the combination of node type identifier and edge type identifier corresponding to the session identifier field. The copyright domain identifier association rule items are used to specify the combination constraint relationship between the event type field value and the session node identifier. The frequency control cycle identifier association rule items are used to specify the cycle division rule between the session identifier field and the event node time attribute. The candidate constraint generation rule set includes rule fields and threshold fields. The rule fields are used to indicate the node attribute set field and edge attribute set field participating in the judgment. The threshold field is used to indicate the numerical threshold or set threshold configured by the management rules. Based on the candidate constraint generation rule set, the associated node identifier and associated edge identifier are extracted from the media terminal knowledge graph to generate a candidate constraint set. The candidate constraints set constraint identifiers are set. The associated node identifier set is written with the extracted node identifiers. The associated edge identifier set is written with the extracted edge identifiers. The constraint judgment conditions are written with the comparison conditions corresponding to the rule fields and the threshold conditions corresponding to the threshold fields. Configure a time window for the candidate constraint set. The time window sets a start timestamp and an end timestamp. The start timestamp is determined by locating the session event edge according to the associated edge identifier set of the candidate constraint and reading the time attribute of the session event edge, taking the minimum value. The end timestamp is determined by locating the session event edge according to the associated edge identifier set of the candidate constraint and reading the time attribute of the session event edge, taking the maximum value. Write the start timestamp and end timestamp into the time window field of the candidate constraint.

[0023] In this embodiment, step four specifically includes: Based on the candidate constraint set, a module key is generated for the candidate constraint. The module key includes a session identifier, a copyright domain identifier, and a frequency control period identifier. The session identifier is determined by locating the session node identifier through the associated node identifier set of the candidate constraint and reading the value of the session identifier field in the session node attribute. The copyright domain identifier is determined by the value of the copyright domain identifier field written in the constraint judgment condition of the candidate constraint according to the management rules. The frequency control period identifier is determined by performing period mapping on the start timestamp of the time window of the candidate constraint according to the period division rules set by the management rules. The period mapping maps the start timestamp of the time window to the period number. The module key is formed by concatenating the values ​​of the session identifier field, the copyright domain identifier field, and the period number. A module mapping is established. The module mapping records the constraint identifier of the candidate constraint and the module key and saves the one-to-one correspondence between the constraint identifier and the module key. Candidate constraint sets are grouped by module key to form module sets. The grouping rule is that candidate constraints with the same module key value are grouped into the same group. A module record is created for each group in the module set. The module record is written with the module key and the constraint identifier sequence of the candidate constraints in the group. The constraint identifier sequence is sorted in ascending order according to the start timestamp of the time window of the candidate constraint. A summary constraint set is generated based on the module set. The summary constraint set generates a summary constraint record for each module record. The summary constraint record is written to the module key and the constraint identifier sequence. The start timestamp of the summary time window is the minimum value of the start timestamp of the candidate constraint time window corresponding to the constraint identifier sequence. The end timestamp of the summary time window is the maximum value of the end timestamp of the candidate constraint time window corresponding to the constraint identifier sequence. The summary constraint record writes the summary time window into the summary time window field and saves it in association with the module key.

[0024] In this embodiment, step five specifically includes: The starting timestamp of the time axis range is determined based on the minimum starting timestamp of the time window of the candidate constraints in the candidate constraint set. The starting timestamp of the time window of the candidate constraint is taken from the starting value of the time window field in the candidate constraint record. The minimum value of the starting timestamp of the time axis range is obtained by comparing all candidate constraints within the candidate constraint set. The ending timestamp of the time axis range is determined based on the maximum value of the ending timestamp of the time window of the candidate constraints in the candidate constraint set. The ending timestamp of the time window of the candidate constraint is taken from the ending value of the time window field in the candidate constraint record. The maximum value of the ending timestamp of the time axis range is obtained by comparing all candidate constraints within the candidate constraint set. An interval tree is built on the time axis range. The interval of the root node of the interval tree is the time axis range. The interval tree node records the starting timestamp and ending timestamp of the node interval. The starting timestamp of the node interval is taken from the start of the time period covered by the node, and the ending timestamp of the node interval is taken from the end of the time period covered by the node. For each interval tree node, an interval constraint subset is constructed. The interval constraint subset contains the basic constraint set, which remains consistent across each interval tree node. The interval constraint subset contains candidate constraints within the candidate constraint set that satisfy the condition that the start timestamp of the time window is not greater than the start timestamp of the node interval and the end timestamp of the time window is not less than the end timestamp of the node interval. The condition is determined by comparing the start value of the candidate constraint time window field with the start timestamp of the node interval and the end value of the candidate constraint time window field with the end timestamp of the node interval. A consistency check is performed on the interval constraint subset, and the consistency status is written to the corresponding interval tree node. The consistency check calculates and summarizes the constraint check conditions for the candidate constraints of the basic constraint set and the interval constraint subset. When all the summarized results are satisfied, the consistency status is written as consistent; when some of the summarized results are not satisfied, the consistency status is written as inconsistent. For interval tree nodes with inconsistent consistency states, interval shrinkage is performed. Interval shrinkage uses the median of the node interval's start and end timestamps as the split timestamp to generate left and right sub-intervals. The median timestamp is the arithmetic mean of the node interval's start and end timestamps, rounded to the nearest integer. The left sub-interval's start timestamp is taken from the node interval's start timestamp, and its end timestamp is taken from the split timestamp. The right sub-interval's start timestamp is taken from the split timestamp, and its end timestamp is taken from the node interval's end timestamp. Left and right child nodes are constructed separately, and consistency checks are performed on each. A node is determined when both the left and right child nodes are consistent. The interval is the minimum conflict time window. When the consistency status of the left child node is inconsistent and the consistency status of the right child node is consistent, the left child node is selected as the node to be shrunk. When the consistency status of the left child node is consistent and the consistency status of the right child node is inconsistent, the right child node is selected as the node to be shrunk. When the consistency status of the left child node is inconsistent and the consistency status of the right child node is inconsistent, the child node with the smaller interval length is selected as the node to be shrunk. The interval length is obtained by subtracting the start timestamp of the node interval from the end timestamp of the node interval. When the interval length of the node to be shrunk is less than or equal to the preset time granularity threshold, the interval of the node to be shrunk is determined as the minimum conflict time window. The preset time granularity threshold is the minimum time length threshold allowed by the minimum conflict time window.

[0025] In this embodiment, step six specifically includes: Based on the start and end timestamps of the minimum conflict time window, the start and end timestamps of the summary time window of each summary constraint record in the summary constraint set are read one by one. Summary constraints that satisfy the condition that the start timestamp of the summary time window is not greater than the start timestamp of the minimum conflict time window and the end timestamp of the summary time window is not less than the end timestamp of the minimum conflict time window are written into the time window summary constraint subset. The time window summary constraint subset maintains the fixed order of the summary constraint identifier sequence. Consistency determination is performed on the basic constraint set and the time window summary constraint subset. For each basic constraint in the basic constraint set, the consistency determination reads the constraint object identifier and locates the node attributes and edge time attributes of the media terminal knowledge graph. Based on the constraint determination conditions of the basic constraint records, the field value range verification, missing value determination verification, and time consistency verification are completed to form the basic constraint determination result. For each summary constraint in the time window summary constraint subset, the consistency determination reads the constraint identifier sequence of the summary constraint record and locates the candidate constraint record. Based on the constraint determination conditions of the candidate constraint record, the threshold comparison and relation constraint verification are completed to form the summary constraint determination result. The consistency determination summarizes the basic constraint determination result and the summary constraint determination result. When all the summary results are satisfied, the consistency state is determined to be consistent. When there are any unsatisfied summaries, the consistency state is determined to be inconsistent. Under the condition of inconsistency in the consistency state, a proof-learning QuickXPlain solution is performed on the time window summary constraint subset. The proof-learning QuickXPlain solution sets the subset of summary constraints to be reduced as the time window summary constraint subset and determines the median index according to the summary constraint identifier sequence. The median index corresponds to the middle position of the summary constraint identifier sequence. Based on the median index, the subset of summary constraints to be reduced is divided into a left partition subset and a right partition subset. The left partition subset contains the summary constraints before the median index, and the right partition subset contains the summary constraints starting at the median index and the summary constraints after the median index. A consistency check is performed on the left partition subset based on the basic constraint set. The consistency status of the left partition subset is determined. If the consistency status of the left partition subset is inconsistent, the summary constraint subset to be reduced is updated to the left partition subset and the median index is recalculated to complete the partition. If the consistency status of the left partition subset is consistent, the right partition subset is updated to the summary constraint subset to be reduced, and the left partition subset is merged into the basic constraint set. The median index is recalculated to complete the partition. If the number of elements in the summary constraint subset to be reduced is 1, the summary constraint subset to be reduced is written into the conflict proof subset and the corresponding summary constraint of the summary constraint subset to be reduced is removed from the basic constraint set before the consistency determination is performed. If there is a consistency status of inconsistency, the partition and update are continued until the conflict proof subset is formed. The minimumity test is performed on the conflict proof subset. The minimumity test uses the basic constraint set and the conflict proof subset as a joint decision set to perform a consistency determination and confirm that the joint consistency state is inconsistent. The minimumity test performs a deletion test on each summary constraint in the conflict proof subset. The deletion test removes the target summary constraint from the conflict proof subset and keeps the remaining summary constraints unchanged. The consistency determination is performed again and the consistency state is confirmed to be consistent. Write the conflict proof subset into the conflict hypergraph to form a proof hyperedge. The proof hyperedge records the summary constraint identifier sequence of the conflict proof subset. The proof hyperedge records the module key sequence corresponding to the summary constraint record within the conflict proof subset. The proof hyperedge records the constraint identifier sequence corresponding to the summary constraint record within the conflict proof subset and arranges them according to the summary constraint identifier sequence. Coverage determination is performed on the interval constraint subset corresponding to the minimum conflict time window. The coverage determination reads the constraint identifiers of the candidate constraints in the interval constraint subset and forms an interval constraint identifier set. The coverage determination performs an inclusion determination on the interval constraint identifier set and the constraint identifier sequence of the proof superedge record. If the inclusion determination is true, the determination is inconsistent, and the module key sequence of the proof superedge record is written into the conflict module set.

[0026] This invention uses the proof-learning QuickXPlain as the core solver in the consistency analysis of knowledge graph constraints in media terminals, and makes three improvements: First, it establishes an interval tree and recursively shrinks the node intervals based on the consistency judgment of interval constraint subsets to obtain the smallest conflict time window that cannot be further shrunk, thus compressing the time range for candidate constraint sets to participate in the solution. Second, it extends the consistency judgment to a conflict proof subset generation mechanism, where the conflict proof subset restores consistency after deleting any summary constraint. Third, it writes the conflict proof subset into the conflict hypergraph to form a proof hyperedge and records the module key and constraint identifier sequence. Fourth, it directly gives an inconsistency conclusion and reuses the obtained proof by judging the coverage of the proof hyperedge by the interval constraint subset. Fifth, it constructs a module mapping and aggregates candidate constraints using session identifier, copyright domain identifier, and frequency control cycle identifier as module keys to form a summary constraint set. It completes the segmentation and solution at the granularity of summary constraints, and then refines the conflict module set into atomic candidate constraints and updates the summary constraint set to obtain the smallest conflict constraint set. Compared with QuickXPlain, the improved version reduces the number of consistency judgment calls, reduces the solution scale, improves the conflict location accuracy, enhances the reusability of conflict interpretation records, and strengthens the determinism of management instruction generation.

[0027] In this embodiment, step seven specifically includes: Based on the module mapping, the constraint identifier sequence corresponding to the module key is located in the conflict module set record. The module mapping records the one-to-one correspondence between constraint identifiers and module keys. The conflict module set records the set of module keys. The module key consists of session identifier, copyright domain identifier, and frequency control period identifier. Based on the module key, the constraint identifier sequence corresponding to the module key is retrieved in the module mapping. Based on the constraint identifier sequence, the candidate constraint record matching the constraint identifier is retrieved in the candidate constraint set. The candidate constraint record is written into the atomic candidate constraint set one by one. The atomic candidate constraint set keeps the constraint identifier, associated node identifier set, associated edge identifier set, constraint judgment condition, and time window field of the candidate constraint record unchanged. The summary constraint set is updated based on the atomic candidate constraint set. The update rule aggregates atomic candidate constraints by module key. The module key is taken from the module key corresponding to the constraint identifier of the atomic candidate constraint in the module mapping. The summary constraint record is written to the module key and written to the aggregated constraint identifier sequence. The constraint identifier sequence is arranged in ascending order according to the start timestamp of the time window of the atomic candidate constraint. The start timestamp of the summary time window is taken as the minimum value of the start timestamp of the time window of the atomic candidate constraint corresponding to the constraint identifier sequence. The end timestamp of the summary time window is taken as the maximum value of the end timestamp of the time window of the atomic candidate constraint corresponding to the constraint identifier sequence. Based on the basic constraint set, minimum conflict time window, and summary constraint set, the proof-learning QuickXPlain is used to solve and map the minimum conflict constraint set. Within the minimum conflict time window, the proof-learning QuickXPlain performs consistency determination and set reduction on the summary constraint set, determines the set of summary constraints that are inconsistent and maps them to the constraint identifier sequence of the summary constraint record. The mapping process takes the atomic candidate constraints corresponding to the constraint identifier sequence of the summary constraint record as the candidate set. The candidate set continues to be reduced by the proof-learning QuickXPlain under the constraint conditions of the basic constraint set to obtain the minimum conflict constraint set. The minimum conflict constraint set is determined to be consistent after removing any atomic candidate constraint in the minimum conflict constraint set. Write the constraint identifier sequence, module key, minimum conflict time window start timestamp, and minimum conflict time window end timestamp of the minimum conflict constraint set into the conflict interpretation record. Create a record item in the conflict interpretation record according to the module key and write the constraint identifier sequence, time window start timestamp, and time window end timestamp into the record item. Generate management instructions based on the conflict interpretation record. The management instructions copy the module key, constraint identifier sequence, time window start timestamp, and time window end timestamp to form the instruction field according to the record item.

[0028] Example 1: To verify the feasibility of this invention in practice, it was applied to a typical media terminal data analysis and management system. This system needs to uniformly manage and constrain event data generated by a large number of terminals during operation. In actual operation, such systems typically need to simultaneously meet management requirements such as session integrity, copyright authorization scope, and frequency control strategies. However, existing engineering practices generally employ a full-scale verification method based on log tables and rule engines. In this traditional method, event records are written to a relational database or log indexing system after field cleaning. Management rules are configured one by one in the form of conditional expressions. The system iterates and verifies all event records and all rules at a fixed time granularity. Once a rule is found to be unsatisfactory, it can only return the rule number and the set of events that were hit, without further determining the precise time range of the conflict or distinguishing which rule combinations triggered the conflict. As the number of rules and the scale of events increase, the verification time increases significantly, and the conflict interpretation results rely on manual log analysis, making it difficult to directly generate executable management instructions.

[0029] In this application scenario, after introducing the method of this invention, the system first standardizes the fields and aligns the time of the event entries generated by the terminal, unifying the expression methods of terminal identifier, session identifier, event type, and timestamp, forming a set of event records with strict time semantics. Then, a media terminal knowledge graph is constructed based on the event records, representing terminals, sessions, and events as nodes, and session and event relationships as edges, explicitly writing time attributes into nodes and edges to structurally characterize the order of events and their associated paths. Based on this, the system no longer directly verifies all events and rules, but first generates a basic constraint set and a candidate constraint set according to the structure and management rules of the knowledge graph, configuring corresponding time windows for candidate constraints, thus naturally limiting rule verification to the time range where conflicts may occur. Furthermore, through a module mapping mechanism, candidate constraints are aggregated into module sets according to session identifier, copyright domain identifier, and frequency control cycle identifier, generating a summary constraint set, thereby compressing the constraint scale at the module level and laying the foundation for subsequent conflict analysis.

[0030] When the system detects inconsistencies in constraints, this invention does not employ the traditional strategy of "expanding the time range and rescanning." Instead, it constructs an interval tree based on candidate constraint time windows, recursively shrinks the time axis, and performs consistency checks on subsets of interval constraints to gradually locate the minimum conflict time window causing the inconsistency. Within the minimum conflict time window, a proof-learning QuickXPlain is introduced to structurally reduce the summary constraint set, generating a subset of conflict proofs that meets the minimumity requirement. This subset of conflict proofs is then written into the conflict hypergraph to form proof hyperedges. Through the coverage check of the proof hyperedges by the subsets of interval constraints, the system can stably identify sets of conflicting modules and further refine these modules into atomic candidate constraints. Finally, the minimum set of conflicting constraints is solved, and a conflict explanation record containing constraint identifiers, module keys, and conflict time windows, along with corresponding management instructions, is automatically generated.

[0031] In this embodiment, the processing performance of the traditional full-rule verification method and the method of the present invention are compared under the same event and rule scale conditions. The traditional method scans all event records at a fixed time granularity and performs rule condition judgments one by one, while the method of the present invention uses knowledge graph modeling, interval tree shrinkage, and proof-learning QuickXPlain to solve the problem jointly. The statistical results are shown in Table 1: Table 1. Performance Comparison of Conflict Localization and Constraint Solving

[0032] The comparison results show that traditional full rule verification methods require repeated scanning of event data over a large time range when conflicts occur. The number of rules involved in the verification is large, the conflict location time window is coarse, and the interpretation results are unstable. Managers need to manually determine the source of the conflict by combining logs. In contrast, this invention significantly reduces the conflict time window through the interval tree mechanism, reduces the scale of constraints involved in the solution through module mapping and summary constraint sets, and uses proof-learning QuickXPlain to generate a minimum conflict constraint set, making the conflict location results stable and directly usable for generating management instructions.

[0033] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for analyzing and managing media terminal data based on knowledge graphs, characterized in that, Includes the following steps: Step 1: Retrieve event records and perform field normalization and time alignment; Step 2: Construct a media terminal knowledge graph based on event records. The media terminal knowledge graph includes entity nodes, relationship edges, and time attributes. Step 3: Generate a basic constraint set and a candidate constraint set based on the pattern constraints and management rules of the media terminal knowledge graph, and associate the candidate constraint set with a time window; Step 4: Establish module mapping. Module keys include session identifier, copyright domain identifier, and frequency control period identifier. Aggregate candidate constraint sets according to module mapping to form module sets, and generate summary constraint sets corresponding to the module sets. Step 5: Build an interval tree, filter the basic constraint set and the candidate constraint set covering the time window interval to form an interval constraint subset, perform consistency judgment on the interval constraint subset and shrink the interval tree to obtain the minimum conflict time window; Step 6: Perform proof learning QuickXPlain on the summary constraint set within the minimum conflict time window. Consistency determination generates conflict proof subsets. Write the conflict proof subsets into the conflict hypergraph to form proof hyperedges. When the interval constraint subsets cover the proof hyperedges, determine inconsistency and obtain the conflict module set. Step 7: Based on the module mapping, refine the conflicting module set into atomic candidate constraints and update the summary constraint set. Solve to obtain the minimum conflicting constraint set, generate conflict interpretation records, and generate management instructions.

2. The method for media terminal data analysis and management based on knowledge graphs according to claim 1, characterized in that, Step one specifically involves: Obtain a set of event entries, which includes a terminal identifier field, a session identifier field, an event type field, a timestamp field, and a payload field. Perform field standardization on the event entries, which includes standardizing field names, field values, data types, and filling in missing values ​​to obtain standardized event entries. A baseline timeline is determined, which is a baseline timestamp sequence. Time alignment is performed on the timestamp fields of standardized event entries based on the baseline timeline. Time alignment includes time scale conversion and time offset correction to obtain aligned timestamps. The standardized event entries are sorted based on the aligned timestamps and aggregated by the terminal identifier field and the session identifier field to obtain the event records.

3. The method for media terminal data analysis and management based on knowledge graphs according to claim 1, characterized in that, Step two specifically involves: Based on the event logs, a set of terminal nodes, a set of session nodes, and a set of event nodes are generated. The set of terminal nodes is determined by the terminal identifier field, the set of session nodes is determined by the session identifier field, and the set of event nodes is determined by the event type field and the aligned timestamp field. The terminal nodes, session nodes, and event nodes are set with node identifiers and node type identifiers. A set of relational edges is generated based on event records. The set of relational edges includes session association edges from terminal nodes to session nodes and session event edges from session nodes to event nodes. Relationship edges are set with edge identifiers and edge type identifiers. The alignment timestamp field is written into the event node time attribute and the session event edge time attribute, the event type field is written into the event node attribute, and the terminal identifier field and session identifier field are written into the corresponding node attribute, forming a media terminal knowledge graph containing entity nodes, relationship edges and time attributes.

4. The method for media terminal data analysis and management based on knowledge graphs according to claim 1, characterized in that, Step three specifically involves: Based on the knowledge graph of media terminals, a pattern constraint rule set is established using node type identifiers, edge type identifiers, node attribute sets, and edge attribute sets. The pattern constraint rule set limits the connection relationship between node type identifiers and edge type identifiers, limits the value range and missing condition of node attribute sets and edge attribute sets, and limits the consistency condition between edge time attributes and node time attributes. A basic constraint set is generated based on the pattern constraint rule set, and the basic constraints of the basic constraint set set the constraint identifier, constraint object identifier, and constraint judgment condition. A candidate constraint generation rule set is established based on management rules. The candidate constraint generation rule set limits the rule fields and threshold fields associated with session identifier, copyright domain identifier, and frequency control cycle identifier. Based on the candidate constraint generation rule set, the associated node identifier and associated edge identifier are extracted from the media terminal knowledge graph to generate a candidate constraint set. The candidate constraint set sets the constraint identifier, associated node identifier set, associated edge identifier set, and constraint judgment conditions. Configure a time window for the candidate constraint set. The time window sets a start timestamp and an end timestamp. The start timestamp is the minimum value of the time attribute of the associated edge, and the end timestamp is the maximum value of the time attribute of the associated edge. Write the time window into the candidate constraints of the candidate constraint set.

5. The method for media terminal data analysis and management based on knowledge graphs according to claim 1, characterized in that, Step four specifically involves: Based on the candidate constraint set, a module key is generated for the candidate constraint. The module key includes a session identifier, a copyright domain identifier, and a frequency control cycle identifier. The session identifier is determined by locating the session node based on the associated node identifier set of the candidate constraint and reading the session identifier field. The copyright domain identifier is determined by assigning a value to the candidate constraint according to the management rules. The frequency control cycle identifier is determined by periodically mapping the start timestamp of the time window of the candidate constraint according to the period division rules set by the management rules. A module mapping is established, and the module mapping records the constraint identifier and the module key. Candidate constraint sets are grouped by module key to form module sets, and the module sets record the module key and constraint identifier sequence. A summary constraint set is generated based on the module set. The summary constraint record of the summary constraint set includes the module key, constraint identifier sequence, and summary time window. The start timestamp of the summary time window is the minimum value of the start timestamp of the time window corresponding to the constraint identifier sequence, and the end timestamp of the summary time window is the maximum value of the end timestamp of the time window corresponding to the constraint identifier sequence.

6. The method for media terminal data analysis and management based on knowledge graphs according to claim 1, characterized in that, Step five specifically involves: The starting timestamp of the time axis range is determined based on the minimum starting timestamp of the time window of the candidate constraints in the candidate constraint set, and the ending timestamp of the time axis range is determined based on the maximum ending timestamp of the time window of the candidate constraints in the candidate constraint set. An interval tree is built on the time axis range, the interval of the root node of the interval tree is the time axis range, and the interval tree node records the starting timestamp and ending timestamp of the node interval. For each interval tree node, an interval constraint subset is constructed. The interval constraint subset contains the basic constraint set. The interval constraint subset contains candidate constraints within the candidate constraint set that satisfy the condition that the start timestamp of the time window is not greater than the start timestamp of the node interval and the end timestamp of the time window is not less than the end timestamp of the node interval. A consistency determination is performed on the interval constraint subset and the consistency status is written to the corresponding interval tree node. For interval tree nodes with inconsistent consistency, interval shrinkage is performed. Interval shrinkage uses the median of the start and end timestamps of the interval as the split timestamp to generate left and right sub-intervals. Left and right child nodes are constructed separately, and consistency is determined separately. When both the left and right child nodes are consistent, the node interval is determined as the minimum conflict time window. When both the left and right child nodes are inconsistent, the left child node is selected as the node to be shrunk. When both the left and right child nodes are consistent, the right child node is selected as the node to be shrunk. When both the left and right child nodes are inconsistent, the child node with the smaller interval length is selected as the node to be shrunk. When the interval length of the node to be shrunk is less than or equal to a preset time granularity threshold, the node interval to be shrunk is determined as the minimum conflict time window.

7. The method for media terminal data analysis and management based on knowledge graphs according to claim 1, characterized in that, Step six specifically involves: Based on the start and end timestamps of the minimum conflict time window, the summary constraints with a start timestamp of the summary time window not greater than the start timestamp of the minimum conflict time window and an end timestamp of the summary time window not less than the end timestamp of the minimum conflict time window are selected from the summary constraint set to form a subset of time window summary constraints. A consistency determination is performed on the basic constraint set and the time window summary constraint subset. The consistency determination calculates and summarizes the constraint determination conditions of the basic constraint set and the constraint determination conditions of the time window summary constraint subset one by one. When all the summary results are satisfied, the consistency state is determined to be consistent. When there are any unsatisfied summary results, the consistency state is determined to be inconsistent. Under the condition of inconsistent consistency, a proof-learning QuickXPlain solution is performed on the time window summary constraint subset. The proof-learning QuickXPlain sets the summary constraint subset to be reduced as the time window summary constraint subset and divides the summary constraint subset to be reduced into a left partition subset and a right partition subset according to the median index of the summary constraint identifier sequence. A consistency determination is performed on the left partition subset based on the basic constraint set. If the consistency status of the left partition subset is inconsistent, the summary constraint subset to be reduced is updated to the left partition subset and the division continues. If the consistency status of the left partition subset is consistent, the right partition subset is used as the summary constraint subset to be reduced and the left partition subset is added to the basic constraint set before the division continues. If the number of elements in the summary constraint subset to be reduced is equal to 1, the summary constraint subset to be reduced is added to the conflict proof subset. Perform a minimality test on the conflict proof subset. The minimality test satisfies that the joint consistency state of the basic constraint set and the conflict proof subset is inconsistent, and the joint consistency state of the basic constraint set and the conflict proof subset after deleting any summary constraint is consistent. Write the conflict proof subset into the conflict hypergraph to form a proof hyperedge. The proof hyperedge records the summary constraint identifier sequence of the conflict proof subset. The proof hyperedge records the module key corresponding to the conflict proof subset. The proof hyperedge records the constraint identifier sequence corresponding to the conflict proof subset. The coverage determination is performed based on the interval constraint subset corresponding to the minimum conflict time window. If the coverage determination satisfies the condition that the interval constraint subset contains the constraint identifier sequence of the proof superedge record, the determination is inconsistent, and the module key of the proof superedge record is written into the conflict module set.

8. The method for media terminal data analysis and management based on knowledge graphs according to claim 1, characterized in that, Step seven specifically involves: Based on the module mapping, locate the constraint identifier sequence corresponding to the module key recorded in the conflict module set. Based on the constraint identifier sequence, locate the candidate constraint in the candidate constraint set. Write the candidate constraint into the atomic candidate constraint set. The atomic candidate constraint retains the constraint identifier, associated node identifier set, associated edge identifier set, constraint decision condition, and time window. The summary constraint set is updated based on the atomic candidate constraint set. The summary constraint record module key and constraint identifier sequence are used. The start timestamp of the summary time window is the minimum value of the start timestamp of the time window corresponding to the constraint identifier sequence. The end timestamp of the summary time window is the maximum value of the end timestamp of the time window corresponding to the constraint identifier sequence. Based on the basic constraint set, minimum conflict time window, and summary constraint set, the proof-learning QuickXPlain is used to solve and map the minimum conflict constraint set. The minimum conflict constraint set satisfies the condition that removing any atomic candidate constraint in the minimum conflict constraint set determines the consistency state as consistent. Write the constraint identifier sequence, module key, minimum conflict time window start timestamp, and minimum conflict time window end timestamp of the minimum conflict constraint set into the conflict interpretation record. Generate management instructions based on the conflict interpretation record. The management instructions include the module key, constraint identifier sequence, minimum conflict time window start timestamp, and minimum conflict time window end timestamp.

Images