A multi-source heterogeneous data real-time governance system based on data mirroring

Abstract

The application discloses a kind of multi-source heterogeneous data real-time management system based on data mirror, comprising: multi-source mirror compiling system, for obtaining multi-source heterogeneous data and generating mirror object set;Commitment characterization system, for generating mirror commitment vector set;Mirror commitment chain arrangement system, for the chain connection of mirror commitment vector set;Double-track consistent checking system, for performing endogenous consistency and exogenous constraint checking;Counterfactual back-throw decision system, for back-throw checking and ruling optimal management action;Governance closed loop recompilation system, for executing management, outputting evidence and recompiling rule.The application adopts the chain management method based on data mirror, realizes multi-source heterogeneous data real-time checking and closed loop management, has the advantages of strong continuity, high traceability and high management accuracy.

Description

Technical Field

[0001] This invention relates to the field of multi-source heterogeneous data processing technology, and in particular to a real-time governance system for multi-source heterogeneous data based on data mirroring. Background Technology

[0002] Existing multi-source heterogeneous data governance technologies typically revolve around data access, format unification, field mapping, rule validation, and result output. They primarily clean, compare, and repair data from relational databases, document databases, message queues, API receipts, log files, and batch-imported files through pre-set templates, rule bases, or quality inspection processes. These solutions can, to some extent, address issues such as inconsistent data formats, inconsistent field naming, and localized quality anomalies from multiple data sources. They have been widely applied in enterprise data platforms, data warehouses, master data management, and data quality governance scenarios, forming a conventional governance path based on rule matching, template constraints, and task closure.

[0003] However, most existing technologies treat the data to be governed as discrete records, data tables, or file objects, lacking a chain-like governance carrier for cross-source, cross-version, and cross-time relationships. This makes it difficult to simultaneously express primary key inheritance, version succession, temporal sequence, source propagation, and field dependencies between objects, resulting in insufficient identification of link breaks, source mismatches, coverage gaps, and version conflicts. Furthermore, existing technologies typically execute governance actions directly after a rule is hit, lacking a pre-verification mechanism based on counterfactual feedback, and also lacking a closed-loop update mechanism to recompile commitment representation rules, source commitment weights, coverage commitment thresholds, and dual-track consistency adjudication rules based on the governance results. Therefore, they struggle to meet the requirements of continuity, traceability, and adaptability in real-time governance scenarios for multi-source heterogeneous data.

[0004] Therefore, how to provide a real-time governance system for multi-source heterogeneous data based on data mirroring is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose a real-time governance system for multi-source heterogeneous data based on data mirroring. This invention adopts a chain-like governance method based on data mirroring to achieve real-time verification and closed-loop governance of multi-source heterogeneous data, which has the advantages of strong continuity, high traceability and high governance accuracy.

[0006] A real-time governance system for multi-source heterogeneous data based on data mirroring according to an embodiment of the present invention includes:

[0007] The multi-source image compilation module is used to acquire heterogeneous data from multiple sources, perform object splitting, time warping, source marking and version binding, and generate a collection of image objects;

[0008] The commitment representation generation module is used to extract structural integrity, temporal validity, cross-source coverage relationship, source inheritance relationship and historical stability relationship based on the mirror object set, and generate a set of mirror commitment vectors.

[0009] The mirror commitment chain orchestration module is used to perform chain concatenation on the mirror commitment vector set based on the primary key inheritance relationship, version inheritance relationship, time sequence relationship, source transmission relationship and field dependency relationship in the mirror object set, and generate a mirror commitment chain set.

[0010] The dual-track consistency verification module is used to perform mirror endogenous consistency verification on the mirror commitment chain set, and to perform exogenous constraint verification in combination with governance rules and standard templates to generate a set of governance candidate actions;

[0011] The Counterfactual Reversal Decision Module is used to apply the set of governance candidate actions to the set of mirror commitment chains, generate a set of counterfactual mirror commitment chains and perform reversal verification, execute a dual-track consistency decision based on the reversal verification results, and generate the optimal set of governance actions.

[0012] The governance closed-loop recompilation module is used to apply the optimal set of governance actions to the set of mirror objects, generate a set of governance results and an explanatory evidence package, and perform commitment recompilation on the generation rules, source commitment weights, coverage commitment thresholds and dual-track consistency adjudication rules of the set of mirror commitment vectors, thus forming a real-time governance closed loop.

[0013] Optionally, the multi-source image compilation module includes:

[0014] Acquire multi-source heterogeneous data and parse the data carrier identifier, business field content, original record primary key, original time information and source system information to generate a multi-source raw data set;

[0015] Perform format unification and field standardization on multi-source raw data sets to generate a unified field data set;

[0016] Object splitting is performed based on a unified field data set to generate a candidate object set;

[0017] Perform time warping on the candidate object set to generate a time-warped object set;

[0018] Perform source tagging on the time-warped object set to generate a source tag object set;

[0019] Perform version binding on the source tag object collection, generating a version binding object collection;

[0020] The collection of version-bound objects is compiled into mirror objects to obtain mirror objects. These mirror objects are then aggregated according to the business primary key index, source index, and version index to generate a collection of mirror objects.

[0021] Optionally, the commitment representation generation module includes:

[0022] Read the object identifier, business primary key, unified field data, object time series, source tag and version binding results from the mirror object collection, and generate the structural integrity calculation input set;

[0023] Based on the input set of structural integrity calculation, a regularization calculation is performed to generate a set of structural integrity results, which is then written back to the corresponding mirror object.

[0024] Read the object time series, version binding results, and source tags from the mirror object collection, and generate a time validity result set;

[0025] Based on the business primary key, unified field data and source marker in the mirror object collection, field coverage alignment is performed on cross-source objects under the same business primary key to generate a cross-source coverage relationship result set.

[0026] Based on the source tags and version binding results in the mirror object collection, generate a source inheritance relationship result set;

[0027] Based on the version binding results, object time series, and unified field data in the mirror object collection, a set of historical stable relationship results is generated.

[0028] The result sets of structural integrity, time validity, cross-source coverage, source inheritance, and historical stability are aggregated and encapsulated according to the mirror object identifier to generate a mirror commitment vector set.

[0029] Optionally, the mirror commitment chain orchestration module includes:

[0030] Read the set of mirror objects and the set of mirror commitment vectors, merge the mirror objects and mirror commitment vectors under the same governance object according to the business primary key, and generate the chain orchestration input set;

[0031] Based on the chain orchestration input set, perform primary key inheritance judgment, version inheritance judgment, time sequence judgment, source transmission judgment and field dependency judgment on mirror objects under the same business primary key to generate a set of mirror object pairs;

[0032] Write relation type tags and relation direction tags into the mirror object pairing set according to primary key inheritance relationship, version inheritance relationship, time sequence relationship, source transit relationship and field dependency relationship respectively, and generate a relation tag pairing set;

[0033] Based on the relation tag pairing set, the mirror commitment vectors corresponding to each object pairing are sorted in front and back order to generate a commitment pairing set;

[0034] Chain the commitment pair set to generate a candidate mirror commitment chain set;

[0035] Based on the candidate mirror commitment chain set, perform in-chain order correction and generate an order correction result set;

[0036] The candidate mirror commitment chain set is rearranged and encapsulated based on the order correction result set to obtain the mirror commitment chain set.

[0037] Optionally, the dual-track consistency verification module includes:

[0038] Read the mirror commitment chain set and mirror commitment vector set, and read the governance rules and standard templates. Perform alignment according to chain identifier, in-chain position and relationship record to generate the mirror endogenous consistency check input set and exogenous constraint check input set;

[0039] Based on the input set of the mirror endogenous consistency verification, the mirror endogenous consistency verification is performed on each mirror commitment chain to generate a set of mirror endogenous consistency results, including chain identifier, chain position, endogenous consistency state, break position marker and break type marker.

[0040] Based on the input set of exogenous constraint verification, exogenous constraint verification is performed on the mirror commitment chain set to generate an exogenous constraint result set including chain identifier, chain position, exogenous constraint status, constraint violation position marker and constraint violation type marker;

[0041] A dual-track intersection verification is performed based on the mirrored endogenous consistency result set and the exogenous constraint result set to generate a dual-track intersection result set.

[0042] Based on the set of results from the convergence of the two tracks, a governance candidate action entry point is established, and conflict actions are mapped according to conflict object, conflict location, and conflict relationship to obtain a set of governance candidate action items.

[0043] The set of governance candidate action items is encapsulated according to chain identifier, position within the chain, conflicting object, action type, triggering basis, and expected repair relationship to generate a set of governance candidate actions.

[0044] Optionally, the conflict action mapping includes generating a field repair action when the conflict relationship is a field specification conflict or a field dependency conflict; generating a source switching action when the conflict relationship is a source transition conflict, or when there is a source switching state, a source rollback state, or a source mismatch state in the result set of the source inheritance relationship; generating a coverage completion action when the conflict relationship is a coverage closure conflict, or when there is a field coverage gap state in the result set of the cross-source coverage relationship; generating a version rollback action when the conflict relationship is a version inheritance conflict, or when there is a version inheritance break, update coverage anomaly, or a previous version reference mismatch in the version binding result; and generating a conflict isolation action when the same conflict object hits two or more types of conflict relationships at the same time, or when there is a mirror endogenous consistency break and an exogenous constraint violation at the same position in the same chain.

[0045] Optionally, the counterfactual reversal adjudication module includes:

[0046] Read the set of governance candidate actions and the set of mirror commitment chains, align them by chain identifier, position within the chain, conflicting object, action type, triggering basis, and expected repair relationship, and generate a return input set;

[0047] Based on the feedback input set, each governance candidate action item is applied to the corresponding mirror commitment chain set to generate a counterfactual mirror commitment chain set.

[0048] Based on the counterfactual mirror commitment chain set, a resubmission verification is performed, generating a resubmission verification result set;

[0049] Based on the set of feedback verification results, perform dual-track consistent adjudication input reconstruction to generate a dual-track consistent adjudication input set;

[0050] Based on the input set of the dual-track consistent decision, a dual-track consistent decision is executed to generate the optimal set of governance actions.

[0051] Optionally, the dual-track consistent adjudication compares the primary key inheritance closure state, version inheritance closure state, time sequence closure state, source transmission closure state, field dependency closure state, field specification closure state, and coverage closure state for multiple governance candidate action entries corresponding to the same conflict object. It prioritizes the governance candidate action entries with fewer unclosed relationship types. When the number of unclosed relationship types is the same, it prioritizes the governance candidate action entries with fewer unclosed position markers. When the number of unclosed position markers is the same, it prioritizes the governance candidate action entries with more conflict relationships corresponding to the expected repair relationship and the dual-track intersection result set. When there is still no closure action for the same conflict object after the aforementioned action mapping, or when it still fails the resubmission verification after a preset number of counterfactual resubmissions, a manual review submission action is generated. The optimal governance action set is obtained by encapsulating the chain identifier, chain position, conflict object, action type, target, adjudication basis, closure state, and expected repair relationship.

[0052] Optionally, the governance closed-loop recompilation module includes:

[0053] Read the optimal governance action set, mirror object set, and mirror commitment vector set, align them according to chain identifier, in-chain position, conflict object, action type, target, adjudication basis, closure state, and expected repair relationship, generate a governance execution input set, and generate a review record input set for the manual review submission actions in the optimal governance action set;

[0054] Based on the governance execution input set, the field repair action, source switching action, overwrite and completion action, version rollback action and conflict isolation action in the optimal governance action set are applied to the corresponding image object set to generate the post-governance image object set.

[0055] Based on the post-governance mirror object set, the results of structural integrity, time validity, cross-source coverage, source inheritance, and historical stability are updated and aggregated on the affected mirror objects to generate a post-governance commitment result set;

[0056] Based on the governance execution input set and the post-governance commitment result set, a governance result set is generated; based on the review record input set, a review record set is generated.

[0057] Based on the set of governance results, the set of review records, the basis for adjudication, and the set of post-governance commitment results, an explanatory evidence package is generated according to the conflict object.

[0058] Based on the governance result set, the review record set, and the interpretation evidence package, the commitment recompilation is performed on the generation rules of the mirror commitment vector set, the source commitment weight, the coverage commitment threshold, and the dual-track consistency adjudication rules to generate a recompiled result set;

[0059] The mirror commitment vector set is updated and encapsulated based on the recompiled result set, and the updated mirror commitment vector set is written back to the commitment representation generation module. The recompiled result set is written back to the dual-track consistency verification module and the counterfactual resubmission adjudication module, forming a real-time governance closed loop.

[0060] The beneficial effects of this invention are:

[0061] This invention, based on a set of mirrored objects, uniformly represents structural integrity, temporal validity, cross-source coverage relationships, source inheritance relationships, and historical stability relationships as a set of mirrored commitment vectors. Furthermore, it forms a set of mirrored commitment chains based on primary key inheritance relationships, version succession relationships, temporal sequence relationships, source transitive relationships, and field dependencies. This allows multi-source heterogeneous data to no longer be passively governed as fragmented records, entries, or messages, but rather participate in subsequent verification, adjudication, and updates as a chain-like, continuous governance carrier. This enables the simultaneous identification of intra-chain relationship breaks, cross-source coverage gaps, source mismatches, and version conflicts during the governance process, improving the completeness of identification and the targeted nature of governance for complex heterogeneous data association problems, and enhancing the data continuity expression capability and traceability of the governance process in real-time governance scenarios.

[0062] This invention further utilizes a dual-track consistency verification module to perform endogenous consistency verification and exogenous constraint verification on the mirror commitment chain set. Based on the dual-track convergence results, it establishes entry points for governance candidate actions. Then, a counterfactual re-adjudication module performs re-adjudication and dual-track consistency adjudication on the counterfactual mirror commitment chain set after the governance candidate action set has been applied. This completes closure verification and optimal selection before the actual execution of governance actions, reducing the risks of erroneous repairs, incorrect switching, and conflict propagation caused by direct governance. Simultaneously, a governance closed-loop recompilation module performs commitment recompilation based on the governance result set, review record set, and interpretation evidence package, adjusting the generation rules, source commitment weights, coverage commitment thresholds, and dual-track consistency adjudication rules of the mirror commitment vector set. This allows the system to continuously adjust rules and weight configurations based on governance results, forming an adaptive closed-loop mechanism for real-time governance of multi-source heterogeneous data. Therefore, it achieves superior overall performance in terms of governance accuracy, link consistency, result interpretability, and subsequent governance stability. Attached Figure Description

[0063] 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:

[0064] Figure 1 This is a flowchart of a real-time governance system for multi-source heterogeneous data based on data mirroring proposed in this invention;

[0065] Figure 2This is a flowchart illustrating the generation and arrangement of a set of mirror commitment vectors to form a set of mirror commitment chains in a real-time governance system for multi-source heterogeneous data based on data mirroring, as proposed in this invention.

[0066] Figure 3 This is a flowchart of a real-time governance closed loop for a multi-source heterogeneous data real-time governance system based on data mirroring proposed in this invention. Detailed Implementation

[0067] 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.

[0068] refer to Figures 1-3 A real-time governance system for multi-source heterogeneous data based on data mirroring includes:

[0069] The multi-source image compilation module is used to acquire heterogeneous data from multiple sources, perform object splitting, time warping, source marking and version binding, and generate a collection of image objects;

[0070] The commitment representation generation module is used to extract structural integrity, temporal validity, cross-source coverage relationship, source inheritance relationship and historical stability relationship based on the mirror object set, and generate a set of mirror commitment vectors.

[0071] The mirror commitment chain orchestration module is used to perform chain concatenation on the mirror commitment vector set based on the primary key inheritance relationship, version inheritance relationship, time sequence relationship, source transmission relationship and field dependency relationship in the mirror object set, and generate a mirror commitment chain set.

[0072] The dual-track consistency verification module is used to perform mirror endogenous consistency verification on the mirror commitment chain set, and to perform exogenous constraint verification in combination with governance rules and standard templates to generate a set of governance candidate actions;

[0073] The Counterfactual Reversal Decision Module is used to apply the set of governance candidate actions to the set of mirror commitment chains, generate a set of counterfactual mirror commitment chains and perform reversal verification, execute a dual-track consistency decision based on the reversal verification results, and generate the optimal set of governance actions.

[0074] The governance closed-loop recompilation module is used to apply the optimal set of governance actions to the set of mirror objects, generate a set of governance results and an explanatory evidence package, and perform commitment recompilation on the generation rules, source commitment weights, coverage commitment thresholds and dual-track consistency adjudication rules of the set of mirror commitment vectors, thus forming a real-time governance closed loop.

[0075] In this embodiment, the multi-source image compilation module includes:

[0076] Acquire multi-source heterogeneous data and parse the data carrier identifier, business field content, original record primary key, original time information and source system information to generate a multi-source raw data set;

[0077] The heterogeneous data sources include relational databases, document databases, message queues, API receipts, log files, and batch import files;

[0078] Perform format unification and field standardization on multi-source raw data sets to generate a unified field data set;

[0079] Object splitting is performed based on a unified field data set to generate a candidate object set;

[0080] The object splitting process identifies object affiliation based on business primary key, composite identifier, record boundary, message boundary, and event boundary. Fields belonging to the same business object are aggregated and encapsulated, and fields that cannot be directly affixed are merged based on the association of adjacent fields, time adjacency, and source context.

[0081] Perform time warping on the candidate object set to generate a time-warped object set;

[0082] Perform source tagging on the time-warped object set to generate a source tag object set;

[0083] The source tag reads the source system identifier, source interface identifier, source carrier type, acquisition channel identifier, and import batch identifier corresponding to each candidate object, and writes them into the source tag at the object level and field level respectively;

[0084] Perform version binding on the source tag object collection, generating a version binding object collection;

[0085] Version binding extracts historical versions, incremental change records, snapshot records, and write-back records under the same business primary key. It establishes the version succession order based on the time sequence, update coverage relationship, and source continuity relationship, and writes the version number, previous version reference, and current version status into the corresponding object.

[0086] Compile the image objects into a collection of version-bound objects to obtain image objects. Then, aggregate them according to the business primary key index, source index, and version index to generate a collection of image objects.

[0087] The compilation of mirror objects encapsulates the version-bound object collection according to object identifier, business primary key, unified field data, object time series, source tag and version binding result.

[0088] In this embodiment, the commitment characterization generation module includes:

[0089] Read the object identifier, business primary key, unified field data, object time series, source tag and version binding results from the mirror object collection, and generate the structural integrity calculation input set;

[0090] The input set for structural integrity calculation extracts field existence status, field missing status, field type matching status, and field position continuity status at the object level and field level, respectively.

[0091] Based on the input set of structural integrity calculation, a regularization calculation is performed to generate a set of structural integrity results, which is then written back to the corresponding mirror object.

[0092] The regularization calculation performs alignment on the field existence status, field missing status, field type matching status, and field position continuity status according to the field regularization name and field position index. It performs complete field, damaged field, and missing field determination on each field, performs object-level aggregation according to the mirror object identifier, and generates a structural integrity result set including object identifier, number of complete fields, number of damaged fields, number of missing fields, and object integrity status.

[0093] Read the object time series, version binding results, and source tags from the mirror object collection, and generate a time validity result set;

[0094] The time validity result set extracts the time start status, time end status, time sequence consistency status, version time connection status, and source time continuity status.

[0095] Based on the business primary key, unified field data and source marker in the mirror object collection, field coverage alignment is performed on cross-source objects under the same business primary key to generate a cross-source coverage relationship result set.

[0096] Among them, field coverage alignment extracts the field coverage range, field coverage overlap status, field coverage gap status, and field coverage complementary status;

[0097] Based on the source tags and version binding results in the mirror object collection, generate a source inheritance relationship result set;

[0098] The source inheritance relationship result set extracts the source continuation status, source switching status, source rollback status, and source mismatch status between the current mirror object and the previous version;

[0099] Based on the version binding results, object time series, and unified field data in the mirror object collection, a set of historical stable relationship results is generated.

[0100] The historical stable relationship result set extracts the field fluctuation status, time progression status, coverage change status and source continuation status of each mirror object under the same business primary key in consecutive versions;

[0101] The result sets of structural integrity, time validity, cross-source coverage, source inheritance, and historical stability are aggregated and encapsulated according to the mirror object identifier to generate a mirror commitment vector set.

[0102] In this embodiment, the mirror commitment chain orchestration module includes:

[0103] Read the set of mirror objects and the set of mirror commitment vectors, merge the mirror objects and mirror commitment vectors under the same governance object according to the business primary key, and generate the chain orchestration input set;

[0104] Based on the chain orchestration input set, perform primary key inheritance judgment, version inheritance judgment, time sequence judgment, source transmission judgment and field dependency judgment on mirror objects under the same business primary key to generate a set of mirror object pairs;

[0105] The primary key inheritance determination is based on the consistency of the business primary key and the difference of the object identifier to determine the inheritance candidate object pair. The version inheritance determination is based on the reference of the previous version, the current version status, and the update coverage relationship to determine the version inheritance object pair. The time sequence determination is based on the record generation time, update time, and synchronization time in the object time sequence to determine the time sequence object pair. The source transmission determination is based on the source system identifier, source interface identifier, source carrier type, and source continuation status to determine the source transmission object pair. The field dependency determination is based on the field regularization name, field position index, and field coverage complementarity status in the unified field data to determine the field dependency object pair.

[0106] Write relation type tags and relation direction tags into the mirror object pairing set according to primary key inheritance relationship, version inheritance relationship, time sequence relationship, source transit relationship and field dependency relationship respectively, and generate a relation tag pairing set;

[0107] Based on the relation tag pairing set, the mirror commitment vectors corresponding to each object pairing are sorted in front and back order to generate a commitment pairing set;

[0108] The preceding and following order arrangement involves aligning the source inheritance relationship results, historical stability relationship results, and time validity results in the preceding mirror commitment vector with the structural integrity results, cross-source coverage relationship results, and source inheritance results in the following mirror commitment vector;

[0109] Chain the commitment pair set to generate a candidate mirror commitment chain set;

[0110] The chained concatenation connects the first and last entries of commitment alignment with continuous primary key inheritance, continuous version inheritance, continuous time sequence, and continuous source transmission, and writes the field dependency relationship into the corresponding connection position.

[0111] Based on the candidate mirror commitment chain set, perform in-chain order correction and generate an order correction result set;

[0112] Among them, the in-chain order correction is performed on candidate mirror commitment chains with multiple version acceptance paths, parallel source transmission paths or cross field dependency paths, in the order of version acceptance priority, time priority, source transmission priority, and field dependency supplementary constraints.

[0113] The candidate mirror commitment chain set is rearranged and encapsulated based on the order correction result set to obtain the mirror commitment chain set.

[0114] Each mirror commitment chain in the mirror commitment chain set includes a chain identifier, a business primary key, the order of mirror objects within the chain, the order of mirror commitment vectors within the chain, a primary key inheritance relationship record, a version inheritance relationship record, a time sequence relationship record, a source transmission relationship record, and a field dependency relationship record.

[0115] In this embodiment, the dual-track consistency verification module includes:

[0116] Read the mirror commitment chain set and mirror commitment vector set, and read the governance rules and standard templates. Perform alignment according to chain identifier, in-chain position and relationship record to generate the mirror endogenous consistency check input set and exogenous constraint check input set;

[0117] The governance rules are provided by a pre-defined governance rule library, including field specification rules, time continuity rules, version inheritance rules, source transmission rules, and coverage closure rules. The standard templates are provided by a pre-defined standard template library, including standard field sets, standard time relationships, standard version relationships, standard source relationships, and standard coverage relationships.

[0118] The mirror endogenous consistency verification input set is generated by binding the mirror commitment chain set and the mirror commitment vector set to the preceding and following chain positions according to the chain identifier and the position within the chain. The exogenous constraint verification input set is generated by aligning the governance rules and standard templates with the mirror objects and mirror commitment vectors corresponding to each chain position according to the chain identifier and the position within the chain.

[0119] Based on the input set of the mirror endogenous consistency verification, the mirror endogenous consistency verification is performed on each mirror commitment chain to generate a set of mirror endogenous consistency results, including chain identifier, chain position, endogenous consistency state, break position marker and break type marker.

[0120] The mirror-based endogenous consistency check is achieved by sequentially checking the primary key inheritance continuity, version inheritance continuity, temporal continuity, source propagation continuity, and field dependency continuity between adjacent chain bits, and combining the structural integrity results, time validity results, cross-source coverage relationship results, source inheritance relationship results, and historical stability relationship results of the corresponding chain bits.

[0121] Based on the input set of exogenous constraint verification, exogenous constraint verification is performed on the mirror commitment chain set to generate an exogenous constraint result set including chain identifier, chain position, exogenous constraint status, constraint violation position marker and constraint violation type marker;

[0122] The exogenous constraint verification will instantiate and match the governance rules and standard templates according to the business primary key, field regularization name, source tag, version binding result and object time series, and check the field specification satisfaction status, time validity satisfaction status, cross-source coverage satisfaction status, source inheritance satisfaction status and historical stable satisfaction status in turn.

[0123] A dual-track intersection verification is performed based on the mirrored endogenous consistency result set and the exogenous constraint result set to generate a dual-track intersection result set.

[0124] The dual-track convergence verification aligns the endogenous consistency state and exogenous constraint state corresponding to the same chain identifier, the same position within the same chain and the same relationship record. It uses the fracture position marker and fracture type marker in the mirrored endogenous consistency result set and the constraint violation position marker and constraint violation type marker in the exogenous constraint result set to jointly determine the conflict object, conflict position and conflict relationship.

[0125] Based on the set of results from the convergence of the two tracks, a governance candidate action entry point is established, and conflict actions are mapped according to conflict object, conflict location, and conflict relationship to obtain a set of governance candidate action items.

[0126] The governance candidate action entry points are established based on the items that exist in the chain, such as primary key inheritance break, version succession break, time sequence break, source transmission break, field dependency break, field specification violation, coverage gap, or source mismatch.

[0127] The set of governance candidate action items is encapsulated according to chain identifier, position within the chain, conflicting object, action type, triggering basis, and expected repair relationship to generate a set of governance candidate actions.

[0128] In this embodiment, the conflict action mapping includes generating a field repair action when the conflict relationship is a field specification conflict or a field dependency conflict; generating a source switching action when the conflict relationship is a source transition conflict, or when there is a source switching state, a source rollback state, or a source mismatch state in the result set of the source inheritance relationship; generating a coverage completion action when the conflict relationship is a coverage closure conflict, or when there is a field coverage gap state in the result set of the cross-source coverage relationship; generating a version rollback action when the conflict relationship is a version inheritance conflict, or when there is a version inheritance break, update coverage anomaly, or a previous version reference mismatch in the version binding result; and generating a conflict isolation action when the same conflict object hits two or more types of conflict relationships at the same time, or when there is a mirror endogenous consistency break and an exogenous constraint violation at the same position in the same chain.

[0129] In this embodiment, the counterfactual reversal adjudication module includes:

[0130] Read the set of governance candidate actions and the set of mirror commitment chains, align them by chain identifier, position within the chain, conflicting object, action type, triggering basis, and expected repair relationship, and generate a return input set;

[0131] The feedback input set binds each governance candidate action entry in the governance candidate action set to its corresponding mirror commitment chain, the order of mirror objects within the chain, the order of mirror commitment vectors within the chain, the primary key inheritance relationship record, the version inheritance relationship record, the time sequence relationship record, the source transmission relationship record, and the field dependency relationship record.

[0132] Based on the feedback input set, each governance candidate action item is applied to the corresponding mirror commitment chain set to generate a counterfactual mirror commitment chain set.

[0133] Based on the counterfactual mirror commitment chain set, a resubmission verification is performed, generating a resubmission verification result set;

[0134] The resubmission verification checks each counterfactual mirror commitment chain sequentially on the primary key inheritance closure status, version inheritance closure status, time sequence closure status, source transmission closure status, field dependency closure status, field specification closure status, and coverage closure status, and records the corresponding closure position marker, unclosed position marker, and unclosed relationship type marker.

[0135] Based on the set of feedback verification results, perform dual-track consistent adjudication input reconstruction to generate a dual-track consistent adjudication input set;

[0136] The dual-track consistent adjudication input reconstruction will align the feedback results of each chain position in the counterfactual mirror commitment chain set with the mirror endogenous consistent result set, exogenous constraint result set and dual-track intersection result set according to the chain identifier and the position within the chain.

[0137] Based on the input set of the dual-track consistent decision, a dual-track consistent decision is executed to generate the optimal set of governance actions.

[0138] In this implementation, the dual-track consistent adjudication compares the primary key inheritance closure state, version inheritance closure state, time sequence closure state, source transmission closure state, field dependency closure state, field specification closure state, and coverage closure state of multiple governance candidate action entries corresponding to the same conflict object. It prioritizes the governance candidate action entries with fewer unclosed relationship types. When the number of unclosed relationship types is the same, it prioritizes the governance candidate action entries with fewer unclosed position markers. When the number of unclosed position markers is the same, it prioritizes the governance candidate action entries with more conflict relationships corresponding to the expected repair relationship and the dual-track intersection result set. When there is still no closure action for the same conflict object after the aforementioned action mapping, or when it still fails the resubmission verification after a preset number of counterfactual resubmissions, a manual review submission action is generated. The optimal governance action set is obtained by encapsulating the chain identifier, chain position, conflict object, action type, target, adjudication basis, closure state, and expected repair relationship.

[0139] In this embodiment, the governance closed-loop recompilation module includes:

[0140] Read the optimal governance action set, mirror object set, and mirror commitment vector set, align them according to chain identifier, in-chain position, conflict object, action type, target, adjudication basis, closure state, and expected repair relationship, generate a governance execution input set, and generate a review record input set for the manual review submission actions in the optimal governance action set;

[0141] Based on the governance execution input set, the field repair action, source switching action, overwrite and completion action, version rollback action and conflict isolation action in the optimal governance action set are applied to the corresponding image object set to generate the post-governance image object set.

[0142] Based on the post-governance mirror object set, the results of structural integrity, time validity, cross-source coverage, source inheritance, and historical stability are updated and aggregated on the affected mirror objects to generate a post-governance commitment result set;

[0143] Based on the governance execution input set and the post-governance commitment result set, a governance result set is generated; based on the review record input set, a review record set is generated.

[0144] The governance result set records the chain identifier, chain position, conflict object, action type, target, relationship state before execution, relationship state after execution, and changes in closed state for each governance execution entry;

[0145] The review record set includes the chain identifier, chain position, conflict object, conflict position, conflict relationship, decision basis, return verification result, unclosed position mark, and unclosed relationship type mark for each manual review submission action record chain;

[0146] Based on the set of governance results, the set of review records, the basis for adjudication, and the set of post-governance commitment results, an explanatory evidence package is generated according to the conflict object.

[0147] The explanatory evidence package includes conflicting objects, conflicting locations, conflicting relationships, action types, objects of action, basis for adjudication, relationship status before execution, relationship status after execution, changes in closed status and corresponding mirror object identifiers, manual review submission markers, return verification results, unclosed location markers, and unclosed relationship type markers;

[0148] Based on the governance result set, the review record set, and the interpretation evidence package, the commitment recompilation is performed on the generation rules of the mirror commitment vector set, the source commitment weight, the coverage commitment threshold, and the dual-track consistency adjudication rules to generate a recompiled result set;

[0149] The commitment recompilation includes the following actions: when the post-execution relational state of the corresponding entry in the governance result set is better than the pre-execution relational state and the closure state change is improvement, the generation rules, source commitment weights, coverage commitment thresholds, or dual-track consistency adjudication rules of the corresponding mirror commitment vector set are adjusted upwards; when the post-execution relational state of the corresponding entry in the governance result set is worse than the pre-execution relational state or the closure state change is deterioration, the generation rules, source commitment weights, coverage commitment thresholds, or dual-track consistency adjudication rules of the corresponding mirror commitment vector set are adjusted downwards; when the post-execution relational state of the corresponding entry in the governance result set remains unchanged and the closure state change is unchanged, the generation rules, source commitment weights, coverage commitment thresholds, or dual-track consistency adjudication rules of the corresponding mirror commitment vector set are frozen; when the post-execution relational state of the corresponding entry in the governance result set experiences version rollback, source rollback, or coverage rollback, the generation rules, source commitment weights, coverage commitment thresholds, or dual-track consistency adjudication rules of the corresponding mirror commitment vector set are rolled back.

[0150] The mirror commitment vector set is updated and encapsulated based on the recompiled result set, and the updated mirror commitment vector set is written back to the commitment representation generation module. The recompiled result set is written back to the dual-track consistency verification module and the counterfactual resubmission adjudication module, forming a real-time governance closed loop.

[0151] Example 1: To verify the feasibility of this invention in practice, it was applied to the data governance scenario of a high-end equipment manufacturing enterprise. The enterprise headquarters and its remote factories simultaneously run multiple business systems, including procurement, production, warehousing, quality inspection, and after-sales service. Daily data comes from relational databases, interface receipts, message queues, equipment logs, and batch-imported files. Due to different update rhythms, inconsistent field naming, and chaotic version overwrite sequences among the systems, duplicate records, source mismatches, time inversions, overwrite gaps, and broken historical links frequently occur under the same business primary key. Manual governance often only allows for patching individual records, making it difficult to promptly detect continuity issues across sources and versions. This is precisely the core problem that this invention aims to solve.

[0152] In practical applications, purchase orders, work orders, batches, inspection records, and inbound / outbound records are first imported into the multi-source mirror compilation module. The original data undergoes object splitting, time normalization, source tagging, and version binding to form a continuously updated set of mirror objects. Then, the commitment representation generation module extracts structural integrity, time validity, cross-source coverage relationships, source inheritance relationships, and historical stability relationships from each mirror object, forming a set of mirror commitment vectors. Subsequently, the mirror commitment chain orchestration module organizes the discrete data into a set of mirror commitment chains according to primary key inheritance relationships, version succession relationships, time sequence relationships, source transit relationships, and field dependencies. Upon entering the dual-track consistency verification stage, the system checks whether the preceding and following entries within the chain are continuous and self-consistent. Simultaneously, based on governance rules and standard templates, it checks field specifications, source consistency, and coverage closure status, mapping conflicting entries to governance candidate actions such as field repair, source switching, coverage completion, version rollback, or conflict isolation. The counterfactual reversal adjudication module does not immediately execute these actions. Instead, it first applies each candidate action to the mirror commitment chain, generates a counterfactual mirror commitment chain, and performs a reversal verification. Then, it selects the optimal governance action from those actions with higher closure and less conflict propagation. After the governance closed-loop recompilation module completes its execution, it also writes the governance results and explanatory evidence back into the commitment representation and adjudication rules, continuously revising the basis for subsequent governance.

[0153] The application covers both peak production periods and stable daily operation periods for enterprises, and is applied to data access nodes in headquarters data centers and remote factories. Using governance logs, conflict location records, version write-back records, manual review records, and result traceability records as verification criteria, it can be seen that source conflicts and version mismatches that previously required manual comparison can be exposed in advance after entering the mirror commitment chain; erroneous overwriting and erroneous merging that were prone to occur can be intercepted after counterfactual verification; and problem chains that were previously scattered across different systems can be completely traced along the business primary key, version, and source direction after generating explanatory evidence packages. This demonstrates that the present invention can more stably maintain data continuity, improve governance accuracy, and enhance the traceability and closed-loop update capabilities of the governance process in real-time governance scenarios of multi-source heterogeneous data.

[0154] Table 1. Comparison of Comprehensive Performance of Real-Time Governance Driven by Mirror Commitment Chain

[0155] Mirror object merge accuracy 93.4% 97.1% Cross-source coverage gap identification rate 82.7% 94.6% Source mismatch recognition rate 84.1% 95.3% Version acceptance fracture recognition rate 80.9% 93.8% Time-sequence conflict recognition rate 86.2% 94.9% False trigger rate of candidate governance actions 8.7% 3.4% One-time closure rate of governance actions 78.5% 91.2% Manual review rate 21.6% 9.8% Average treatment time per batch 18.7 min 11.4 min Post-treatment regression rate 6.9% 2.1% Result link traceability completeness rate 88.3% 97.5% Continuous operation stability score 91.5 96.2

[0156] As shown in Table 1, the proposed solution significantly improves the identification metrics, particularly in the identification rates of cross-source coverage gaps, source mismatches, version continuity breaks, and temporal conflicts, increasing from 82.7%, 84.1%, 80.9%, and 86.2% to 94.6%, 95.3%, 93.8%, and 94.9%, respectively. This result indicates that while traditional rule-based governance schemes can perform field-level comparisons and local rule verifications, they tend to fragment problems into single-point anomaly handling when dealing with continuous relationships across sources, versions, and timelines, making it difficult to detect link-level conflicts in a timely manner. The proposed solution, by first forming a set of mirrored objects and then constructing a set of mirrored commitment vectors and a set of mirrored commitment chains, organizes the originally scattered data objects into a continuous governance carrier with primary key inheritance relationships, version continuity relationships, temporal relationships, source transmission relationships, and field dependencies. Therefore, it can more completely expose link breaks and coverage gaps, resulting in higher identification capabilities.

[0157] Regarding governance quality, the false trigger rate of candidate governance actions in this invention decreased from 8.7% to 3.4%, the first-time closure rate of governance actions increased from 78.5% to 91.2%, and the post-governance rollback rate decreased from 6.9% to 2.1%. These data indicate that traditional solutions directly execute repairs after rule matching, which, while having a shorter processing path, easily introduces new conflicts during source switching, version overriding, and overriding completion, leading to subsequent corrections or even rollbacks. This invention, after generating candidate governance actions, does not execute them directly but instead uses a counterfactual mirror commitment chain set for verification, prioritizing governance actions with more closed relationships and fewer unclosed relationships. Therefore, it significantly reduces false repairs, incorrect overriding, and erroneous rollbacks. The proportion of manual review decreased from 21.6% to 9.8%, further demonstrating that this invention moves many complex link conflicts that originally relied on experience-based judgment to the automatic adjudication stage, reducing the intensity of human intervention and improving the automation level of the governance process.

[0158] In terms of governance efficiency and maintainability, the average governance time per batch of the proposed solution has been reduced from 18.7 min to 11.4 min, the traceability and completeness rate of the result chain has increased from 88.3% to 97.5%, and the continuous operation stability score has increased from 91.5 to 96.2. The reduction in governance time is not simply achieved by reducing processing depth, but because the mirror commitment chain orchestration module pre-organizes the preceding versions, source transmission, and field dependencies into a structured chain. The dual-track consistency verification module can more quickly locate conflict objects, conflict locations, and conflict relationships, reducing redundant scanning and judgment. The significant improvement in the traceability and completeness rate is due to the formation of a continuous closed-loop record between the governance result set, the explanation evidence package, and the commitment recompilation result set. This ensures that every field repair, source switch, overwrite completion, version rollback, and conflict isolation can be traced back to the specific chain position, specific conflict relationship, and specific adjudication basis. The improved stability score during continuous operation indicates that the present invention not only performs well in a single governance task, but also can dynamically correct the generation rules of the mirror commitment vector set, the source commitment weight, the coverage commitment threshold, and the dual-track consistent adjudication rules through the commitment recompilation mechanism under continuous operation conditions, thereby maintaining the consistency and stability of the subsequent governance process.

[0159] As can be seen from Table 1, this invention, in the real-time governance of multi-source heterogeneous data, not only improves the ability to identify key issues such as cross-source coverage gaps, source mismatch, version continuity breaks, and time sequence conflicts, but also significantly reduces governance erroneous triggers and post-governance rollbacks through dual-track consistency verification and counterfactual reversal adjudication, further shortening the governance time and enhancing the traceability of the process. This demonstrates that this invention can balance identification accuracy, action closure, execution efficiency, and continuous stability in complex real-time governance scenarios, making it suitable for multi-source heterogeneous data governance application environments with high requirements for continuity, reliability, and interpretability.

[0160] The above are merely preferred embodiments 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 real-time governance system for multi-source heterogeneous data based on data mirroring, characterized in that, include: The multi-source image compilation module is used to acquire heterogeneous data from multiple sources, perform object splitting, time warping, source marking and version binding, and generate a collection of image objects; The commitment representation generation module is used to extract structural integrity, temporal validity, cross-source coverage relationship, source inheritance relationship and historical stability relationship based on the mirror object set, and generate a set of mirror commitment vectors. The mirror commitment chain orchestration module is used to perform chain concatenation on the mirror commitment vector set based on the primary key inheritance relationship, version inheritance relationship, time sequence relationship, source transmission relationship and field dependency relationship in the mirror object set, and generate a mirror commitment chain set. The dual-track consistency verification module is used to perform mirror endogenous consistency verification on the mirror commitment chain set, and to perform exogenous constraint verification in combination with governance rules and standard templates to generate a set of governance candidate actions; The Counterfactual Reversal Decision Module is used to apply the set of governance candidate actions to the set of mirror commitment chains, generate a set of counterfactual mirror commitment chains and perform reversal verification, execute a dual-track consistency decision based on the reversal verification results, and generate the optimal set of governance actions. The governance closed-loop recompilation module is used to apply the optimal set of governance actions to the set of mirror objects, generate a set of governance results and an explanatory evidence package, and perform commitment recompilation on the generation rules, source commitment weights, coverage commitment thresholds and dual-track consistency adjudication rules of the set of mirror commitment vectors, thus forming a real-time governance closed loop.

2. The real-time governance system for multi-source heterogeneous data based on data mirroring according to claim 1, characterized in that, The multi-source image compilation module includes: Acquire multi-source heterogeneous data and parse the data carrier identifier, business field content, original record primary key, original time information and source system information to generate a multi-source raw data set; Perform format unification and field standardization on multi-source raw data sets to generate a unified field data set; Object splitting is performed based on a unified field data set to generate a candidate object set; Perform time warping on the candidate object set to generate a time-warped object set; Perform source tagging on the time-warped object set to generate a source tag object set; Perform version binding on the source tag object collection, generating a version binding object collection; The collection of version-bound objects is compiled into mirror objects to obtain mirror objects. These mirror objects are then aggregated according to the business primary key index, source index, and version index to generate a collection of mirror objects.

3. The real-time governance system for multi-source heterogeneous data based on data mirroring according to claim 1, characterized in that, The commitment representation generation module includes: Read the object identifier, business primary key, unified field data, object time series, source tag and version binding results from the mirror object collection, and generate the structural integrity calculation input set; Based on the input set of structural integrity calculation, a regularization calculation is performed to generate a set of structural integrity results, which is then written back to the corresponding mirror object. Read the object time series, version binding results, and source tags from the mirror object collection, and generate a time validity result set; Based on the business primary key, unified field data and source marker in the mirror object collection, field coverage alignment is performed on cross-source objects under the same business primary key to generate a cross-source coverage relationship result set. Based on the source tags and version binding results in the mirror object collection, generate a source inheritance relationship result set; Based on the version binding results, object time series, and unified field data in the mirror object collection, a set of historical stable relationship results is generated. The result sets of structural integrity, time validity, cross-source coverage, source inheritance, and historical stability are aggregated and encapsulated according to the mirror object identifier to generate a mirror commitment vector set.

4. The real-time governance system for multi-source heterogeneous data based on data mirroring according to claim 1, characterized in that, The mirror commitment chain orchestration module includes: Read the set of mirror objects and the set of mirror commitment vectors, merge the mirror objects and mirror commitment vectors under the same governance object according to the business primary key, and generate the chain orchestration input set; Based on the chain orchestration input set, perform primary key inheritance judgment, version inheritance judgment, time sequence judgment, source transmission judgment and field dependency judgment on mirror objects under the same business primary key to generate a set of mirror object pairs; Write relation type tags and relation direction tags into the mirror object pairing set according to primary key inheritance relationship, version inheritance relationship, time sequence relationship, source transit relationship and field dependency relationship respectively, and generate a relation tag pairing set; Based on the relation tag pairing set, the mirror commitment vectors corresponding to each object pairing are sorted in front and back order to generate a commitment pairing set; Chain the commitment pair set to generate a candidate mirror commitment chain set; Based on the candidate mirror commitment chain set, perform in-chain order correction and generate an order correction result set; The candidate mirror commitment chain set is rearranged and encapsulated based on the order correction result set to obtain the mirror commitment chain set.

5. A real-time governance system for multi-source heterogeneous data based on data mirroring according to claim 1, characterized in that, The dual-track consistency verification module includes: Read the mirror commitment chain set and mirror commitment vector set, and read the governance rules and standard templates. Perform alignment according to chain identifier, in-chain position and relationship record to generate the mirror endogenous consistency check input set and exogenous constraint check input set; Based on the input set of the mirror endogenous consistency verification, the mirror endogenous consistency verification is performed on each mirror commitment chain to generate a set of mirror endogenous consistency results, including chain identifier, chain position, endogenous consistency state, break position marker and break type marker. Based on the input set of exogenous constraint verification, exogenous constraint verification is performed on the mirror commitment chain set to generate an exogenous constraint result set including chain identifier, chain position, exogenous constraint status, constraint violation position marker and constraint violation type marker; A dual-track intersection verification is performed based on the mirrored endogenous consistency result set and the exogenous constraint result set to generate a dual-track intersection result set. Based on the set of results from the convergence of the two tracks, a governance candidate action entry point is established, and conflict actions are mapped according to conflict object, conflict location, and conflict relationship to obtain a set of governance candidate action items. The set of governance candidate action items is encapsulated according to chain identifier, position within the chain, conflicting object, action type, triggering basis, and expected repair relationship to generate a set of governance candidate actions.

6. A real-time governance system for multi-source heterogeneous data based on data mirroring according to claim 5, characterized in that, The conflict action mapping includes generating a field repair action when the conflict relationship is a field specification conflict or a field dependency conflict; generating a source switching action when the conflict relationship is a source transition conflict, or when there is a source switching state, a source rollback state, or a source mismatch state in the result set of the source inheritance relationship; generating a coverage completion action when the conflict relationship is a coverage closure conflict, or when there is a field coverage gap state in the result set of the cross-source coverage relationship; generating a version rollback action when the conflict relationship is a version inheritance conflict, or when there is a version inheritance break, update coverage anomaly, or a previous version reference mismatch in the version binding result; and generating a conflict isolation action when the same conflict object hits two or more types of conflict relationships at the same time, or when there is a mirror endogenous consistency break and an exogenous constraint violation at the same position in the same chain.

7. A real-time governance system for multi-source heterogeneous data based on data mirroring according to claim 1, characterized in that, The counterfactual resubmission adjudication module includes: Read the set of governance candidate actions and the set of mirror commitment chains, align them by chain identifier, position within the chain, conflicting object, action type, triggering basis and expected repair relationship, and generate a return input set; Based on the feedback input set, each governance candidate action item is applied to the corresponding mirror commitment chain set to generate a counterfactual mirror commitment chain set. Based on the counterfactual mirror commitment chain set, a resubmission verification is performed, generating a resubmission verification result set; Based on the set of feedback verification results, perform dual-track consistent adjudication input reconstruction to generate a dual-track consistent adjudication input set; Based on the input set of the dual-track consistent decision, a dual-track consistent decision is executed to generate the optimal set of governance actions.

8. A real-time governance system for multi-source heterogeneous data based on data mirroring according to claim 7, characterized in that, The dual-track consistent adjudication compares the primary key inheritance closure state, version inheritance closure state, time sequence closure state, source transmission closure state, field dependency closure state, field specification closure state, and coverage closure state of multiple governance candidate action entries corresponding to the same conflict object. It prioritizes the governance candidate action entries with fewer unclosed relationship types. When the number of unclosed relationship types is the same, it prioritizes the governance candidate action entries with fewer unclosed position markers. When the number of unclosed position markers is the same, it prioritizes the governance candidate action entries with more conflict relationships corresponding to the expected repair relationship and the dual-track intersection result set. When there is still no closure action for the same conflict object after the aforementioned action mapping, or when it still fails the resubmission verification after a preset number of counterfactual resubmissions, a manual review submission action is generated. The optimal governance action set is obtained by encapsulating the chain identifier, chain position, conflict object, action type, target, adjudication basis, closure state, and expected repair relationship.

9. A real-time governance system for multi-source heterogeneous data based on data mirroring according to claim 1, characterized in that, The governance closed-loop recompilation module includes: Read the optimal governance action set, mirror object set, and mirror commitment vector set, align them according to chain identifier, in-chain position, conflict object, action type, target, adjudication basis, closure state, and expected repair relationship, generate a governance execution input set, and generate a review record input set for the manual review submission actions in the optimal governance action set; Based on the governance execution input set, the field repair action, source switching action, overwrite and completion action, version rollback action and conflict isolation action in the optimal governance action set are applied to the corresponding image object set to generate the post-governance image object set. Based on the post-governance mirror object set, the results of structural integrity, time validity, cross-source coverage, source inheritance, and historical stability are updated and aggregated on the affected mirror objects to generate a post-governance commitment result set; Based on the governance execution input set and the post-governance commitment result set, a governance result set is generated; based on the review record input set, a review record set is generated. Based on the set of governance results, the set of review records, the basis for adjudication, and the set of post-governance commitment results, an explanatory evidence package is generated according to the conflict object. Based on the governance result set, the review record set, and the interpretation evidence package, the commitment recompilation is performed on the generation rules of the mirror commitment vector set, the source commitment weight, the coverage commitment threshold, and the dual-track consistency adjudication rules to generate a recompiled result set; The mirror commitment vector set is updated and encapsulated based on the recompiled result set, and the updated mirror commitment vector set is written back to the commitment representation generation module. The recompiled result set is written back to the dual-track consistency verification module and the counterfactual resubmission adjudication module, forming a real-time governance closed loop.

Images