A big data-based intelligent civil affairs supervision and management system

Abstract

The application discloses a kind of wisdom civil affairs supervision and management systems based on big data, it is related to civil affairs informatization and data supervision technical field, by collecting the multi-source data of civil affairs object in identity, household, relief, asset, fund distribution and service performance, construct static continuous feature and dynamic continuous feature, generate the composite identity index containing identity stability fingerprint and behavior coherence fingerprint, and based on the index carries out candidate association and two-way countercheck verification, to identify primary key legal conflict and generate conflict isolation identification;In the case where conflict is not triggered, the multi-source data is fused to build civil affairs object data chain, and the association path and source information are written in the data chain, finally generate relief qualification, asset anomaly, fund distribution and service performance supervision result;The application can effectively avoid different object data mismerger, improve the reliability and traceability of data association accuracy and supervision result.

Description

Technical Field

[0001] This invention relates to the field of civil affairs informatization and data supervision technology, specifically to a smart civil affairs supervision and management system based on big data. Background Technology

[0002] As the level of informatization in civil affairs operations continues to improve, social assistance, elderly care services, fund disbursement, and asset verification are increasingly relying on multi-source data for collaborative management. However, existing technologies typically use a single identity identifier to directly link data from different systems. In cases of duplicate ID numbers, historical data entry errors, or reuse of temporary numbers, data from different civil affairs recipients can be incorrectly merged, leading to data chain confusion. This can result in biased assessments of eligibility for assistance, distorted identification of asset anomalies, and misjudgments in fund supervision. Furthermore, existing methods lack the comprehensive verification capability to consider both static structures such as household registration migration and family relationship evolution, and dynamic behaviors such as subsidy disbursement rhythms and service performance trajectories. This makes it difficult to effectively identify potential conflicts and abnormal relationships in cross-system data, resulting in insufficient accuracy and traceability of supervisory results. Summary of the Invention

[0003] The purpose of this invention is to provide a smart civil affairs supervision and management system based on big data to address the shortcomings of the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a smart civil affairs supervision and management system based on big data, comprising: The data acquisition module collects multi-source data on civil affairs recipients in terms of identity, household registration, assistance, assets, fund disbursement, and service performance. After field normalization, time calibration, and anomaly cleaning, a basic dataset is formed. The continuous feature construction module uses identity identifiers as the initial anchor point to construct static continuous features such as household registration migration, family relationship evolution, and assistance application history, and constructs dynamic continuous features such as subsidy receipt rhythm, service usage trajectory, and cross-domain business activity sequence. The composite identity index generation module generates an identity stability fingerprint based on static continuous features and a behavioral coherence fingerprint based on dynamic continuous features, and then combines the two with the initial anchor point to form a composite identity index. The candidate association module performs candidate association on multi-source data based on the composite identity index to obtain a set of objects to be merged. The conflict rebuttal verification module performs bidirectional rebuttal verification of the set of objects to be merged, including static stability and dynamic coherence. When the initial anchor points are the same but the identity stability fingerprint shows a break, or the behavioral coherence fingerprint shows an incompatible trajectory, it determines that there is a valid primary key conflict and generates a conflict isolation flag to prohibit merging. The data chain fusion module, when neither bidirectional rebuttal verification is triggered, merges the corresponding data into a civil affairs object data chain and writes the associated path, source, and verification result. The intelligent supervision output module generates results for monitoring eligibility for assistance, asset anomalies, fund disbursement, and service performance based on the data chain of civil affairs beneficiaries.

[0005] Preferably, the data acquisition module includes: Data from multiple sources, including civil affairs, household registration management, social assistance, asset verification, government subsidy disbursement, and service performance, is collected separately. Each data entry is then accompanied by the data source, collection time, transaction time, and data batch identifier. Unify the field names, data formats, and coding rules of identity, time, administrative division, relief category, and service category fields from different sources; The normalized data undergoes integrity, consistency, duplication, and validity checks. Data that cannot be corrected or has significant conflicts is written to the abnormal data area to form the basic dataset.

[0006] Preferably, the continuous feature construction module includes: Using the normalized identity field in the basic dataset as the initial anchor point, extract the household registration history field, family relationship field, and relief history field, and form a static event sequence according to the time of business occurrence; The system segments and compresses household registration migration records in static event sequences, expands family relationship change records hierarchically, and divides relief application records into stages to generate static continuous features. Using the fund disbursement time sequence field and service performance time sequence field as input, the subsidy disbursement records are mapped at periodic intervals, the service records are reorganized for trajectory coherence, and dynamic continuous features are formed by combining the order of cross-domain business occurrences.

[0007] Preferably, the continuous feature construction module includes the following when forming dynamic continuous features: Subsidy disbursement records are grouped according to subsidy items and receiving account identifiers, and the corresponding disbursement rhythm is determined when the interval between three or more consecutive disbursements falls within the range of 25 to 35 days per month, 80 to 100 days per quarter, or 350 to 380 days per year. Service records are sorted according to service time, service confirmation time, and service item registration order, and service locations are converted to a unified administrative division level; When the interval between adjacent services deviates from the established distribution rhythm by more than 20%, or when adjacent service locations are in different administrative divisions and the time interval is less than 1 day and lacks remote service or agency identification, the corresponding time period is marked as abnormal to obtain dynamic continuous characteristics.

[0008] Preferably, the composite identity index generation module includes: Structural compression and sequential encoding are performed on the household registration migration sequence, family relationship hierarchy and assistance stage division results in static continuous features. Key structural fragments with a duration of no less than 180 days or no more than 2 changes within 12 consecutive months are extracted to generate identity stability fingerprints. The subsidy disbursement rhythm, service fulfillment trajectory, and cross-domain business sequence in the dynamic continuous features are reconstructed and path-corrected. Behavioral segments that appear at least 3 times and whose intervals between adjacent segments deviate from the corresponding period range by no more than 20% are extracted to generate behavioral coherence fingerprints. The identity stability fingerprint and behavior coherence fingerprint are segmented and aligned according to the identity anchor segment, time location segment, structural content segment, source segment, and anomaly marker segment, and then written together with the initial anchor into the identifier structure to form a composite identity index.

[0009] Preferably, the candidate association modules include: Using the initial anchor point in the composite identity index as the grouping key, the multi-source data is initially aggregated to obtain several candidate object subsets with the same initial anchor point. In the candidate subset, the household registration segment, relationship segment, and rescue segment in the identity stability fingerprint are compared structurally. When at least two consecutive segments have the same order, the same segment type, and a time overlap ratio of 70% to 100%, the first candidate association set is formed. Perform sequential consistency verification on the behavioral coherence fingerprints in the first candidate association set, remove object pairs with broken behavioral trajectories or conflicting business sequences, and merge and aggregate the retained objects to obtain the set of objects to be merged.

[0010] Preferably, the conflict rebuttal verification module includes: Extract identity stability fingerprints and behavioral coherence fingerprints from the set of objects to be merged, and form a two-way verification sequence according to the initial anchor point, fingerprint version, update time, source layer and collection batch; Continuous counter-verification processing is performed on identity stability fingerprints. When household registration fragments overlap in different administrative regions for more than 30 days, or when there are mutually exclusive family relationship levels within the same time period, or when there is a withdrawal in the assistance stage before the disbursement and no record of reapplication, a static conflict marker is generated. For behavioral coherence fingerprints, trajectory reversal processing is performed. When adjacent business locations are across prefecture-level cities and the time difference is less than 1 day, or across provincial administrative regions and the time difference is less than 2 days, and there is a lack of remote processing, off-site settlement, or entrusted processing markers, dynamic conflict markers are generated.

[0011] Preferably, the conflict rebuttal verification module includes: When static or dynamic conflict markers exist, a conflict isolation marker is generated based on the initial anchor point, merge identifier, conflict fragment location, conflict type, and processing batch. Remove the set of objects with conflict isolation markers from the mergeable path, while retaining the original source, bidirectional check sequence, conflict markers, and decision criteria; When neither static nor dynamic conflict markers exist, the corresponding set of objects is marked as merging-ready, and the initial anchor point, merge identifier, bidirectional verification sequence, and verification result identifier are output.

[0012] Preferably, the data link fusion module includes: The set of objects to be merged that have not triggered conflict markers is used as input. Data from each source is collected according to the initial anchor point, and an ordered data sequence is formed based on the business occurrence time, business type order, source trust level, and collection batch. Align the positions of data of the same business type in the ordered data sequence, and determine the retention priority of duplicate records with the same business number, time position, object identifier and business status, and generate a set of data fragments to remove redundancy; The set of redundant data segments is concatenated in chronological order to form a civil affairs object data chain, and the associated path information, data source information and two-way reversal verification result identifier are written into the civil affairs object data chain.

[0013] Preferably, the intelligent monitoring output module includes: Read identity fragments, household registration fragments, family relationship fragments, relief fragments, asset verification fragments, fund disbursement fragments, and service performance fragments from the data chain of civil affairs beneficiaries, and determine the scope of supervision for the same civil affairs beneficiary based on the initial anchor point and the merge identifier; The results of the supervision of eligibility for assistance are generated based on the assistance segment, household registration segment, family relationship segment, and asset verification segment, and the results of the supervision of abnormal assets are generated based on the correspondence between the asset registration time and the assistance period. The system generates fund disbursement supervision results based on the correspondence between fund disbursement segments and relief approval segments, and generates service performance supervision results based on service performance segments, service dispatch segments, and recipient status segments.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention constructs a composite identity index constrained by both static and dynamic continuous features, transforming the original planar association method relying solely on a single identity identifier into a multi-dimensional physically constrained association method based on temporal continuity, spatial consistency, and structural stability. Household registration migration, family relationships, and assistance stages reflect the long-term structural attributes of an object in the real world, with changes exhibiting low-frequency, continuous objective characteristics. In contrast, subsidy disbursement rhythms, service fulfillment trajectories, and cross-domain business sequences reflect the behavioral paths of an object in time and space, with changes constrained by real-world activities, making it difficult for them to undergo disordered jumps across regions or states in a short period. This invention synchronously models these two types of features and performs bidirectional reversal verification before data association, preventing the merging of records that do not meet physical continuity and behavioral consistency requirements, fundamentally avoiding erroneous merging problems caused by primary key legal conflicts. The direct technical effect is that data from different sources but with inconsistent physical trajectories are automatically isolated, the data chain maintains the consistency of real objects, and the reliability of data association is improved from the source.

[0015] 2. This invention introduces an ordered time-series processing method, along with hierarchical source retention and conflict marker embedding, during the data chain construction process. This enables the merged civil affairs object data chain to possess not only a continuous time structure but also complete source paths and verification information. All business data is sorted according to actual occurrence time and established with hierarchical relationships based on business type, allowing the data chain to accurately reflect the behavioral evolution of objects in reality. Simultaneously, by selecting or retaining duplicate or conflicting data through source trust levels and time priority rules, the data chain retains multi-source information while avoiding information overwriting. Unlike existing methods that only output result data, the data chain generated by this invention inherently includes association paths and conflict determination criteria. Therefore, when generating subsequent results for assistance eligibility, asset anomalies, and fund disbursement supervision, it is possible to directly trace back to specific business segments and data sources. The resulting technical effects are: not only improved accuracy of regulatory results but also significantly enhanced interpretability and traceability, reducing manual review costs. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart of a smart civil affairs supervision and management system module based on big data according to the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] For examples, please refer to Figure 1 As shown in this embodiment, a smart civil affairs supervision and management system based on big data includes: In this invention, the data acquisition module is used to collect multi-source data on civil affairs subjects in terms of identity, household registration, relief, assets, fund disbursement and service performance, and to perform field normalization, time calibration and anomaly cleaning on the collected data to form a basic dataset that can be used for subsequent composite identity index construction.

[0020] Specifically, the civil affairs beneficiaries include individuals, families, elderly care service recipients, assistance recipients, and institutions related to civil affairs operations. The data collection module establishes data interfaces with the civil affairs business system, household registration management system, social assistance system, asset verification system, fiscal subsidy disbursement system, elderly care service system, and third-party collaborative systems to collect the beneficiaries' identity data, household registration data, assistance data, asset data, fund disbursement data, and service performance data. Identity data includes at least name, identification, gender, date of birth, and contact information; household registration data includes at least registered address, migration records, and family member relationships; assistance data includes at least application time, assistance category, approval status, and duration of benefits; asset data includes at least verification results for real estate, vehicles, business entities, or other assets; fund disbursement data includes at least subsidy items, disbursement time, disbursement amount, and receiving account identifier; and service performance data includes at least service items, service location, service time, number of services, and service confirmation results.

[0021] Regarding data collection methods, the data acquisition module uses API calls, database synchronization, batch file imports, or message subscriptions to acquire data based on the openness conditions of different data sources. For fund disbursement data and service performance data with high real-time requirements, an incremental collection method is used, collecting only newly added or changed data based on business timestamps or data change identifiers. For periodically updated data such as household registration migration and asset verification, a scheduled collection method is used to update data according to a preset cycle. Each piece of collected data is accompanied by the data source, collection time, business occurrence time, and data batch identifier for subsequent traceability.

[0022] After data collection is completed, the data collection module performs field normalization on the multi-source data. Field normalization includes standardizing field names, data formats, and encoding rules. For example, "ID number," "ID card number," and "identity number" from different systems are uniformly mapped to an identity identifier field; date data in different formats is standardized to a standard date format; and administrative divisions, assistance categories, and service categories expressed in different ways are converted into a unified code. When the same field has multiple sources, its source identifier is retained, and both the original field value and the normalized field value are stored simultaneously.

[0023] Subsequently, the data acquisition module performs time calibration on the normalized data. This time calibration involves distinguishing between the business occurrence time, system recording time, and data acquisition time, and determining the effective time for subsequent processing according to a preset time priority rule. For data with cross-system delays, the data acquisition module does not directly substitute the acquisition time for the business occurrence time. Instead, it determines the position of the data in the historical sequence of civil affairs recipients based on the business document time, approval time, disbursement time, or service confirmation time. This avoids errors in the order of household registration changes, assistance applications, fund disbursements, or service performance records due to data synchronization delays.

[0024] The data acquisition module performs anomaly cleaning on the time-calibrated data. Anomaly cleaning includes integrity verification, consistency verification, duplication verification, and validity verification. Integrity verification identifies data with missing key fields such as identity identifier, business time, and business type; consistency verification identifies data with obvious contradictions between name and identity identifier, date of birth and identity identifier, and registered address and administrative division code; duplication verification identifies duplicate records from the same or different sources with highly consistent business numbers, identity identifiers, issuance batches, and service records; validity verification identifies data exceeding the scope of business rules, such as negative issuance amounts, service times later than the collection time, and new service records being generated for cancelled entities.

[0025] For abnormal data discovered during the cleaning process, the data acquisition module handles the data according to the type of abnormality: for data with incorrect format but which can be corrected, it corrects the data according to preset rules and records the correction traces; for data with missing key fields but which can be supplemented by data from the same source or trusted related data, it completes the fields and marks the source of the supplementation; for data that cannot be corrected or has major conflicts, it is not included in the direct fusion process, but is written into the abnormal data area, and the original data, the cause of the abnormality and the source information are retained for subsequent conflict verification or manual verification.

[0026] After the above processing, the data acquisition module generates a basic dataset. This basic dataset includes at least the following fields: unified identity field for civil affairs recipients, household registration history field, assistance history field, asset verification field, fund disbursement time series field, service performance time series field, data source field, effective time field, and anomaly marker field. This basic dataset preserves the original source relationships of multi-source data while forming a unified data representation structure. It provides a data foundation for subsequent static continuous features based on household registration migration, family relationship evolution, and assistance application history, as well as dynamic continuous features based on subsidy receipt rhythm, service usage trajectory, and cross-domain business activity sequence.

[0027] The continuous feature construction module uses identity identifiers as the initial anchor point to construct static continuous features such as household registration migration, family relationship evolution, and assistance application history, and constructs dynamic continuous features such as subsidy receipt rhythm, service usage trajectory, and cross-domain business activity sequence.

[0028] In one implementation, the unified identity field of civil affairs objects in the basic dataset is used as the initial anchor point. The household registration history field, family relationship field and assistance history field under the same initial anchor point are extracted, and an event item is created for each record. The event item includes at least the event type, business occurrence time, source identifier, administrative division, related personnel identifier and business number.

[0029] The time of a business transaction is prioritized based on the time the business document was created, followed by the time the approval was completed, and then the time the data was collected. When multiple time fields exist for the same record, the valid time is determined according to the above priority order. For records that lack a valid time but have a business batch number, the missing time is supplemented by records in the same batch that already have a valid time; records that cannot be supplemented are retained but are not included in the order reordering.

[0030] The events are then arranged from earliest to latest according to their validity time. If the validity times are the same, they are sorted in the following order: event type priority, source identifier trust level, and business number. The event type priority is: change of household registration, change of family relationship, and change of assistance application. The sorted event items constitute a static event sequence with time constraints.

[0031] Based on static event sequences, household registration migration records, family relationship change records, and assistance application records are processed continuously. When segmenting and compressing household registration migration records, two adjacent records with the same administrative division, consistent standardized household registration address, and a time interval not exceeding 30 days are merged into the same household registration segment. The segment start time is the earlier record's effective time, and the segment end time is the later record's effective time. If the administrative division or address standardization results are different, they are retained as separate segments. When processing family relationship changes hierarchically, relationships such as spouse, parents, children, guardians, and co-applicants are expanded into independent relationship segments according to their relationship hierarchy, and the relationship's effective time and termination time are written into the corresponding segments. When the same relationship overlaps in different sources, segments with more up-to-date effective times and higher source credibility are prioritized for retention.

[0032] When processing relief application records into stages, the application, acceptance, review, approval, disbursement, suspension, and withdrawal stages are divided into continuous stages. Duplicate records with the same business number and stage status are deleted. Stages with overlapping time intervals and mutually exclusive statuses are marked for conflict, and conflict-marked records are not included in the static continuous feature. After processing, a continuous and conflict-free static continuous feature is formed.

[0033] Using the time-series fields of fund disbursement and service performance in the basic dataset as input, the subsidy disbursement records and service records are dynamically sorted in sequence.

[0034] When mapping subsidy disbursement records to periodic intervals, the records are first grouped by subsidy project and receiving account identifier, and then the time interval between two adjacent disbursement records is calculated as the number of natural days between the later disbursement time and the previous disbursement time. When three or more consecutive intervals fall within the same period range, the disbursement rhythm of the group of records is determined. The period range includes a monthly period of 25 to 35 days, a quarterly period of 80 to 100 days, and an annual period of 350 to 380 days.

[0035] When reconstructing service records to ensure consistency, they are arranged from earliest to latest by service time, and the service locations are converted to a unified administrative division level. For multiple service records on the same day, they are sorted by service confirmation time; if a confirmation time is missing, they are sorted by the order of service item registration. Subsequently, the paths are rearranged based on the order of cross-regional business occurrences, connecting the fund disbursement location, service location, and business processing location in chronological order to form a dynamic event sequence with temporal consistency.

[0036] Based on dynamic event sequences, consistency screening is performed on the intervals between adjacent business events. First, the time interval between two adjacent dynamic events is calculated as the number of natural days between the effective time of the later event and the effective time of the earlier event. Then, the location and business type of the two adjacent dynamic events are compared. If the adjacent disbursement intervals for the same subsidy project deviate from the established disbursement rhythm by more than 20%, the corresponding time period is marked as rhythm abnormal. If two adjacent service locations are in different administrative divisions and the time interval is less than one day, and there are no remote service or agency identifications, the corresponding time period is marked as trajectory abnormal. If the order of business types is inconsistent with the actual business process, for example, if an exit record appears first and then the first disbursement record within the same relief period appears, the corresponding time period is marked as sequence abnormal.

[0037] For time periods with abnormal rhythms, abnormal trajectories, and abnormal sequences, if the data can be corrected using data of a high level of source credibility, the corrected events are used for concatenation. If the data cannot be corrected, either removal or retention of markers is performed. Removed records do not participate in subsequent continuous representation, while retained marker records are only used as evidence against the contrary. The processed dynamic event sequences are then reconnected in chronological order to obtain dynamic continuous features reflecting behavioral continuity, which are used for subsequent construction of composite identity indexes.

[0038] The composite identity index generation module generates identity stability fingerprints based on static continuous features and behavioral coherence fingerprints based on dynamic continuous features, and combines the two with the initial anchor point to form a composite identity index.

[0039] Using static continuous features as input, the system first reads the household registration migration sequence, family relationship hierarchy, and assistance stage division results, and then forms household registration fragments, relationship fragments, and assistance fragments respectively.

[0040] When compressing the structure of household registration segments, segments with the same continuous administrative division, the same standardized household registration address, and an interval of no more than 30 days are merged. After merging, the start time, end time, administrative division code, and address standardization code are retained. When compressing the family relationship hierarchy, spouses, parents, children, guardians, and joint applicants are assigned fixed level codes, and relationship segments with the same related persons, the same relationship type, and continuous time intervals are merged. When compressing the results of the assistance stage division, the stage code and stage time are retained according to the business sequence of application, acceptance, review, approval, disbursement, suspension, and withdrawal.

[0041] During sequential encoding, segments are arranged from earliest to latest based on their effective time. If the time is the same, they are arranged in the order of household registration segment, relationship segment, and assistance segment to obtain a static structure string. Key structural segments are extracted through stability conditions, which are that the segment duration is not less than 180 days, or the number of changes does not exceed 2 within 12 consecutive months. For segments that meet the stability conditions, the segment type, level code, time position, and source identifier are retained, and then they are concatenated in order of appearance to obtain an identity stability fingerprint that can represent the long-term stable state of the identity.

[0042] Using dynamic and continuous features as input, the system first reads the subsidy disbursement rhythm, service fulfillment trajectory, and cross-domain business sequence, and then forms disbursement segments, service segments, and cross-domain segments respectively.

[0043] When reconstructing the subsidy disbursement rhythm, the subsidy items and receiving account identifiers are grouped together, and the number of natural days between two adjacent disbursement times is calculated. If the intervals fall within the same range of 25 to 35 days in a monthly cycle, 80 to 100 days in a quarterly cycle, or 350 to 380 days in an annual cycle for more than three consecutive times, then the group is reconstructed into a disbursement cycle segment, and the cycle type, the first and last disbursement times, and the number of disbursements are recorded.

[0044] When organizing service delivery routes, service locations are converted to a unified administrative division level and arranged sequentially according to service time, service confirmation time, and service item registration order; multiple similar service records within the same district on the same day are merged into one service segment. When organizing cross-regional business sequences, the fund disbursement location, service location, and business processing location are connected by time, while retaining changes in the administrative division level of adjacent locations.

[0045] Repeatable and coherent behavioral segments are determined by the number of occurrences and time intervals. Segments that occur at least 3 times and whose intervals between adjacent segments deviate from the corresponding period range by no more than 20% are retained. They are then combined in the order of occurrence to obtain a behavioral coherence fingerprint that can characterize the consistency of behavior.

[0046] When segmenting and aligning identity stability fingerprints with behavioral coherence fingerprints, a uniform length rule is first set. The uniform length rule divides each fingerprint into an identity anchor segment, a time location segment, a structural content segment, a source segment, and an anomaly marker segment. The length of each segment is represented by a fixed number of characters. The fixed number of characters can be set to 8, 16, or 32 bits depending on the scale of the deployed data, and should remain consistent during the same batch of processing.

[0047] If the actual number of characters in a segment is less than the fixed number of characters, a space marker is added to the end of the segment; if the actual number of characters in a segment exceeds the fixed number of characters, the segment with the earliest time, the higher source credibility level, and the longer duration is retained, and the excess part is written into the extended segment.

[0048] After alignment, the correspondence between the identity stability fingerprint and the behavior coherence fingerprint is compared segment by segment in terms of time position, source identifier, and anomaly marker. If the corresponding segments have different sources but are time-continuous, a source difference identifier is written; if the corresponding segments are not time-continuous or have anomaly markers, a position difference identifier is written. After segment alignment and difference segment position marking, a two-layer fingerprint structure is formed.

[0049] When embedding the two-layer fingerprint structure into the identification structure of the corresponding civil affairs object using the initial anchor point as the primary index, a primary index item is first established using the normalized identity field. Then, under the primary index item, the identity stability fingerprint, behavioral coherence fingerprint, segment alignment result, and difference segment position identifier are written sequentially. The identification structure includes at least the initial anchor point, fingerprint version, static layer, dynamic layer, alignment layer, source layer, and update time. The fingerprint version increases according to the processing batch. The static layer stores the identity stability fingerprint, the dynamic layer stores the behavioral coherence fingerprint, the alignment layer stores the two-layer fingerprint structure, and the source layer stores the data source and collection batch corresponding to each segment.

[0050] After writing is complete, a uniqueness check is performed on multiple identifier structures under the same initial anchor point. If both the identity stability fingerprint and the behavioral coherence fingerprint are consistent, they are retained as the same composite identity index. If the initial anchor points are consistent but any fingerprint has a break or an incompatible trajectory, multiple composite identity indexes are retained and an identifier to be verified is attached. Through the above processing, a composite identity index that can be used for cross-source data association and conflict verification is obtained.

[0051] The candidate association module performs candidate association on multi-source data based on the composite identity index to obtain a set of objects to be merged.

[0052] Using the composite identity index as input, the initial anchor point, identity stability fingerprint, behavioral coherence fingerprint, source layer, and update time are read from each index, and the initial anchor point is used as the grouping key to perform preliminary aggregation of multi-source data.

[0053] During data aggregation, records with identical initial anchor points are first placed into the same subset of candidate objects. Then, the original hierarchical relationships are preserved according to data source, business type, and collection batch to avoid directly overwriting field values ​​in the initial grouping stage. Records with empty initial anchor points, abnormal formats, or placeholder numbers are not included in the normal grouping path but are written to the pending completion area. For records with multiple composite identity index versions under the same initial anchor point, they are arranged from most recent to oldest update time, and the fingerprint differences between versions are preserved.

[0054] After initial aggregation, several candidate object subsets are formed. Each candidate object subset contains at least one civil affairs object record with the same initial anchor point and its corresponding two-layer fingerprint structure, providing input for subsequent object pair screening.

[0055] In each subset of candidate objects, the identity stability fingerprint and behavioral coherence fingerprint corresponding to each object are extracted. First, structural fragment comparison is performed on the identity stability fingerprint. Structural fragment comparison uses household registration fragments, relationship fragments, and assistance fragments in the static layer as comparison units, comparing fragment type, time location, administrative division code, relationship level code, assistance stage code, and source layer information in sequence. The condition for consecutive fragment overlap is set as follows: adjacent fragments have the same arrangement order, the same fragment type, overlapping time intervals, and the overlapping duration accounts for 70% to 100% of the shorter fragment's duration; when the data source credibility level is lower than the preset credibility level, this proportion is increased to 80% to 100%. If at least two consecutive fragments in the identity stability fingerprints of two objects meet the above conditions, and there are no mutually exclusive relationship fragments at the same time location, then the two objects are identified as an object pair. All object pairs that meet the conditions are aggregated to form the first candidate association set.

[0056] Based on the first candidate association set, the behavioral coherence fingerprints corresponding to the object pairs are further read, and sequential consistency checks are performed. Sequential consistency checks are performed according to the order of occurrence of disbursement fragments, service fragments, and cross-domain fragments in the dynamic layer. First, the disbursement rhythm of the same subsidy project is compared to see if it is within the same period range. Then, the service fulfillment trajectory is compared to see if it can be connected continuously in chronological order. Finally, the order of cross-domain business is compared to see if there are any inconsistencies.

[0057] Incompatible scenarios include: two adjacent business locations located in different administrative regions with a business time interval of less than one day, and lacking remote service or agency markings; mutually exclusive disbursement statuses for the same subsidy program within the same time period; and an exit status occurring before the initial disbursement status within the same relief period with no record of reapplication. If no such behavioral trajectory breaks or sequence conflicts exist, the object pair is retained in the second candidate association set; if any incompatible scenario exists, it is removed from the first candidate association set, and the reason for removal, the corresponding segment location, and the data source are recorded.

[0058] When aggregating objects in the second candidate association set using merge identifiers, an aggregation chain is first established based on the association relationships between object pairs. When object A and object B meet the association conditions, and object B and object C meet the association conditions, A, B, and C are included in the same aggregation chain, and the chain is further checked to see if there is a recorded reason for removal between any two objects. If no reason for removal exists, the same merge identifier is assigned to the aggregation chain; if a reason for removal exists, the chain is split into multiple merge identifiers according to the segment position corresponding to the reason for removal. Each merge identifier retains the object source, initial anchor point, identity stability fingerprint, behavioral coherence fingerprint, associated object pairs, continuous segment overlap records, and sequence consistency verification records. After aggregation, a set of objects to be merged containing data from multiple sources and having association relationships is formed. This set of objects to be merged, along with the merge identifier, fingerprint difference positions, and verification records, is output for subsequent conflict rebuttal verification processing.

[0059] The conflict rebuttal verification module performs bidirectional rebuttal verification of the set of objects to be merged, checking both static stability and dynamic coherence. When the initial anchor points are the same but the identity stability fingerprint shows a break, or the behavioral coherence fingerprint shows an incompatible trajectory, it determines that there is a valid primary key conflict and generates a conflict isolation flag, prohibiting merging.

[0060] Taking the set of objects to be merged as input, the initial anchor point, identity stability fingerprint and behavior coherence fingerprint of each object are read, and the objects are arranged in the same group according to the initial anchor point.

[0061] For multiple objects under the same initial anchor point, the sorting order is first determined by fingerprint version and update time, and then supplemented by source layer and collection batch. When the update time is the same, the data source with the higher source trust level is given priority. The trust level can be pre-configured from level 1 to level 5 according to the data source, with level 5 being the highest level. Legal business sources such as household registration and fiscal disbursement can be configured to level 4 or 5, while manually supplemented or historically migrated data can be configured to level 1 to level 3.

[0062] After the arrangement is completed, the identity stability fingerprint is placed in the static verification layer, the behavior coherence fingerprint is placed in the dynamic verification layer, and the static verification layer and dynamic verification layer of the same object are aligned according to the time position.

[0063] At least the fragment type, start time, end time, administrative division, business type, source identifier, and anomaly marker are recorded at each time location, thereby forming a comparable two-way verification sequence.

[0064] Based on a two-way verification sequence, continuous reversal verification processing is performed on the identity stability fingerprint. This processing uses the order of household registration migration, family relationship hierarchy, and assistance stage as the objects of reversal verification, and checks segment by segment whether there are static structures that cannot be simultaneously established between different objects under the same initial anchor point.

[0065] When checking the order of household registration migration, if the end time of the preceding household registration segment is later than the start time of the following segment, and the two segments have different administrative divisions and address codes, and the overlap period exceeds 30 days, it is considered a broken household registration segment. When checking the family relationship hierarchy, if the same related person has mutually exclusive relationships within the same time period, such as a spouse and parents coexisting, or a guardian and ward having opposite relationships, and the overlap period reaches more than 50% of the duration of the shorter relationship segment, it is considered a relationship exclusion segment. When checking the relief stage, if the exit stage is earlier than the disbursement stage but there is no reapplication stage, or the suspension stage overlaps with the normal disbursement stage for more than one disbursement cycle within the same relief period, it is considered a conflicting relief stage segment. When any of the above segments exist, a static conflict marker is written, and the location, conflict type, and source identifier of the conflict segment are recorded.

[0066] When performing trajectory reversal verification on behavioral coherence fingerprints, the verification is based on the order of the business occurrence time, location, and business type. Each segment of the dynamic verification layer is checked to see if there are any behavioral segments with overlapping times or disjointed paths.

[0067] During time overlap checks, the start and end times of adjacent or parallel segments are compared. If the same subsidy project has two mutually exclusive disbursement statuses within the same disbursement cycle, or the same service project appears in two different administrative regions within the same service time period, and there are no agency, remote service, or supplementary record markers, then it is considered a time overlap segment.

[0068] When checking for unconnectable paths, the administrative division level and time difference of two adjacent business locations are read. If the location crosses a prefecture-level city and the time difference is less than 1 day, or crosses a provincial-level administrative region and the time difference is less than 2 days, and there is a lack of remote processing, off-site settlement or entrusted processing markers, it is identified as an unconnectable path segment.

[0069] During the business type sequence check, if the preceding business that subsequent business depends on is missing—for example, fund disbursement before approval completion, or service fulfillment before service order dispatch—and cannot be corrected by records with a source trust level of 4 or higher, it is identified as a sequence conflict segment. If any of the above segments exist, a dynamic conflict flag is written, and the corresponding time, location change, and business type are recorded.

[0070] When static or dynamic conflict markers exist, a conflict isolation identifier is generated for the corresponding set of objects to be merged. The conflict isolation identifier is formed by sequentially concatenating the initial anchor point, merge identifier, conflict fragment position, conflict type, and processing batch, and is used to distinguish records that may represent different civil affairs objects under the same initial anchor point.

[0071] After generating a conflict isolation marker, the corresponding object set is removed from the mergeable path, while retaining the original source, bidirectional verification sequence, static conflict marker, dynamic conflict marker, and judgment criteria. This object set can only enter conflict review or subsequent supplementary verification and does not participate in the current data chain construction. When neither the static conflict marker nor the dynamic conflict marker exists, the corresponding object set is marked as merging-ready, and the initial anchor point, merge marker, bidirectional verification sequence, and verification result marker are output as inputs for subsequent civil affairs object data chain construction.

[0072] The data chain fusion module merges the corresponding data into a civil affairs object data chain when neither bidirectional rebuttal verification is triggered, and writes the associated path, source and verification result.

[0073] The set of objects to be merged that have not triggered conflict flags after conflict rebuttal verification is taken as input. The initial anchor point, merge identifier, source identifier, business type, business occurrence time and verification result identifier are read from it, and the data from each source are sorted uniformly according to the initial anchor point.

[0074] During sorting, all records under the same civil affairs recipient are first aggregated according to the initial anchor point, and then arranged from earliest to latest according to the time of the transaction. For transactions with the same time of occurrence, they are sorted in the following order: transaction type, source reliability level, and collection batch. The transaction type order is set as identity, household registration, family relationship, assistance application, assistance approval, fund disbursement, service performance, and asset verification. The source reliability level is configured from level 1 to level 5, with level 5 being the highest. Statutory registration and government-disbursed sources have higher priority than manually entered sources. After sorting, an ordered data sequence is formed with time as the main line and transaction type as the secondary line.

[0075] Based on ordered data sequences, position alignment is performed on data of the same business type from different sources. During position alignment, records with the same business type and whose business occurrence time falls within the same business cycle are placed in the same time position; relief application, approval, and disbursement records are aligned based on the same business number or the same relief period, service performance records are aligned based on service item, service time, and service location, and asset verification records are aligned based on verification time and asset category.

[0076] For overlapping or duplicate records, first determine if the business number, time location, object identifier, and business status are consistent. If they are consistent, retain the record with the higher source credibility level; if the credibility levels are the same, retain the record with the newer update time. If the field values ​​are not completely consistent but the business meanings can coexist, retain the multiple values ​​and mark their source; if the business meanings are mutually exclusive but no conflict flag is triggered, retain the primary record and write the remaining records into the difference field. After processing, a set of redundant data fragments is obtained.

[0077] The set of data segments, after redundancy removal, is concatenated in chronological order. During concatenation, each data segment retains its start time, end time, business type, business status, source identifier, and object association information. If there is a business succession relationship between adjacent segments, a succession relationship identifier is written in. For example, the rescue application segment is connected to the rescue approval segment, the rescue approval segment is connected to the fund disbursement segment, and the service dispatch segment is connected to the service fulfillment segment.

[0078] For segments that are continuous in time but different in business type, they are connected in the actual order of occurrence. For segments with gaps in time, no new business content is created; only the start and end times of the gap are recorded, along with the sources of the preceding and following segments. After concatenation, a data chain of civil affairs objects with a continuous time structure is formed. This data chain can express the continuous business processes of the same civil affairs object in terms of identity, household registration, assistance, funds, assets, and services.

[0079] The association path information, data source information, and two-way rebuttal verification result identifier are written into the civil affairs object data chain. The association path information includes the initial anchor point, composite identity index, candidate object subset, first candidate association set, second candidate association set, object set to be merged, and merging identifier, which is used to record the processing path of data from candidate association to merging into a chain; the data source information includes source identifier, collection batch, update time, source credibility level, and original business number, which is used to locate the source of each data segment; the two-way rebuttal verification result identifier includes static conflict mark status, dynamic conflict mark status, verification time, and processing batch. After writing, a traceable fused data structure is formed, which serves as the input for the subsequent generation of assistance eligibility, asset anomaly, fund disbursement, and service performance supervision results.

[0080] The intelligent supervision output module generates results for monitoring eligibility for assistance, asset anomalies, fund disbursement, and service performance based on the data chain of civil affairs beneficiaries.

[0081] In one implementation, the intelligent supervision output module takes the civil affairs object data chain as input, reads the identity fragment, household registration fragment, family relationship fragment, assistance fragment, asset verification fragment, fund disbursement fragment, service performance fragment, related path information, data source information and two-way reciprocal verification result identifier, and determines the complete supervision scope of the same civil affairs object according to the initial anchor point and merging identifier.

[0082] Data already marked with conflict isolation indicators will not be included in the automatic monitoring result generation process and will only be retained as a record pending verification. Data that has not triggered static or dynamic conflict indicators will be included in the monitoring and processing process for assistance eligibility, asset anomalies, fund disbursement, and service performance.

[0083] The results of the assistance eligibility monitoring are generated based on assistance segments, household registration segments, family relationship segments, and asset verification segments. During processing, the current valid assistance period is first determined, and then the family members, household registration status, assistance category, and asset verification results within that period are retrieved. When there are changes in family relationship, household registration relocation, death cancellation, assistance withdrawal, or new asset records during the assistance period, the corresponding items are written into the eligibility change item. The eligibility change item is compared with the assistance period based on the time of occurrence. If the change occurred before the assistance approval, it is marked as an admission verification item; if the change occurred during the assistance period, it is marked as a continuing eligibility verification item, thus forming the assistance eligibility monitoring result.

[0084] The asset anomaly monitoring results are generated based on asset verification segments and relief segments. During processing, the verification results of real estate, vehicles, business entities, and other assets are entered into the data chain according to the verification time and compared with the corresponding relief period. When the asset registration time falls within the preset verification window before and after the relief application, the asset-related information is recorded. The preset verification window can be set to 180 days before the application date to 180 days after approval, and the specific value can be configured between 90 and 365 days. If the asset-related information is inconsistent with the relief category requirements, an asset anomaly monitoring result is generated, recording the asset category, registration time, data source, and associated relief period.

[0085] The fund disbursement monitoring results are generated based on the fund disbursement segment and the relief approval segment. During processing, it is first verified whether the disbursement time is within the valid approval period, and then verified whether the recipient, subsidy item, receiving account identifier, and disbursement batch are consistent with the valid relief segment in the data chain. If there are disbursements before approval, disbursements after withdrawal, duplicate disbursements in the same period, or abnormal changes in the receiving account, a fund disbursement monitoring result is generated, and the anomaly type, batch involved, disbursement time, and source identifier are written.

[0086] Service performance supervision results are generated based on service performance segments, service dispatch segments, and recipient status segments. During processing, the service is checked against the dispatch content according to the service item, service time, service location, and confirmation result. Service performance supervision results are generated when there are instances of undispatched services, unfulfilled dispatches, service times earlier than dispatch times, service locations that do not match recipient status, or missing confirmation results. Finally, the eligibility for assistance, asset anomalies, fund disbursement, and service performance supervision results are written into the same regulatory record according to the civil affairs recipient data chain, along with associated path information and data source information, for subsequent query, review, and processing.

[0087] Example 2: To verify the effectiveness of this application in identifying legitimate primary key conflicts, reducing erroneous merging, and improving the accuracy of smart civil affairs supervision results, a set of anonymized sample data consistent with civil affairs business fields was constructed in a test environment. This sample data included multi-source records such as identity, household registration, assistance, assets, fund disbursement, and service performance. During testing, the traditional single-identity direct association method was used as a control method, while the processing method of this application, which employs a composite identity index and bidirectional reverse verification of static stability and dynamic coherence, was used as the method of this application. Both methods used the same input data, and the results of manual review were used as the comparison benchmark.

[0088] Table 1 shows the composition of the sample data used in this embodiment. In the aforementioned sample data, after manual verification, 43 sets of primary key legal conflicts were identified, all with the same initial anchor point but corresponding to different civil affairs entities. These conflicts primarily stemmed from duplicate historical ID numbers, reuse of temporary numbers, early manual entry errors, and placeholder numbers for special entities. Traditional methods of directly associating data with a single identity identifier, when processing this type of data, only merge data based on a unified identity field, failing to distinguish between different real entities under the same initial anchor point. This easily leads to the merging of relief records, asset records, fund disbursement records, and service records under the same entity name. The method in this application further introduces identity stability fingerprints and behavioral coherence fingerprints on top of the initial anchor point, performing bidirectional reciprocal verification through static and dynamic continuous features.

[0089] Table 2 shows the processing status of one group of primary key valid conflict samples. In the scenario shown in Table 2, if a single identity identifier is used for direct association, Subject A's low-income assistance record will be merged with Subject B's real estate record, potentially generating an abnormal asset result of "the low-income assistance recipient owns real estate," which could affect the assessment of eligibility for assistance. When processing this set of records, the method in this application first generates an identity stability fingerprint based on the household registration migration sequence, family relationship hierarchy, and assistance stage division results; then, it generates a behavioral coherence fingerprint based on the subsidy disbursement rhythm, service performance trajectory, and cross-domain business sequence. Because there are long-term parallel household registration segments with different administrative divisions under the same initial anchor point, and the fund disbursement location and service performance location exhibit incompatible trajectories within the same time period, static and dynamic conflict markers are triggered, generating a conflict isolation marker to prevent Subject A and Subject B from entering the same civil affairs object data chain.

[0090] Table 3 Comparison results between the comparative method and the method of this application in handling legitimate primary key conflicts. As shown in Table 3, under the same input data conditions, the direct association method using a single identity identifier could not identify legitimate primary key conflicts, and all 43 conflicting objects were directly merged. The method in this application uses identity stability fingerprints and behavioral coherence fingerprints for reverse verification, identifying 41 legitimate primary key conflicts and writing the corresponding object sets into the conflict isolation path, preventing them from entering the current data chain fusion. The two unidentified samples, after analysis, were found to be due to severe data gaps in the source data, i.e., a lack of sufficient household registration migration and service fulfillment segments. These can be addressed by supplementing the data source or through manual review. Because the conflicting objects were isolated, the number of incorrectly merged object groups decreased from 43 to 0, the number of incorrectly generated asset anomaly results decreased from 37 to 3, and the number of errors affecting eligibility for assistance decreased from 29 to 2.

[0091] Furthermore, the results of the fund disbursement and service performance supervision were compared, and the test results are shown in Table 4.

[0092] Table 4 Test Results Table 4 shows that the direct association method using a single identity generated 550 supervisory anomalies, of which 384 were confirmed as valid after manual review. The method in this application generated 416 supervisory anomalies, of which 377 were confirmed as valid after manual review. The number of valid anomalies confirmed by the two methods is similar, but the method in this application reduces the output of erroneous anomalies caused by primary key conflicts, incompatible trajectories, and static structure breaks. In other words, the method in this application does not simply reduce the number of supervisory results, but rather, while retaining the main valid anomalies, removes invalid anomalies caused by error merging from the supervisory results or transfers them to the conflict isolation path.

[0093] In this embodiment, the traceability of the civil affairs object data chain is also verified. For each supervision result generated by the method of this application, an initial anchor point, composite identity index, candidate object subset, set of objects to be merged, merge identifier, data source information, and two-way rebuttal verification result identifier are written. A traceability check of 100 supervision results shows that the original data source, collection batch, business occurrence time, and data segment involved in the merging can all be located. The comparison method only saves the merged object number and some source fields; when an erroneous merge occurs, it is difficult to directly distinguish whether the abnormal result originates from a relief segment, an asset segment, or a service segment.

[0094] Table 5 Data Link Traceability Verification Results In summary, this embodiment demonstrates that, in cases of primary key legal conflicts such as duplicate ID card numbers, reused temporary numbers, or placeholder numbers, the method of this application can construct a composite identity index through static and dynamic continuous features and perform bidirectional reverse verification before data fusion. For objects with the same initial anchor point but broken identity stability fingerprints or behavioral coherence fingerprints exhibiting incompatible trajectories, the method of this application generates conflict isolation markers and blocks the merging processing path, thereby preventing different civil affairs objects from being incorrectly merged into the same data chain. The rescue eligibility, asset anomaly, fund disbursement, and service performance supervision results generated based on this data chain can reduce false alarms caused by erroneous associations and retain the association path, source information, and verification results, facilitating subsequent querying, review, and handling.

[0095] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A smart civil affairs supervision and management system based on big data, characterized in that, include: The data acquisition module collects multi-source data on civil affairs recipients in terms of identity, household registration, assistance, assets, fund disbursement, and service performance. After field normalization, time calibration, and anomaly cleaning, a basic dataset is formed. The continuous feature construction module uses identity identifiers as the initial anchor point to construct static continuous features such as household registration migration, family relationship evolution, and assistance application history, and constructs dynamic continuous features such as subsidy receipt rhythm, service usage trajectory, and cross-domain business activity sequence. The composite identity index generation module generates an identity stability fingerprint based on static continuous features and a behavioral coherence fingerprint based on dynamic continuous features, and then combines the two with the initial anchor point to form a composite identity index. The candidate association module performs candidate association on multi-source data based on the composite identity index to obtain a set of objects to be merged. The conflict rebuttal verification module performs bidirectional rebuttal verification of the set of objects to be merged, including static stability and dynamic coherence. When the initial anchor points are the same but the identity stability fingerprint shows a break, or the behavioral coherence fingerprint shows an incompatible trajectory, it determines that there is a valid primary key conflict and generates a conflict isolation flag to prohibit merging. The data chain fusion module, when neither bidirectional rebuttal verification is triggered, merges the corresponding data into a civil affairs object data chain and writes the associated path, source, and verification result. The intelligent supervision output module generates results for monitoring eligibility for assistance, asset anomalies, fund disbursement, and service performance based on the data chain of civil affairs beneficiaries.

2. The intelligent civil affairs supervision and management system based on big data according to claim 1, characterized in that, The data acquisition module includes: Data from multiple sources, including civil affairs, household registration management, social assistance, asset verification, government subsidy disbursement, and service performance, is collected separately. Each data entry is then accompanied by the data source, collection time, transaction time, and data batch identifier. Unify the field names, data formats, and coding rules of identity, time, administrative division, relief category, and service category fields from different sources; The normalized data undergoes integrity, consistency, duplication, and validity checks. Data that cannot be corrected or has significant conflicts is written to the abnormal data area to form the basic dataset.

3. The intelligent civil affairs supervision and management system based on big data according to claim 1, characterized in that, The continuous feature construction module includes: Using the normalized identity field in the basic dataset as the initial anchor point, extract the household registration history field, family relationship field, and relief history field, and form a static event sequence according to the time of business occurrence; The system segments and compresses household registration migration records in static event sequences, expands family relationship change records hierarchically, and divides relief application records into stages to generate static continuous features. Using the fund disbursement time sequence field and service performance time sequence field as input, the subsidy disbursement records are mapped at periodic intervals, the service records are reorganized for trajectory coherence, and dynamic continuous features are formed by combining the order of cross-domain business occurrences.

4. The intelligent civil affairs supervision and management system based on big data according to claim 3, characterized in that, The continuous feature construction module includes the following when forming dynamic continuous features: Subsidy disbursement records are grouped according to subsidy items and receiving account identifiers, and the corresponding disbursement rhythm is determined when the interval between three or more consecutive disbursements falls within the range of 25 to 35 days per month, 80 to 100 days per quarter, or 350 to 380 days per year. Service records are sorted according to service time, service confirmation time, and service item registration order, and service locations are converted to a unified administrative division level; When adjacent services deviate from the established distribution rhythm by more than 20%, or when adjacent service locations are in different administrative divisions and the time interval is less than 1 day and lacks remote service or agency identification, the corresponding time period is marked as abnormal to obtain dynamic continuous characteristics.

5. The intelligent civil affairs supervision and management system based on big data according to claim 1, characterized in that, The composite identity index generation module includes: Structural compression and sequential encoding are performed on the household registration migration sequence, family relationship hierarchy and assistance stage division results in static continuous features. Key structural fragments with a duration of no less than 180 days or no more than 2 changes within 12 consecutive months are extracted to generate identity stability fingerprints. The subsidy disbursement rhythm, service fulfillment trajectory, and cross-domain business sequence in the dynamic continuous features are reconstructed and path-corrected. Behavioral segments that appear at least 3 times and whose intervals between adjacent segments deviate from the corresponding period range by no more than 20% are extracted to generate behavioral coherence fingerprints. The identity stability fingerprint and behavior coherence fingerprint are segmented and aligned according to the identity anchor segment, time location segment, structural content segment, source segment, and anomaly marker segment, and then written together with the initial anchor into the identifier structure to form a composite identity index.

6. The intelligent civil affairs supervision and management system based on big data according to claim 1, characterized in that, The candidate association module includes: Using the initial anchor point in the composite identity index as the grouping key, the multi-source data is initially aggregated to obtain several candidate object subsets with the same initial anchor point. In the candidate subset, the household registration segment, relationship segment, and rescue segment in the identity stability fingerprint are compared structurally. When at least two consecutive segments have the same order, the same segment type, and a time overlap ratio of 70% to 100%, the first candidate association set is formed. Perform sequential consistency verification on the behavioral coherence fingerprints in the first candidate association set, remove object pairs with broken behavioral trajectories or conflicting business sequences, and merge and aggregate the retained objects to obtain the set of objects to be merged.

7. The intelligent civil affairs supervision and management system based on big data according to claim 1, characterized in that, The conflict rebuttal verification module includes: Extract identity stability fingerprints and behavioral coherence fingerprints from the set of objects to be merged, and form a two-way verification sequence according to the initial anchor point, fingerprint version, update time, source layer and collection batch; Continuous counter-verification processing is performed on identity stability fingerprints. When household registration fragments overlap in different administrative regions for more than 30 days, or when there are mutually exclusive family relationship levels within the same time period, or when there is a withdrawal in the assistance stage before the disbursement and no record of reapplication, a static conflict marker is generated. For behavioral coherence fingerprints, trajectory reversal processing is performed. When adjacent business locations are across prefecture-level cities and the time difference is less than 1 day, or across provincial administrative regions and the time difference is less than 2 days, and there is a lack of remote processing, off-site settlement, or entrusted processing markers, dynamic conflict markers are generated.

8. The intelligent civil affairs supervision and management system based on big data according to claim 7, characterized in that, The conflict rebuttal verification module includes: When static or dynamic conflict markers exist, a conflict isolation marker is generated based on the initial anchor point, merge identifier, conflict fragment location, conflict type, and processing batch. Remove the set of objects with conflict isolation markers from the mergeable path, while retaining the original source, bidirectional check sequence, conflict markers, and decision criteria; When neither static nor dynamic conflict markers exist, the corresponding set of objects is marked as merging-ready, and the initial anchor point, merge identifier, bidirectional verification sequence, and verification result identifier are output.

9. The intelligent civil affairs supervision and management system based on big data according to claim 1, characterized in that, The data link fusion module includes: The set of objects to be merged that have not triggered conflict markers is taken as input. Data from each source is collected according to the initial anchor point, and an ordered data sequence is formed according to the business occurrence time, business type order, source trust level and collection batch. Align the positions of data of the same business type in the ordered data sequence, and determine the retention priority of duplicate records with the same business number, time position, object identifier and business status, and generate a set of data fragments without redundancy; The set of redundant data segments is concatenated in chronological order to form a civil affairs object data chain, and the associated path information, data source information and two-way reversal verification result identifier are written into the civil affairs object data chain.

10. A smart civil affairs supervision and management system based on big data as described in claim 1, characterized in that, The intelligent supervision output module includes: Read identity fragments, household registration fragments, family relationship fragments, relief fragments, asset verification fragments, fund disbursement fragments, and service performance fragments from the data chain of civil affairs beneficiaries, and determine the scope of supervision for the same civil affairs beneficiary based on the initial anchor point and the merge identifier; The results of supervision on eligibility for assistance are generated based on the assistance segment, household registration segment, family relationship segment, and asset verification segment, and the results of supervision on asset anomalies are generated based on the correspondence between asset registration time and assistance period. The system generates fund disbursement supervision results based on the correspondence between fund disbursement segments and relief approval segments, and generates service performance supervision results based on service performance segments, service dispatch segments, and recipient status segments.

Images