A building material data quality automatic checking method and system based on agent cooperation

Abstract

The application discloses a kind of based on agent cooperation's building material data quality automatic checking method and system, it is related to data management technical field, including following steps: step one: acquisition building material business flow data, generates building material check basic data set;Step two: execution standardization processing, establish object association index and generate building material standardization record set;Step three: configuration collaborative check node, generate check evidence proposition set;Step four: build standard constraint lattice, generate proposition node mapping table;Step five: establish check evidence coverage, generate overlapping record of coverage;Step six: execution consistency splicing, generate coverage conflict unit;Step seven: build improved Dubois-Prade check arbitration model, generate conflict quality transfer object;Step eight: generate building material data quality arbitration result.The application improves building material data anomaly positioning, relationship traceability and conflict arbitration capability.

Description

Technical Field

[0001] This invention relates to the field of data governance technology, and in particular to an automatic verification method and system for building material data quality based on agent collaboration. Background Technology

[0002] With the increasing demand for digital management of engineering construction and full-process traceability of building materials, the quality verification and anomaly location technologies for building materials business flow data have received widespread attention. Existing building materials data management platforms mainly rely on manual sampling, fixed rule matching, or single-table field verification to perform compliance judgments on heterogeneous records generated during procurement, warehousing, testing, and settlement. However, in practical applications, the following problems commonly exist: The inconsistent field representations, time sequences, and source identifiers of building material objects across different business stages make it difficult for existing standardized processing methods to establish stable object association indexes. This leads to record breaks and unclear verification scopes for the same building material object during cross-stage transfers. Furthermore, the lack of a unified standard constraint structure among different verification rules means that existing methods can typically only output field-level alerts when conflicts arise in field values, object relationships, and business sequences, making it difficult to pinpoint the connections between abnormal fields. Regarding consistency, conflict, and missing states among multi-source verification evidence, traditional evidence fusion methods often use ordinary union propositions to handle conflict quality, lacking constraint transfer mechanisms oriented towards building material standard levels and verification coverage. This results in insufficient interpretability of arbitration results and difficulty in tracing verification sources.

[0003] Therefore, how to provide an automatic verification method and system for building material data quality based on intelligent agent collaboration is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] One objective of this invention is to propose an automatic verification method and system for building material data quality based on agent collaboration. This invention utilizes collaborative verification nodes, standard constraint lattices, and an improved Dubois-Prade verification arbitration model to standardize, generate evidence propositions, perform consistency splicing, and arbitrate conflict quality in building material business flow data. It has the advantages of accurate anomaly location, traceable verification source, and refined conflict handling.

[0005] An automatic verification method for building material data quality based on intelligent agent collaboration according to an embodiment of the present invention includes the following steps: Step 1: Collect building materials business flow data and generate a basic dataset for building materials verification; Step 2: Perform normalization processing on the basic dataset for building material verification, establish an object association index according to the building material object identifier, and generate a standardized record set of building materials; Step 3: Configure the collaborative verification node based on the object association index, so that the collaborative verification node reads the standardized record set of building materials and generates a set of verification evidence propositions; Step 4: Construct a standard constraint grid based on the building material standard constraint relationship, map the set of verification evidence propositions to the proposition nodes in the standard constraint grid, and generate a proposition node mapping table; Step 5: Establish the verification evidence coverage range based on the data processing boundaries of the collaborative verification nodes, and generate overlapping records based on the overlapping relationship between the verification evidence coverage ranges; Step 6: Perform consistency splicing on the verification evidence proposition set based on the overlapping records and proposition node mapping table, and write the splicing failure positions into the overlapping conflict unit; Step 7: Construct an improved Dubois-Prade verification arbitration model, which replaces the ordinary union proposition with the least common superordinate proposition in the standard constraint lattice when fusing conflict evidence propositions, and limits the conflict quality transfer object by covering conflict units; Step 8: Generate building material data quality arbitration results based on the conflict quality transfer object, and output the abnormal field location, abnormal relationship location, and verification source identifier.

[0006] Optionally, step one specifically includes: Collect building materials business flow data, and extract building materials object identifier, flow link identifier, business entity identifier, business time identifier, data source identifier and quality association field from the building materials business flow data to form the original flow record set; Based on the data source identifier, the original flow record set is aggregated from the source, and based on the business time identifier, the source aggregation result is arranged in time sequence to generate a time sequence flow record set; Based on the building material object identifier, the time-series flow record set is merged into an object scope, and the merging result is written into the object record unit corresponding to the building material object identifier. Perform field validity checks and record consistency checks on the object record units, retain the records that pass the checks, and associate and organize the records that pass the checks according to the building material object identifier, circulation link identifier and business time identifier to generate the building material verification basic dataset.

[0007] Optionally, step two specifically involves: Read the verification-passed records from the building materials verification basic dataset, establish field conversion rules based on the data source identifier, convert the quality-related fields into standard field names, standard field types, and standard field value formats, and generate field-standardized records; Based on the business time identifier, the execution time benchmark of the standardized field records is unified; based on the flow step identifier, the execution step sequence of the standardized field records is organized to generate a standardized flow record set. Extract building material object identifiers from the standardized circulation record set, write standardized circulation records with the same building material object identifier into the same object index item, and establish a correspondence between the object index item and the circulation link identifier, business entity identifier, business time identifier and data source identifier to generate an object association index; Based on the object association index, the standardized flow record set is organized into objects. The results of the object association organization are written into the building material standardized record unit according to the building material object identifier. The building material standardized record unit is then collected to generate the building material standardized record set.

[0008] Optionally, step three specifically includes: Extract the correspondence between building material object identifiers, object index items and building material standardized record units from the object association index, and divide the building material standardized record units into verification ranges according to the circulation link identifier and business entity identifier to form verification task fragments; The calibration results are written to the node configuration record, and the node configuration record is configured to the collaborative verification node. The collaborative verification node performs standard field value comparison, object association comparison, and business flow order comparison on the verification task fragment based on the node configuration record, and generates node verification results; Verification evidence propositions are generated based on the consistency, conflict, and missing states in the node verification results. These propositions are then bound to building material object identifiers, collaborative verification nodes, and node configuration records. Finally, a set of verification evidence propositions is compiled.

[0009] Optionally, step four specifically involves: Extract standard field names, standard field types, standard field value formats, node verification results, and node configuration records from the building materials standard record set and the verification evidence proposition set. Generate building materials standard constraint relationships according to the direction of connection between standard field names and the field scope in the node configuration records. Write the constraint start point, constraint end point and constraint direction in the building material standard constraint relationship into the grid structure record, and connect the grid structure records according to the constraint direction to form a standard constraint grid; Convert the consistent, conflicting, and missing states in the validation evidence propositions into proposition identifiers, and write the proposition identifiers into proposition nodes in the standard constraint cell that match the standard field names; Establish a mapping relationship between the proposition node and the building material object identifier, collaborative verification node, node configuration record, and standard field name to generate a proposition node mapping table.

[0010] Optionally, step five specifically includes: Extract the field carrying range, business flow range and object association position from the node configuration record corresponding to the collaborative verification node. Generate field coverage boundary according to the field carrying range, generate process coverage boundary according to the business flow range, generate object coverage boundary according to the object association position, and write the field coverage boundary, process coverage boundary and object coverage boundary into the data processing boundary. Based on the data processing boundary, the verification evidence propositions in the verification evidence proposition set are covered and labeled. The coverage labeling results are then bound to the building material object identifier and collaborative verification node to form the verification evidence coverage range. Based on the proposition node mapping table, locate the proposition node corresponding to the scope of verification evidence coverage, and perform overlap comparison on the field coverage boundary, link coverage boundary and object coverage boundary between the scopes of verification evidence coverage to generate the coverage boundary overlap result; The overlapping results of the coverage boundaries are associated with the collaborative verification nodes, verification evidence propositions, proposition nodes, and building material object identifiers to generate an overlapping record.

[0011] Optionally, step six specifically includes: Extract the overlapping results of the coverage boundary, the verification evidence propositions, the proposition nodes and the building material object identifiers from the overlapping records. Based on the proposition node mapping table, perform object range matching on the proposition nodes to form a set of overlapping proposition nodes under the same building material object identifier. Based on the constraint direction of the overlapping proposition node set in the standard constraint lattice, the consistent, conflicting, and missing states in the verification evidence propositions are aligned according to the position of the proposition nodes to generate proposition alignment records. Based on the proposition alignment record, perform consistency splicing on the set of verification evidence propositions. Write the splicing result that satisfies the constraint direction of the consistent state into the splicing pass record, and write the splicing result that does not satisfy the constraint direction of the conflicting state and the missing state into the splicing failure record. Extract the building material object identifier, collaborative verification node, proposition node, verification evidence proposition, and coverage boundary overlap result from the splicing failure record, and write the extracted results to the splicing failure location to generate coverage conflict unit.

[0012] Optionally, step seven specifically includes: Extract verification evidence propositions from the set of verification evidence propositions, determine the proposition nodes corresponding to the verification evidence propositions based on the proposition node mapping table, and convert the verification evidence propositions into basic probability assignment records; Based on the basic probability assignment record, Dubois-Prade conflict identification is performed. Verification evidence propositions with empty intersection of proposition nodes are identified as conflict evidence propositions, and conflict evidence proposition pairs are formed according to the building material object identifier. Locate the proposition nodes corresponding to the conflict evidence proposition pairs in the standard constraint lattice, search for common bearing nodes along the constraint direction in the standard constraint lattice, and determine the common bearing node with the smallest hierarchical distance as the least common superordinate proposition; The conflict quality transfer terms assigned to ordinary union propositions in the Dubois-Prade combination rules are replaced with conflict quality transfer terms corresponding to the least common superordinate proposition, and the conflict quality transfer objects are limited according to the covering conflict unit to generate an improved Dubois-Prade verification arbitration model.

[0013] Optionally, step eight specifically includes: Extract conflict quality transfer objects, basic probability assignment records, least common superordinate propositions, and covering conflict units from the improved Dubois-Prade verification arbitration model, establish association relationships according to building material object identifiers, and generate arbitration input records; Based on the arbitration input record, the proposition quality and conflict quality in the conflict quality transfer object of the basic probability assignment record are written into the least common superordinate proposition to generate the proposition arbitration quality record; Based on the proposition node mapping table and the coverage conflict unit, reverse positioning is performed on the proposition arbitration quality record. The standard field name associated with the proposition node is determined as the abnormal field positioning, and the overlap result of the constraint direction and coverage boundary in the standard constraint cell is determined as the abnormal relationship positioning. The collaborative verification node is determined as the verification source identifier. The building material object identifier, proposition arbitration quality record, abnormal field location, abnormal relationship location, and verification source identifier are associated and arranged to generate the building material data quality arbitration result.

[0014] An automatic verification system for building material data quality based on intelligent agent collaboration according to an embodiment of the present invention includes: The data acquisition module is used to collect data on the flow of building materials business and generate a basic dataset for building materials verification. The standardization processing module is used to perform standardization processing on the basic dataset for building material verification, establish object association indexes, and generate a standardized record set for building materials. The collaborative verification module is used to configure collaborative verification nodes based on object association indexes and generate a set of verification evidence propositions. The constraint mapping module is used to construct a standard constraint lattice, map the set of validation evidence propositions to proposition nodes, and generate a proposition node mapping table. The coverage analysis module is used to establish the coverage range of verification evidence based on the data processing boundary and generate overlapping coverage records. The splicing conflict module is used to perform consistent splicing based on the overlapping records and the proposition node mapping table to generate overlapping conflict units. The arbitration modeling module is used to build an improved Dubois-Prade verification arbitration model and generate conflict quality transfer objects. The results output module is used to generate building material data quality arbitration results based on the conflict quality transfer object, and output the abnormal field location, abnormal relationship location, and verification source identifier.

[0015] The beneficial effects of this invention are: This invention standardizes building material business flow data and establishes object association indexes based on building material object identifiers. This allows building material records from different flow stages, data sources, and business entities to be grouped into the same building material object scope, thereby reducing record breaks caused by inconsistent field expressions, inconsistent time sequences, and scattered source identifiers. By configuring collaborative verification nodes and generating verification evidence proposition sets, this invention can form structured verification evidence by comparing field values, object association relationships, and business flow sequences. This makes building material data quality verification no longer limited to single-field alarms, improving the automatic verification capability of building material data across stages and sources.

[0016] This invention further establishes constraint relationships between verification evidence through standard constraint lattices, proposition node mapping tables, verification evidence coverage, and conflict coverage units. During conflict evidence proposition fusion, an improved Dubois-Prade verification arbitration model replaces ordinary union propositions with least common superordinate propositions. Furthermore, conflict coverage units limit the objects of conflict quality transfer, ensuring that conflict quality falls back to the specific receiving level within the building material standard constraint relationships. The resulting building material data quality arbitration results can simultaneously output anomaly field location, anomaly relationship location, and verification source identifier, improving conflict handling accuracy, anomaly attribution clarity, and verification result traceability. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of an automatic verification method for building material data quality based on intelligent agent collaboration proposed in this invention. Figure 2 This is a schematic diagram of the standard constraint lattice construction for an automatic verification method for building material data quality based on intelligent agent collaboration proposed in this invention. Figure 3 This is a schematic diagram illustrating the conflict quality transfer of a building material data quality automatic verification method based on agent collaboration proposed in this invention. Detailed Implementation

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

[0019] refer to Figures 1-3 An automatic verification method for building material data quality based on agent collaboration includes the following steps: Step 1: Collect building materials business flow data and generate a basic dataset for building materials verification; Step 2: Perform normalization processing on the basic dataset for building material verification, establish an object association index according to the building material object identifier, and generate a standardized record set of building materials; Step 3: Configure the collaborative verification node based on the object association index, so that the collaborative verification node reads the standardized record set of building materials and generates a set of verification evidence propositions; Step 4: Construct a standard constraint grid based on the building material standard constraint relationship, map the set of verification evidence propositions to the proposition nodes in the standard constraint grid, and generate a proposition node mapping table; Step 5: Establish the verification evidence coverage range based on the data processing boundaries of the collaborative verification nodes, and generate overlapping records based on the overlapping relationship between the verification evidence coverage ranges; Step 6: Perform consistency splicing on the verification evidence proposition set based on the overlapping records and proposition node mapping table, and write the splicing failure positions into the overlapping conflict unit; Step 7: Construct an improved Dubois-Prade verification arbitration model, which replaces the ordinary union proposition with the least common superordinate proposition in the standard constraint lattice when fusing conflict evidence propositions, and limits the conflict quality transfer object by covering conflict units; Step 8: Generate building material data quality arbitration results based on the conflict quality transfer object, and output the abnormal field location, abnormal relationship location, and verification source identifier.

[0020] In this embodiment, step one specifically includes: Collect building materials business flow data, take each business record in the building materials business flow data as the collection object, read the building materials object identifier, flow link identifier, business entity identifier, business time identifier, data source identifier and quality association field corresponding to the collection object, and write the reading results into the original record row in the collection order, and collect the original record rows to form the original flow record set; Based on the data source identifier, the original flow record set is aggregated by source. The original record lines with the same data source identifier are grouped into the same source group. Then, based on the business time identifier, the original record lines in each source group are sorted by time. The sorted source groups are connected to form a time-series flow record set. Based on the building material object identifier, the time-series flow record set is merged into object scopes. Time-series flow records with the same building material object identifier are located in the same object scope, and the time-series flow records in the same object scope are written into the object record unit corresponding to the building material object identifier. Perform field validity checks and record consistency checks on the object record unit. Field validity checks are used to determine whether the building material object identifier, circulation link identifier, business entity identifier, business time identifier, data source identifier, and quality association fields meet the record generation requirements. Record consistency checks are used to determine whether the circulation link identifier and business time identifier within the object record unit maintain the same order. Records that pass the checks are retained, and the records that pass the checks are associated and organized according to the building material object identifier, circulation link identifier, and business time identifier to generate the building material verification basic dataset.

[0021] In this embodiment, step two specifically includes: Read the verification-passed records from the building materials verification basic dataset, call the field conversion rules according to the data source identifier, and record the correspondence between quality-related fields and standard field names, standard field types and standard field value formats. Write the field names in the quality-related fields into standard field names, write the data types corresponding to the field values ​​in the quality-related fields into standard field types, and write the expression forms of the field values ​​in the quality-related fields into standard field value formats to form standardized field records. Based on the business time identifier, the time base of the field normalization record is unified. The business time identifier in the field normalization record is converted into the time position under the same time base, and the field normalization record is arranged according to the time position. Based on the flow link identifier, the flow link sequence of the arranged field normalization record is sorted, so that the field normalization record has both time position relationship and flow link sequence relationship, and a normalized flow record set is generated. Extract building material object identifiers from the standardized circulation record set, write standardized circulation records with consistent building material object identifiers into the same object index item, record the binding relationship between the building material object identifier and the corresponding standardized circulation record in the object index item, and establish corresponding relationships between the object index item and the circulation link identifier, business entity identifier, business time identifier and data source identifier respectively, so that the object index item can locate the link position, entity source, time position and data source of the building material object in the business circulation process, and generate an object association index; Based on the object association index, the standardized circulation record set is organized by object association. The standardized circulation record corresponding to the object index item is read according to the building material object identifier. Then, the read standardized circulation record is associated and arranged according to the circulation link identifier and business time identifier. The association and arrangement results are written into the building material standardized record unit. The building material standardized record unit is collected to generate the building material standardized record set.

[0022] In this embodiment, step three specifically includes: Extract the correspondence between building material object identifiers, object index items and building material standardized record units from the object association index. First, determine the scope of objects to be verified by building material object identifiers. Then, locate the record position of building material standardized record units in the building material business flow by object index items. Divide the scope of the process according to the flow link identifier and the scope of the main body according to the business entity identifier. The intersection of the scope of the process and the scope of the main body forms the verification scope dividing boundary. Separate the verification task fragment from the building material standardized record unit according to the verification scope dividing boundary. The field carrying range, business flow range, and object association position are defined for the verification task segment. The field carrying range is determined by the standard field name, standard field type, and standard field value format contained in the verification task segment. The business flow range is determined by the flow link identifier and business time identifier. The object association position is determined by the building material object identifier and object index item. The field carrying range, business flow range, and object association position are written into the node configuration record, and the node configuration record is configured to the collaborative verification node. The collaborative verification node performs standard field value comparison, object association comparison, and business flow order comparison on the verification task fragment based on the node configuration record. The standard field value comparison is used to determine whether the standard field value format within the same field range matches. The object association comparison is used to determine whether the binding relationship between the building material object identifier and the object index item is consistent. The business flow order comparison is used to determine whether the arrangement relationship between the flow link identifier and the business time identifier is consistent. The comparison results are collected to generate the node verification result. Verification evidence propositions are generated based on the consistency, conflict, and missing states in the node verification results. Consistency states correspond to affirmative propositions formed by comparing verification task segments. Conflict states correspond to conflicting propositions formed by differences in values, associations, and sequences in verification task segments. Missing states correspond to missing propositions formed by missing fields, relationships, and times in verification task segments. Verification evidence propositions are then bound to building material object identifiers, collaborative verification nodes, and node configuration records. Verification evidence proposition sets are then compiled.

[0023] In this embodiment, step four specifically includes: The standard field names, standard field types, and standard field value formats are extracted from the building materials standardization record set. The node verification results and node configuration records are extracted from the verification evidence proposition set. The standard field names participating in the constraint generation are determined according to the field carrying range in the node configuration record. Then, the start and end connection relationships between fields are established according to the direction of acceptance in the building materials business flow according to the standard field names. The standard field types and standard field value formats are used as the value restriction conditions of the connection relationship to generate building materials standard constraint relationships. Write the starting standard field name in the building material standard constraint relationship to the constraint start point, write the receiving standard field name in the building material standard constraint relationship to the constraint end point, write the connection direction from the starting standard field name to the receiving standard field name to the constraint direction, write the constraint start point, constraint end point and constraint direction to the same cell structure record, and perform a directed connection on the cell structure record according to the constraint direction to form a standard constraint cell. The consistent state in the verification evidence proposition is converted into a proposition identifier indicating that the comparison has passed; the conflict state in the verification evidence proposition is converted into a proposition identifier indicating that the comparison is inconsistent; the missing state in the verification evidence proposition is converted into a proposition identifier indicating that the field is missing; then, the proposition node is located in the standard constraint cell according to the standard field name associated with the proposition identifier, and the proposition identifier is written into the located proposition node. Establish an object mapping relationship between the proposition node and the building material object identifier, establish a source mapping relationship between the proposition node and the collaborative verification node, establish a configuration mapping relationship between the proposition node and the node configuration record, establish a field mapping relationship between the proposition node and the standard field name, and write the object mapping relationship, source mapping relationship, configuration mapping relationship and field mapping relationship into the same mapping record, and collect the mapping records to generate the proposition node mapping table.

[0024] In this embodiment, step five specifically includes: Extract the field carrying range, business flow range, and object association location from the node configuration record corresponding to the collaborative verification node. Determine the start and end range of the field according to the standard field name, standard field type, and standard field value format recorded in the field carrying range and generate the field coverage boundary. Determine the start and end range of the process according to the flow process identifier and business time identifier recorded in the business flow range and generate the process coverage boundary. Determine the start and end range of the object according to the building material object identifier and object index item recorded in the object association location and generate the object coverage boundary. Write the field coverage boundary, process coverage boundary, and object coverage boundary into the data processing boundary. Based on the data processing boundary, the verification evidence propositions in the verification evidence proposition set are covered and labeled. The standard field names associated with the verification evidence propositions are matched with the field coverage boundary. The flow link identifiers associated with the verification evidence propositions are matched with the link coverage boundary. The building material object identifiers associated with the verification evidence propositions are matched with the object coverage boundary. The matching results are then bound to the building material object identifiers and collaborative verification nodes to form the verification evidence coverage range. Based on the proposition node mapping table, locate the proposition node corresponding to the verification evidence coverage. Extract the correspondence between the verification evidence coverage according to the node position of the proposition node in the standard constraint cell. Then, perform overlap comparison on the overlapping field range between field coverage boundaries, the overlapping flow range between link coverage boundaries, and the overlapping object range between object coverage boundaries to generate coverage boundary overlap results. The overlapping results of the coverage boundary are associated and organized with the collaborative verification nodes, verification evidence propositions, proposition nodes, and building material object identifiers. This allows the overlapping results of the same coverage boundary to record the collaborative verification nodes, verification evidence propositions, corresponding proposition nodes, and corresponding building material object identifiers that participated in the overlap. The associated and organized results are then used to generate the overlapping coverage record.

[0025] In this embodiment, step six specifically includes: Extract the overlapping results of the coverage boundary, verify the evidence propositions, proposition nodes and building material object identifiers from the overlapping records. According to the building material object identifiers, the overlapping records are classified into the same object range. Then, based on the proposition node mapping table, the mapping relationship between the proposition nodes and the building material object identifiers is checked. Proposition nodes with consistent mapping relationships are classified into the overlapping proposition node set under the same building material object identifier. Based on the node positions of the overlapping proposition node set in the standard constraint grid, the constraint direction is read, the proposition node arrangement position is determined according to the constraint direction, and then the consistent state, conflict state and missing state in the verification evidence proposition are written into the corresponding proposition node arrangement position, so that each proposition node arrangement position forms a state mark, and the state marks are collected to generate a proposition alignment record. Based on the proposition alignment record, a consistent splicing is performed on the set of verification evidence propositions. The status markers of the arrangement positions of adjacent proposition nodes are compared sequentially along the constraint direction in the standard constraint grid. The splicing result that maintains the continuity of the constraint direction in the consistent state is written into the splicing pass record. The splicing result that breaks the continuity of the constraint direction in the conflict state and interrupts the continuity of the constraint direction in the missing state is written into the splicing failure record. Extract building material object identifiers, collaborative verification nodes, proposition nodes, verification evidence propositions, and coverage boundary overlap results from the splicing failure records. Write the extracted results into the splicing failure position according to the position of the proposition nodes in the splicing failure records, and bind the splicing failure position with the corresponding coverage boundary overlap results to generate coverage conflict units.

[0026] In this embodiment, step seven specifically includes: Extract verification evidence propositions from the set of verification evidence propositions, find the proposition nodes corresponding to the verification evidence propositions based on the proposition node mapping table, take the consistent state, conflict state and missing state in the verification evidence propositions as basic probability assignment objects respectively, and write the mapping relationship between the verification evidence propositions and proposition nodes into the same assignment record, so that the basic probability assignment record simultaneously records the building material object identifier, verification evidence proposition, proposition node and proposition state. Dubois-Prade conflict identification is performed based on the basic probability assignment records. First, the basic probability assignment records are collected according to the building material object identifier. Then, it is compared whether there is an intersection between the proposition nodes under the same building material object identifier. The verification evidence propositions whose proposition nodes have no intersection and whose proposition states cannot be aligned are identified as conflict evidence propositions. The conflict evidence propositions are then grouped into pairs according to the building material object identifier to form conflict evidence proposition pairs. Locate the proposition nodes corresponding to the conflicting evidence proposition pair in the standard constraint grid. Track the connecting nodes that the two proposition nodes in the conflicting evidence proposition pair can reach along the constraint direction in the standard constraint grid. Then, compare the node overlap of the connecting nodes corresponding to the two proposition nodes. The common connecting node that simultaneously connects to the two proposition nodes and has the smallest sum of hierarchical distances relative to the two proposition nodes is determined as the least common superordinate proposition. The conflict quality transfer terms calculated for conflict evidence propositions in the Dubois-Prade combination rules are replaced with transfer objects. The bearing position corresponding to the ordinary union proposition is replaced with the bearing position corresponding to the least common superior proposition. Then, the bearing position is limited by the splicing failure position in the covering conflict unit, the overlapping result of the covering boundary, the proposition node and the building material object identifier. This ensures that the conflict quality transfer terms are only written to the least common superior proposition limited by the covering conflict unit, thus generating an improved Dubois-Prade verification arbitration model.

[0027] This invention structurally improves the Dubois-Prade verification arbitration model. It transforms verification evidence propositions into basic probability assignment records that simultaneously record building material object identifiers, verification evidence propositions, proposition nodes, and proposition states via a proposition node mapping table. Then, conflict identification is performed using building material object identifiers as the aggregation scope. This ensures that conflict judgment no longer relies on simple synthesis of evidence quality values, but instead combines two conditions—empty intersection of proposition nodes and inconsistent proposition states—to determine conflicting evidence proposition pairs. Furthermore, this invention traces the common receiving nodes of corresponding proposition nodes in the standard constraint lattice along the constraint direction of the conflicting evidence proposition pairs, and identifies the common receiving node with the smallest sum of hierarchical distances as the least common superior proposition. This replaces the conflict quality bearing position in the traditional Dubois-Prade combination rules, which is oriented towards ordinary union propositions, with the least common superior proposition bearing position, which has the meaning of building material standard constraints. Conflict quality is no longer transferred to a broad union proposition, but instead falls back to a standard constraint node that can express the level of abnormality. This invention also utilizes the splicing failure location, overlap result of the coverage boundary, proposition node, and building material object identifier in the covering conflict unit to limit the conflict quality transfer item, ensuring that conflict quality can only be written into the least common superior proposition limited by the covering conflict unit. This avoids repeated alarms from different collaborative verification nodes for the same field conflict, improving the constraint and interpretability of conflict evidence fusion. Through the above improvements, this invention can upgrade the building material data quality verification result from a simple field alarm to the synchronous output of abnormal field location, abnormal relationship location, and verification source identifier. It can clearly identify whether the conflict arises from field values, object association relationships, or business flow order, improving the accuracy of cross-stage building material record conflict arbitration, reducing the workload of manual review, and enhancing the business traceability and explanatory power of the building material data quality arbitration result.

[0028] In this embodiment, step eight specifically includes: The conflict quality transfer objects, basic probability assignment records, least common superior propositions, and covering conflict units are extracted from the improved Dubois-Prade verification arbitration model. The extracted results are then aggregated according to the building material object identifier. The conflict quality transfer objects, basic probability assignment records, least common superior propositions, and covering conflict units under the same building material object identifier are written into the same arbitration record position to establish the association between the arbitration record positions and generate arbitration input records. Based on the arbitration input record, read the proposition quality from the basic probability assignment record and read the conflict quality from the conflict quality transfer object. Merge the proposition quality and conflict quality according to the building material object identifier and the least common superior proposition. Write the merged quality value into the quality field corresponding to the least common superior proposition to generate the proposition arbitration quality record. Based on the proposition node mapping table and the coverage conflict unit, reverse positioning is performed on the proposition arbitration quality record. First, the proposition node associated with the proposition arbitration quality record is determined through the proposition node mapping table. Then, the standard field name is located by the proposition node and anomaly field location is formed. The position of the conflicting field is located by the overlap result of the constraint direction and the coverage boundary in the standard constraint cell and anomaly relationship location is formed. The verification source identifier is formed by the collaborative verification node in the coverage conflict unit. The building material object identifier, the proposition arbitration quality record, the abnormal field location, the abnormal relationship location, and the verification source identifier are written into the same arbitration result record. The arbitration result record is then collected and arranged in order according to the building material object identifier to generate the building material data quality arbitration result.

[0029] An automatic verification system for building material data quality based on intelligent agent collaboration includes: The data acquisition module is used to collect data on the flow of building materials business and generate a basic dataset for building materials verification. The standardization processing module is used to perform standardization processing on the basic dataset for building material verification, establish object association indexes, and generate a standardized record set for building materials. The collaborative verification module is used to configure collaborative verification nodes based on object association indexes and generate a set of verification evidence propositions. The constraint mapping module is used to construct a standard constraint lattice, map the set of validation evidence propositions to proposition nodes, and generate a proposition node mapping table. The coverage analysis module is used to establish the coverage range of verification evidence based on the data processing boundary and generate overlapping coverage records. The splicing conflict module is used to perform consistent splicing based on the overlapping records and the proposition node mapping table to generate overlapping conflict units. The arbitration modeling module is used to build an improved Dubois-Prade verification arbitration model and generate conflict quality transfer objects. The results output module is used to generate building material data quality arbitration results based on the conflict quality transfer object, and output the abnormal field location, abnormal relationship location, and verification source identifier.

[0030] Example 1: To verify the feasibility of this invention in implementation, it was applied to a data quality verification scenario for centralized procurement and on-site acceptance of building materials. The business data covered various stages including procurement registration, on-site acceptance, warehousing, testing reports, settlement review, and supplier files. The data to be verified included building material object identifiers, circulation stage identifiers, business entity identifiers, business time identifiers, data source identifiers, and quality-related fields. Before implementation, the business system primarily used a traditional rule-based threshold verification scheme as a comparison method. This scheme established fixed verification rules based on field names, judged for missing fields in a single table, incorrect field formats, and excessive values, and directly output field alarms when anomalies were detected. This scheme did not establish object-related indexes, construct standard constraint cells, generate verification evidence propositions, or use the Dubois-Prade verification arbitration model to handle conflicting evidence propositions. Because the same building material object may have inconsistent field naming, inconsistent batch identification, and unstable association between test report number and warehousing record in different business processes, traditional rule threshold verification schemes can detect some field errors, but it is difficult to explain which link between procurement, warehousing, testing and settlement the anomaly occurred in, and it is also difficult to distinguish the situation where the same anomaly is repeatedly triggered by different rules.

[0031] In the application of this invention, the system first collects building material business flow data and generates a basic dataset for building material verification. The basic dataset is then standardized, and an object association index is established based on the building material object identifier. This allows records of the same building material object in different flow stages to be merged into the corresponding object index item, further forming a standardized building material record set. Subsequently, the system configures collaborative verification nodes based on the object association index, enabling these nodes to generate a set of verification evidence propositions around the standardized building material record set. For building material objects such as steel bars, cement, waterproof membranes, and insulation boards, the collaborative verification nodes compare field values, object association relationships, and business flow order, respectively, and generate verification evidence propositions corresponding to consistent, conflicting, and missing states based on the comparison results. The system then constructs a standard constraint grid based on the building material standard constraint relationships, maps the set of verification evidence propositions to proposition nodes in the standard constraint grid, and generates a proposition node mapping table, allowing field anomalies to enter the standard constraint structure for association judgment.

[0032] During the coverage analysis phase, the system establishes the coverage scope of verification evidence based on the data processing boundaries of the collaborative verification nodes, and generates coverage overlap records based on the overlap relationships between the verification evidence coverage scopes. For a set of overlapping proposition nodes under the same building material object, the system performs consistent splicing based on the coverage overlap records and the proposition node mapping table, and writes the splicing failure positions into the coverage conflict unit. For example, in a batch of steel bar records, the specification fields of the purchase registration and warehousing records can be spliced ​​consistently, but the strength grade in the test report does not match the standard field value format. The system locates this conflict as an abnormal relationship between the standard field value and the test report. In a batch of cement records, the test report number exists, but the corresponding object association position is missing in the warehousing process. The system locates this conflict as an abnormal object association between the warehousing record and the test report. Finally, the system constructs an improved Dubois-Prade verification arbitration model, which replaces the ordinary union proposition with the least common superior proposition in the standard constraint lattice when fusing conflict evidence propositions, and limits the conflict quality transfer object with the coverage conflict unit, outputting the abnormal field location, abnormal relationship location, and verification source identifier.

[0033] To ensure consistency in the comparison criteria, the same building materials business flow data was used as input during the verification process. Both the traditional rule-based threshold verification scheme and the system of this invention processed the same batch of 48,260 business records, with manual review conclusions serving as the result verification benchmark. The traditional rule-based threshold verification scheme only outputs field alarms based on preset field rules, while the system of this invention completes object association, evidence proposition generation, consistency concatenation, and conflict quality arbitration under the same input conditions. The verification results of the two schemes are shown in Table 1: Table 1 Comparison of Building Material Data Verification Results

[0034] As shown in Table 1, the traditional rule-based threshold verification scheme can detect anomalies at the field level with the same number of records. However, due to the lack of object association indexes and standard constraint cells, the number of successfully associated building material objects is low, the accuracy rate of anomaly relationship location is only 54.8%, and the number of duplicate alarms reaches 913. The system of this invention uses object association indexes to group cross-stage records into the same building material object range, uses standard constraint cells and proposition node mapping tables to map verification evidence propositions to the standard constraint structure, covers the failed splicing positions of conflict unit records, and uses the improved Dubois-Prade verification arbitration model to transfer conflict quality to the conflict quality transfer object corresponding to the least common superior proposition. Therefore, the number of successfully associated building material objects increases to 9241, the accuracy rate of anomaly relationship location increases to 89.7%, the number of duplicate alarms decreases to 276, and the time for manual review of a single batch is reduced from 6.4 hours to 2.1 hours. The above results show that the present invention can solve the problems of fragmented object records, difficulty in locating cross-stage conflicts, and difficulty in tracing the source of verification in building materials business flow data, and can improve the accuracy and interpretability of building materials data quality verification results.

[0035] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for automatic verification of building material data quality based on agent collaboration, characterized in that, Includes the following steps: Step 1: Collect building materials business flow data and generate a basic dataset for building materials verification; Step 2: Perform normalization processing on the basic dataset for building material verification, establish an object association index according to the building material object identifier, and generate a standardized record set of building materials; Step 3: Configure the collaborative verification node based on the object association index, so that the collaborative verification node reads the standardized record set of building materials and generates a set of verification evidence propositions; Step 4: Construct a standard constraint grid based on the building material standard constraint relationship, map the set of verification evidence propositions to the proposition nodes in the standard constraint grid, and generate a proposition node mapping table; Step 5: Establish the verification evidence coverage range based on the data processing boundaries of the collaborative verification nodes, and generate overlapping records based on the overlapping relationship between the verification evidence coverage ranges; Step 6: Perform consistency splicing on the verification evidence proposition set based on the overlapping records and proposition node mapping table, and write the splicing failure positions into the overlapping conflict unit; Step 7: Construct an improved Dubois-Prade verification arbitration model, which replaces the ordinary union proposition with the least common superordinate proposition in the standard constraint lattice when fusing conflict evidence propositions, and limits the conflict quality transfer object by covering conflict units; Step 8: Generate building material data quality arbitration results based on the conflict quality transfer object, and output the abnormal field location, abnormal relationship location, and verification source identifier.

2. The automatic verification method for building material data quality based on intelligent agent collaboration according to claim 1, characterized in that, Step one specifically involves: Collect building materials business flow data, and extract building materials object identifier, flow link identifier, business entity identifier, business time identifier, data source identifier and quality association field from the building materials business flow data to form the original flow record set; Based on the data source identifier, the original flow record set is aggregated from the source, and based on the business time identifier, the source aggregation result is arranged in time sequence to generate a time sequence flow record set; Based on the building material object identifier, the time-series flow record set is merged into an object scope, and the merging result is written into the object record unit corresponding to the building material object identifier. Perform field validity checks and record consistency checks on the object record units, retain the records that pass the checks, and associate and organize the records that pass the checks according to the building material object identifier, circulation link identifier and business time identifier to generate the building material verification basic dataset.

3. The automatic verification method for building material data quality based on agent collaboration according to claim 1, characterized in that, Step two specifically involves: Read the verification-passed records from the building materials verification basic dataset, establish field conversion rules based on the data source identifier, convert the quality-related fields into standard field names, standard field types, and standard field value formats, and generate field-standardized records; Based on the business time identifier, the execution time benchmark of the standardized field records is unified; based on the flow step identifier, the execution step sequence of the standardized field records is organized to generate a standardized flow record set. Extract building material object identifiers from the standardized circulation record set, write standardized circulation records with the same building material object identifier into the same object index item, and establish a correspondence between the object index item and the circulation link identifier, business entity identifier, business time identifier and data source identifier to generate an object association index; Based on the object association index, the standardized flow record set is organized into objects. The results of the object association organization are written into the building material standardized record unit according to the building material object identifier. The building material standardized record unit is then collected to generate the building material standardized record set.

4. The automatic verification method for building material data quality based on agent collaboration according to claim 1, characterized in that, Step three specifically involves: Extract the correspondence between building material object identifiers, object index items and building material standardized record units from the object association index, and divide the building material standardized record units into verification ranges according to the circulation link identifier and business entity identifier to form verification task fragments; The calibration results are written to the node configuration record, and the node configuration record is configured to the collaborative verification node. The collaborative verification node performs standard field value comparison, object association comparison, and business flow order comparison on the verification task fragment based on the node configuration record, and generates node verification results; Verification evidence propositions are generated based on the consistency, conflict, and missing states in the node verification results. These propositions are then bound to building material object identifiers, collaborative verification nodes, and node configuration records. Finally, a set of verification evidence propositions is compiled.

5. The automatic verification method for building material data quality based on intelligent agent collaboration according to claim 1, characterized in that, Step four specifically involves: Extract standard field names, standard field types, standard field value formats, node verification results, and node configuration records from the building materials standard record set and the verification evidence proposition set. Generate building materials standard constraint relationships according to the direction of connection between standard field names and the field scope in the node configuration records. Write the constraint start point, constraint end point and constraint direction in the building material standard constraint relationship into the grid structure record, and connect the grid structure records according to the constraint direction to form a standard constraint grid; Convert the consistent, conflicting, and missing states in the validation evidence propositions into proposition identifiers, and write the proposition identifiers into proposition nodes in the standard constraint cell that match the standard field names; Establish a mapping relationship between the proposition node and the building material object identifier, collaborative verification node, node configuration record, and standard field name to generate a proposition node mapping table.

6. The automatic verification method for building material data quality based on intelligent agent collaboration according to claim 1, characterized in that, Step five specifically involves: Extract the field carrying range, business flow range and object association position from the node configuration record corresponding to the collaborative verification node. Generate field coverage boundary according to the field carrying range, generate process coverage boundary according to the business flow range, generate object coverage boundary according to the object association position, and write the field coverage boundary, process coverage boundary and object coverage boundary into the data processing boundary. Based on the data processing boundary, the verification evidence propositions in the verification evidence proposition set are covered and labeled. The coverage labeling results are then bound to the building material object identifier and collaborative verification node to form the verification evidence coverage range. Based on the proposition node mapping table, locate the proposition node corresponding to the scope of verification evidence coverage, and perform overlap comparison on the field coverage boundary, link coverage boundary and object coverage boundary between the scopes of verification evidence coverage to generate the coverage boundary overlap result; The overlapping results of the coverage boundaries are associated with the collaborative verification nodes, verification evidence propositions, proposition nodes, and building material object identifiers to generate an overlapping record.

7. The automatic verification method for building material data quality based on intelligent agent collaboration according to claim 1, characterized in that, Step six specifically involves: Extract the overlapping results of the coverage boundary, the verification evidence propositions, the proposition nodes and the building material object identifiers from the overlapping records. Based on the proposition node mapping table, perform object range matching on the proposition nodes to form a set of overlapping proposition nodes under the same building material object identifier. Based on the constraint direction of the overlapping proposition node set in the standard constraint lattice, the consistent, conflicting, and missing states in the verification evidence propositions are aligned according to the position of the proposition nodes to generate proposition alignment records. Based on the proposition alignment record, perform consistency splicing on the set of verification evidence propositions. Write the splicing result that satisfies the constraint direction of the consistent state into the splicing pass record, and write the splicing result that does not satisfy the constraint direction of the conflicting state and the missing state into the splicing failure record. Extract the building material object identifier, collaborative verification node, proposition node, verification evidence proposition, and coverage boundary overlap result from the splicing failure record, and write the extracted results to the splicing failure location to generate coverage conflict unit.

8. The automatic verification method for building material data quality based on intelligent agent collaboration according to claim 1, characterized in that, Step seven specifically involves: Extract verification evidence propositions from the set of verification evidence propositions, determine the proposition nodes corresponding to the verification evidence propositions based on the proposition node mapping table, and convert the verification evidence propositions into basic probability assignment records; Based on the basic probability assignment record, Dubois-Prade conflict identification is performed. Verification evidence propositions with empty intersection of proposition nodes are identified as conflict evidence propositions, and conflict evidence proposition pairs are formed according to the building material object identifier. Locate the proposition nodes corresponding to the conflict evidence proposition pairs in the standard constraint lattice, search for common bearing nodes along the constraint direction in the standard constraint lattice, and determine the common bearing node with the smallest hierarchical distance as the least common superordinate proposition; The conflict quality transfer terms assigned to ordinary union propositions in the Dubois-Prade combination rules are replaced with conflict quality transfer terms corresponding to the least common superordinate proposition, and the conflict quality transfer objects are limited according to the covering conflict unit to generate an improved Dubois-Prade verification arbitration model.

9. The automatic verification method for building material data quality based on agent collaboration according to claim 1, characterized in that, Step eight specifically involves: Extract conflict quality transfer objects, basic probability assignment records, least common superordinate propositions, and covering conflict units from the improved Dubois-Prade verification arbitration model, establish association relationships according to building material object identifiers, and generate arbitration input records; Based on the arbitration input record, the proposition quality and conflict quality in the conflict quality transfer object of the basic probability assignment record are written into the least common superordinate proposition to generate the proposition arbitration quality record; Based on the proposition node mapping table and the coverage conflict unit, reverse positioning is performed on the proposition arbitration quality record. The standard field name associated with the proposition node is determined as the abnormal field positioning, and the overlap result of the constraint direction and coverage boundary in the standard constraint cell is determined as the abnormal relationship positioning. The collaborative verification node is determined as the verification source identifier. The building material object identifier, proposition arbitration quality record, abnormal field location, abnormal relationship location, and verification source identifier are associated and arranged to generate the building material data quality arbitration result.

10. An automatic verification system for building material data quality based on agent collaboration, comprising executing the automatic verification method for building material data quality based on agent collaboration as described in any one of claims 1 to 9, characterized in that, include: The data acquisition module is used to collect data on the flow of building materials business and generate a basic dataset for building materials verification. The standardization processing module is used to perform standardization processing on the basic dataset for building material verification, establish object association indexes, and generate a standardized record set for building materials. The collaborative verification module is used to configure collaborative verification nodes based on object association indexes and generate a set of verification evidence propositions. The constraint mapping module is used to construct a standard constraint lattice, map the set of validation evidence propositions to proposition nodes, and generate a proposition node mapping table. The coverage analysis module is used to establish the coverage range of verification evidence based on the data processing boundary and generate overlapping coverage records. The splicing conflict module is used to perform consistent splicing based on the overlapping records and the proposition node mapping table to generate overlapping conflict units. The arbitration modeling module is used to build an improved Dubois-Prade verification arbitration model and generate conflict quality transfer objects. The results output module is used to generate building material data quality arbitration results based on the conflict quality transfer object, and output the abnormal field location, abnormal relationship location, and verification source identifier.

Images