A data fusion processing method for the entire vaccine engineering lifecycle

Abstract

This invention discloses a data fusion processing method for the entire lifecycle of vaccine engineering, belonging to the field of data fusion technology. The method includes: receiving fusion queries for the entire vaccine engineering lifecycle and binding them to target application scenarios and output calibers; accessing multi-source real-time data based on the fusion queries and performing unified preprocessing and encapsulation to generate a unified event stream for the entire lifecycle; filtering event subsets matching the target application scenario under transferable gating control information constraints and constructing target reference and source comparison datasets to generate stable correction weights; and storing the fusion results, acquisition configuration descriptions, gating reason explanations, and weight stabilization information in a versioned manner to generate a replayable fusion operation record. This invention, by forming a continuous and consistent batch cross-stage event chain and reducing chain breaks and missing associations, makes it easier to verify and locate stage identifiers, event type identifiers, and event occurrence times under the same output caliber.

Description

Technical Field

[0001] This invention relates to the field of data fusion technology, and in particular to a data fusion processing method for the entire lifecycle of vaccine engineering. Background Technology

[0002] In existing technologies, vaccine engineering covers stages such as raw material preparation, packaging, testing, release, and cold chain transportation. Real-time data comes from multiple sources, including production execution records, quality inspection records, environmental monitoring records, and cold chain transportation records. A common method is to pull data from various source interfaces according to the query time range, perform field mapping, unit conversion, time format conversion, and missing data completion, and then sort by event occurrence time or aggregate by stage to generate monitoring reports or traceability analysis results.

[0003] However, in the process of processing real-time multi-source data for vaccine engineering, existing technologies often suffer from inconsistent field definitions, time semantics, and cross-stage association rules, leading to difficulties in aligning cross-source data in the same batch and easy breakage of cross-stage events. At the same time, the lack of interpretable screening constraints and weight correction mechanisms for different target application scenarios makes it easy for low-quality or inapplicable sources to be mixed in, resulting in fluctuations in fusion results and insufficient replayability. Summary of the Invention

[0004] This invention provides a data fusion processing method for the entire lifecycle of vaccine engineering, in order to solve the problems of inconsistent field definitions and time semantics in multi-source real-time data of vaccine engineering, unstable cross-stage batch event correlation, and the lack of interpretable constraints and corrections for scene migration, which leads to fluctuations in fusion results and difficulty in traceability.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] A data fusion processing method for the entire lifecycle of vaccine engineering includes receiving a fusion query for the entire lifecycle of vaccine engineering and binding the target application scenario and output caliber, accessing multi-source real-time data according to the fusion query and performing unified preprocessing and encapsulation to generate a unified event stream for the entire lifecycle;

[0007] A batch digital thread index is constructed based on a unified event stream throughout the entire lifecycle, and the events throughout the entire lifecycle of the batch are linked together to generate a cross-stage event set.

[0008] Inject collection configuration descriptions into cross-stage event sets, complete the verification and correction of collection configuration information through an interactive wizard, and generate an event set and configuration profile base with collection configuration descriptions.

[0009] For each fusion query, the configuration difference between the source-side configuration profile and the target-side configuration profile is calculated based on the configuration profile, and transferable gating control information and gating reason explanation are generated.

[0010] Based on the constraints of transferable gating control information, a subset of events matching the target application scenario is selected and a target reference and source comparison dataset is constructed to generate stable correction weights.

[0011] Real-time reweighted fusion of cross-stage event sets is performed on the batch digital thread index using stabilization correction weights, and the fusion result is output.

[0012] The fusion results, acquisition configuration descriptions, gating reason explanations, and weight stabilization information are versioned and stored to generate a replayable fusion operation record.

[0013] As a preferred embodiment of the data fusion processing method for the entire vaccine engineering lifecycle described in this invention, the specific steps for generating a unified event stream throughout the entire lifecycle are as follows:

[0014] Based on the fusion query, the target application scenario and output scope are analyzed to determine the source set and time range of multi-source real-time data, a multi-source access list is generated, multi-source real-time data is accessed according to the multi-source access list and the source identifier and collection time metadata are written, and data with metadata is output.

[0015] The data with metadata is standardized by performing field caliber consistency and time semantic consistency to generate standardized data. Then, a unified preprocessing encapsulation is performed through a unified event field structure to output a unified event stream with full lifecycle, including batch identifier, stage identifier, event type identifier, event occurrence time and source identifier.

[0016] As a preferred embodiment of the data fusion processing method for the entire vaccine engineering lifecycle described in this invention, the specific steps for constructing a batch digital thread index and generating a cross-stage event set based on a unified event stream throughout the entire lifecycle are as follows:

[0017] Based on the unified event flow throughout the entire lifecycle, batch identifiers and stage identifiers are extracted, an index key set is generated, and the event records are sorted according to the event occurrence time to output the batch stage event sequence.

[0018] For the event sequence in the batch stage, use the event type identifier to establish the event jump relationship and generate the batch number thread index;

[0019] By using batch number thread indexes to connect and associate batch identifiers with full-cycle event records in the full-cycle unified event stream, a cross-stage event set is generated.

[0020] As a preferred embodiment of the data fusion processing method for the entire vaccine engineering lifecycle described in this invention, the specific steps for generating the event set and configuration profile basis with collection configuration descriptions are as follows:

[0021] Based on the cross-stage event set, basic event entries are constructed, source traceability information and quality tags are generated, and the basic event entry set is organized into event source entries and collection context entries.

[0022] The event source entries and collection context entries are merged using event type identifiers and stage identifiers to generate a draft collection configuration description.

[0023] By collecting the initial draft of the configuration description, the items to be verified for the configuration information are output, and the interactive wizard is used to generate the verification task.

[0024] Receive verification input for the verification task, associate the verification input with the source traceability information, record and update the initial draft of the collection configuration description, and generate a collection configuration description that has passed verification.

[0025] The verified collection configuration descriptions are backfilled into the cross-stage event set, the event set with collection configuration descriptions is output, and the event set with collection configuration descriptions and collection configuration information are aggregated to generate the basic configuration profile.

[0026] As a preferred embodiment of the data fusion processing method for the entire vaccine engineering lifecycle described in this invention, the specific steps for generating transferable gating control information and gating cause explanations are as follows:

[0027] For each fusion query, the target application scenario and output scope are read, and the profile extraction list containing source-side scope items and target-side scope items is generated by associating it with the configuration profile base.

[0028] Based on the source-side scope entries and target-side scope entries in the image extraction list, extract the source-side configuration image and target-side configuration image from the configuration image base to generate a dual-side configuration image.

[0029] Perform item-level comparison on the two-sided configuration profiles to generate a set of difference items. Associate the set of difference items with the stage identifier and event type identifier of the cross-stage event set to generate a description of the impact of the difference.

[0030] The difference impact description is used to perform gating determination on the transferability of differences, generate transferable gating control information, and summarize and organize it with the difference impact description to generate gating cause explanations.

[0031] As a preferred embodiment of the data fusion processing method for the entire vaccine engineering lifecycle described in this invention, the specific steps for filtering and matching a subset of events for the target application scenario based on transferable gating control information constraints and constructing a target reference and source comparison dataset are as follows.

[0032] By utilizing transferable gating control information, gating and filtering are performed on the cross-stage event set to generate an event subset that matches the target application scenario. Then, field alignment and time aggregation are performed on the event subset according to the output caliber to generate the target reference dataset.

[0033] Based on the output criteria, perform field alignment and time aggregation on event records in the cross-stage event set that did not enter the event subset to generate a source comparison dataset;

[0034] Perform stratified alignment of the target reference dataset and the source control dataset by the control key, generate a set of correction factors and output the initial correction weights;

[0035] Perform weight stabilization processing on the initial correction weights to generate weight stabilization information, and output the stabilized correction weights.

[0036] As a preferred embodiment of the data fusion processing method for the entire vaccine engineering cycle described in this invention, the gating cause explanation refers to the generation of configuration differences by summarizing the stage identifier and event type identifier in an itemized manner, and associating the difference impact description and corresponding transferable gating control information for each configuration difference item.

[0037] As a preferred embodiment of the data fusion processing method for the entire vaccine engineering cycle described in this invention, the specific steps of performing real-time reweighted fusion of cross-stage event sets on the batch digital thread index using stabilization correction weights and outputting the fusion result are as follows:

[0038] By utilizing the output scope of the fusion query binding, the summary granularity and field set are determined, and the summary constraints are output.

[0039] Based on the batch digital thread index, locate the event link corresponding to the batch identifier in the cross-stage event set, generate link extraction conditions, extract event records along the event jump relationship, and output the batch stage event sequence in sorted by the event occurrence time.

[0040] By establishing a weight mapping relationship using stabilization correction weights, stage identifiers, and event type identifiers, and generating weight assignment rules, the stabilization correction weights are written into each event sequence of the batch stage to generate a weighted event sequence.

[0041] Real-time incremental aggregation is performed on the weighted event sequence using aggregation constraints to generate fusion results.

[0042] As a preferred embodiment of the data fusion processing method for the entire vaccine engineering lifecycle described in this invention, the specific steps for generating a replayable fusion operation record are as follows:

[0043] Generate runtime version identifiers from fusion queries;

[0044] Based on the running version identifier, the target application scenario, output caliber and fusion result are associated and stored, and index information is generated for the collection configuration description, gating reason explanation, weight stabilization information and the weighted event sequence corresponding to the batch digital thread index.

[0045] The index information is associated with the running version identifier to perform versioned storage, generating a replayable fusion running record.

[0046] As a preferred embodiment of the data fusion processing method for the entire vaccine engineering cycle described in this invention, the gating determination refers to generating a gating score based on the difference impact description, using the effective proportion of the target side configuration profile as the gating threshold, writing "allow" in the transferable gating control information when the gating score is not greater than the gating threshold, and writing "prohibit" in the transferable gating control information when the gating score is greater than the gating threshold.

[0047] The beneficial effects of this invention are as follows:

[0048] In the context of digital data processing, by generating a unified event stream throughout the entire lifecycle and constructing a batch digital thread index to connect and associate cross-stage event records in the same batch, the field caliber and temporal semantics of multi-source real-time data are unified, forming a continuous and consistent batch cross-stage event link and reducing broken links and missing associations. This makes it easier to verify and locate stage identifiers, event type identifiers, and event occurrence times under the same output caliber, thereby reducing fluctuations in fusion results and improving interpretability, traceability, and replayability. Attached Figure Description

[0049] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a flowchart illustrating the data fusion and processing method for the entire lifecycle of vaccine engineering.

[0051] Figure 2 A flowchart for generating a unified event flow throughout the entire lifecycle.

[0052] Figure 3 A flowchart for constructing a batch numeric thread index.

[0053] Figure 4 A flowchart for generating transferable gating control information. Detailed Implementation

[0054] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0055] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0056] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0057] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a data fusion processing method for the entire lifecycle of vaccine engineering, including the following steps:

[0058] S1: Receives fusion queries for the entire vaccine engineering lifecycle and binds them to target application scenarios and output standards. Based on the fusion queries, it accesses real-time data from multiple sources and performs unified preprocessing and encapsulation to generate a unified event stream for the entire lifecycle.

[0059] S1.1: Based on the fusion query, analyze the target application scenario and output scope, determine the source set and time range of multi-source real-time data, generate a multi-source access list, access multi-source real-time data according to the multi-source access list and write source identifier and collection time metadata, and output data with metadata.

[0060] It receives fusion queries throughout the entire vaccine engineering lifecycle, reads the original content of the fusion query, and the original content of the fusion query includes the target application scenario field, the output caliber field, and the time range field.

[0061] Using JSONPath, the target application scenario field path, output caliber field path, and time range field path are located in the original content of the fusion query, and a record of the required field positions containing the target application scenario field path, output caliber field path, and time range field path is generated.

[0062] The existence of each field is checked one by one based on the required field location record; the check result is written to the check result field of the required field location record; when any check result is missing, the query receiving interface returns the name of the missing field and terminates the subsequent data processing of the fusion query.

[0063] Based on the target application scenario field path, the original content of the fusion query is parsed to extract field values, the target application scenario field values ​​are output and written into the fusion query processing context, and the fusion query-target application scenario binding relationship is generated.

[0064] Based on the output caliber field path, the original content of the fusion query is parsed to extract field values. The output includes a list of caliber fields, the meaning of each caliber field, the unit of each caliber field, the set of stage identifier values, and the set of event type identifier values.

[0065] Write the value of the output caliber field into the fusion query processing context to generate a fusion query-output caliber binding relationship.

[0066] Based on the time range field path, the original content of the fusion query is parsed to extract field values, and the output is the time range field value containing the start time field value and the end time field value.

[0067] Perform UTC millisecond timestamp validity and sequence checks on the values ​​of the start time field and the end time field, respectively;

[0068] When any validity check fails, the query receiving interface returns a time range field parsing failure flag. When the sequence check determines that the start time field value is greater than the end time field value, the query receiving interface returns a time range field sequence error flag.

[0069] If a time range field parsing failure flag or a time range field order error flag appears, the subsequent data processing of the fusion query will be terminated.

[0070] Set the value of the start time field as the start time, and set the value of the end time field as the end time.

[0071] Write the start and end times into the fusion query processing context to generate the fusion query - time range determination result.

[0072] Based on the binding relationship between fusion query and target application scenario, the binding relationship between fusion query and output caliber, and the result of fusion query and time range determination, the source set of multi-source real-time data is determined.

[0073] For each source identifier in the source set, a multi-source access list record is created. The multi-source access list record contains the source identifier, source data interface location information, source-side event occurrence time field name, source-side batch identifier field name, source-side event type field name, source-side stage identifier field name, and a description of the correspondence between the source field and the output caliber field.

[0074] All multi-source access list records are aggregated to form a multi-source access list and written into the fusion query processing context.

[0075] Under the constraints of the fusion query processing context, a real-time fetch request is initiated based on the source data interface location information of the multi-source access list records to obtain the data record set from the source side.

[0076] The data record set from the source side is used as multi-source real-time data, and source identifier metadata is written into each data record in the multi-source real-time data.

[0077] The local record time that initiates the real-time fetch request is converted into a UTC millisecond timestamp and written into the collection time metadata of each data record in the multi-source real-time data.

[0078] Data records that simultaneously contain source identifier metadata and collection time metadata are designated as metadata-enabled data records.

[0079] S1.2: Perform field caliber consistency and time semantic consistency on data with metadata to generate standardized data, and perform unified preprocessing encapsulation through a unified event field structure to output a unified event stream with full lifecycle, including batch identifier, stage identifier, event type identifier, event occurrence time and source identifier.

[0080] Based on the source identifier in the metadata data record, a matching multi-source access list record is retrieved in the multi-source access list, and an association relationship is established between the metadata data record and the multi-source access list record for each metadata data record.

[0081] Based on the data records with metadata in the association relationship, read the values ​​of batch-related fields by the batch identifier field name on the source side, read the values ​​of stage-related fields by the stage identifier field name on the source side, and read the values ​​of event-related fields by the event type identifier field name on the source side.

[0082] Based on the correspondence between source fields and output caliber fields in the multi-source access list record, the values ​​of batch-related fields, stage-related fields, and event-related fields are converted into the field value expression form defined by the output caliber. The conversion process includes field naming consistency, enumeration value consistency, and unit expression consistency.

[0083] Field naming consistency is achieved by converting source field names into output caliber field names based on the list of caliber field values ​​in the output caliber field.

[0084] The enumeration value consistency is achieved by converting the source field value to the output caliber field value according to the meaning of the caliber field value in the output caliber field value;

[0085] Unit representation consistency is achieved by converting the unit representation of the source field value to the output caliber unit representation based on the caliber field value in the output caliber field.

[0086] The transformed field values, the source identifier and the data acquisition time metadata in the data record are written into the same record to generate a normalized data record. All normalized data records are then collected and output as a normalized data record set.

[0087] It should be noted that the standardized data record, as a record after the field caliber has been consistent, includes the values ​​of batch-related fields, the converted values ​​of stage-related fields and event-related fields, source identifiers, and metadata of the collection time.

[0088] The source identifier is determined by the source identifier metadata. The batch-related field values ​​represent the identification information of the same batch of objects in the vaccine project. The stage-related field values ​​represent the identification information of the stage to which the event belongs. The event-related field values ​​represent the identification information of the event category.

[0089] Based on the association between data records with metadata and multi-source access list records, the name of the source-side event occurrence time field is determined for each normalized data record in the normalized data record set.

[0090] Retrieves the original value of the event occurrence time from the normalized data record based on the event occurrence time field name on the source side.

[0091] Perform standardized format processing on the original values ​​of the event occurrence time from the source side to output a unified time representation;

[0092] When the original value of the event occurrence time on the source side is empty or the unified time expression is empty, the data acquisition time metadata is read and the data acquisition time metadata is determined as the event occurrence time. At the same time, the time source mark field is written to the data acquisition rollback time.

[0093] When the unified time expression is not null, the unified time expression is determined as the event occurrence time, and the time source marker field is written to the source side time.

[0094] Write the event occurrence time into the event occurrence time field of the normalized data record, and output a set of normalized data records with semantically consistent completion time.

[0095] For each normalized data record in the set of normalized data records that have achieved semantic consistency over time, a mapping match is performed on the set of stage identifier values ​​based on the values ​​of the stage-related fields in the normalized data record:

[0096] When a match is found, the stage identifier is determined and written to the normalized data record;

[0097] When the value of a stage-related field does not match the value set of stage identifiers, the stage identifier field is set to null, the mapping failure flag field is written to stage mapping failure, and the quality flag is set to exception.

[0098] Based on the values ​​of event-related fields in the normalized data records, a mapping match is performed within the set of event type identifier values:

[0099] When a match is found, the event type identifier is determined and written to a normalized data record;

[0100] When the value of an event-related field does not match the value set of the event type identifier, the event type identifier field is set to null, the mapping failure flag field is written to the event type mapping failure field, and the quality flag is set to exception.

[0101] For each normalized data record in the normalized data record set, an event record is generated by encapsulating it according to a unified event field structure (batch identifier, stage identifier, event type identifier, event occurrence time, source identifier), and all event records are collected and output as an event record set.

[0102] The event record set is sorted by the event occurrence time, and the sorted event records are concatenated in sequence to generate a unified event stream for the entire lifecycle.

[0103] The unified event flow throughout the entire lifecycle contains only five types of fields: batch identifier, stage identifier, event type identifier, event occurrence time and source identifier.

[0104] S2: Construct a batch digital thread index based on a unified event stream throughout the entire lifecycle, and connect and associate the events throughout the entire lifecycle of the batch to generate a cross-stage event set;

[0105] S2.1: Based on the unified event flow throughout the entire lifecycle, extract the batch identifier and stage identifier, generate an index key set, sort the event records according to the event occurrence time, and output the batch stage event sequence.

[0106] Based on the event records in the unified event stream throughout the entire lifecycle, the batch identifier, stage identifier, event type identifier, event occurrence time, and source identifier are read for each event record to generate a set of event records to be processed.

[0107] For each event record in the event record set to be processed, read the batch identifier and stage identifier; combine the batch identifier and stage identifier to generate an index key; deduplicate all index keys and output the index key set.

[0108] For each index key in the index key set, filter the event records in the event record set that match both the batch identifier and the stage identifier. Sort the records in ascending order by the event occurrence time. When the event occurrence times are the same, sort them in the following order by stage identifier, event type identifier, and source identifier. Output the sorted list of event records corresponding to the index key.

[0109] Combine the index key with the corresponding sorted event record list into a batch stage event sequence.

[0110] S2.2: For the event sequence of the batch stage, use the event type identifier to establish the event jump relationship and generate the batch number thread index.

[0111] Read the list of sorted event records corresponding to all index keys based on the batch stage event sequence; for each batch identifier, collect all event records matching the batch identifier to generate a list of all batch event records corresponding to the batch identifier;

[0112] The list of all event records corresponding to the batch identifier is sorted in ascending order by the event occurrence time.

[0113] When events occur at the same time, they are stably sorted according to stage identifier, event type identifier, and source identifier, and the batch full-cycle event sequence corresponding to the batch identifier is output.

[0114] For each batch of full-cycle event sequence, read the event records one by one in sorted order; for two adjacent event records, read the event type identifiers respectively, and generate a jump correspondence from the preceding event type identifier to the following event type identifier.

[0115] Write the batch identifier, stage identifier, event type identifier, event occurrence time, and source identifier of the preceding and subsequent event records into the event jump relationship. Collect all event jump relationships in the full-cycle event sequence of the batch and output the event jump relationship set.

[0116] For each batch identifier, locate the event record with the earliest occurrence time in the full-cycle event sequence of the batch, and use the batch identifier, stage identifier, event type identifier, event occurrence time, and source identifier of the event record as the starting location information corresponding to the batch identifier;

[0117] For each event jump relationship in the event jump relationship set, establish an index mapping relationship between the batch identifier, stage identifier, event type identifier, event occurrence time, and source identifier of the preceding event record and the batch identifier, stage identifier, event type identifier, event occurrence time, and source identifier of the following event record;

[0118] The batch identifier is compiled with the corresponding starting location information and index mapping relationship, and the batch numeric thread index is output.

[0119] S2.3: Utilize the batch number thread index to connect the full-cycle event records in the full-cycle unified event stream corresponding to the batch identifier, and generate a cross-stage event set.

[0120] For each batch identifier in the batch digital thread index, read the starting location information corresponding to the batch identifier, search for event records in the full-cycle unified event stream that match the starting location information along with the batch identifier, stage identifier, event type identifier, event occurrence time, and source identifier, and output the batch start event record.

[0121] The batch start event record is used as the current event record and recorded as the first event record in the interconnected chain.

[0122] Based on the current event record, retrieve the location information of the subsequent event record corresponding to the current event record in the index mapping relationship of the batch digital thread index.

[0123] For each subsequent event record location information, search the unified event stream throughout the entire lifecycle for event records that simultaneously match the batch identifier, stage identifier, event type identifier, event occurrence time, and source identifier to obtain candidate subsequent event records.

[0124] Based on the event occurrence time of the current event record as the time base, only candidate subsequent event records whose event occurrence time is greater than or equal to the time base and whose batch identifier matches the batch identifier in the index mapping relationship are retained.

[0125] Sort the candidate subsequent event records that meet the conditions in ascending order of event occurrence time, select the earliest one as the next connected event record, update the current event record and record the connected order.

[0126] If no matching candidate subsequent event record exists, the process of linking the corresponding batch identifier is terminated. During the linking process, the event occurrence time after semantic consistency of completion time is used as the time reference.

[0127] The jump order of stage identifiers and event type identifiers is determined based on the index mapping relationship constructed from the full-cycle event sequence of the batch, and the order of cross-stage event records within the batch is kept consistent and time-synchronized on a unified time axis.

[0128] Collect all the interconnected event records in the interconnected process, generate a cross-stage event set corresponding to the batch identifier, and then collect and output the cross-stage event sets corresponding to all batch identifiers as a cross-stage event set.

[0129] S3: Inject cross-stage event set collection configuration description, complete the verification and correction of collection configuration information through interactive wizard, and generate event set and configuration profile basic with collection configuration description;

[0130] S3.1: Construct basic event entries based on cross-stage event sets, generate source traceability information and quality tags, and organize the basic event entry set into event source entries and collection context entries.

[0131] Based on the cross-stage event set, the source identifier, stage identifier, event type identifier, and event occurrence time are used to generate basic event entries for each event record, and all basic event entries are aggregated into a basic event entry set.

[0132] For each event basic entry in the event basic entry set, based on the source identifier, retrieve the matching multi-source access list record in the multi-source access list, read the source data interface location information and the correspondence description from the source field to the output caliber field, combine them to generate source traceability information, and write it into the corresponding event basic entry.

[0133] Based on the set of basic event entries, the criterion for determining the resolvability of the event occurrence time is limited to the timestamp field being a UTC millisecond timestamp and being resolvable;

[0134] The criteria for determining the matching of stage identifiers are limited to the set of stage identifier values ​​that the value of the stage identifier field belongs to the set of stage identifier values ​​in the output caliber field.

[0135] The criteria for determining the event type identifier matching check are limited to the event type identifier field value belonging to the set of event type identifier values ​​in the output caliber field value set;

[0136] The criterion for source identifier matching verification is limited to the existence of a unique matching record for the source identifier field value in the multi-source access list.

[0137] A quality mark is generated as valid when all four checks pass.

[0138] When any check fails, a quality flag is generated as an exception; the quality flag is written to the corresponding event base entry.

[0139] Based on the source identifier, aggregate the source traceability information and quality markers from the event base entry set, and output the event source entry;

[0140] Based on the event basic entry set, the event occurrence time, source identifier, and quality marker in the event basic entry set are collected by stage identifier and event type identifier, and the collection context entry is output.

[0141] S3.2: Merge the event source entries and collection context entries using event type identifiers and stage identifiers to generate a draft collection configuration description.

[0142] Based on the event source entries and the collection context entries, the event source entries and the collection context entries are grouped according to the combination of event type identifier and stage identifier, generating several grouped entry sets corresponding to the combination of event type identifier and stage identifier.

[0143] For each group of entries, the source traceability information, quality markers, and collection-related information in the collection context entries are summarized to generate the corresponding collection configuration description unit.

[0144] All data acquisition configuration description units are combined according to the order of the stage identifier value set and the event type identifier value set in the output caliber field. A draft data acquisition configuration description is generated.

[0145] S3.3: By collecting the initial draft of the configuration description, output the items to be verified for the collection configuration information, and enter the interactive wizard to generate the verification task.

[0146] For each entry in the initial draft of the data collection configuration description, extract the data collection configuration information to be verified entries. The entries to be verified include stage identifiers, event type identifiers, source identifier sets, and source traceability information sets.

[0147] Based on the initial draft of the data collection configuration description, the data collection frequency field, data collection location field, data collection object field, and data collection time schedule field are read one by one to generate the values ​​of the fields to be verified.

[0148] When there are missing or inconsistent source traceability information sets corresponding to the same stage identifier and the same event type identifier, the corresponding entries are identified as entries to be verified, and all entries to be verified are collected and output as a set of entries to be verified.

[0149] Verification tasks are generated based on the set of items to be verified, and the collection configuration information, source traceability information and quality marks corresponding to each item to be verified are displayed one by one through an interactive wizard.

[0150] S3.4: Receive the verification input for the verification task, associate the verification input with the source traceability information, record and update the initial draft of the collection configuration description, and generate the collection configuration description that has passed verification.

[0151] Receive verification input for the verification task, associate the verification input with the corresponding source traceability information, use the verification item identifier field to locate the item to be verified, and use the source identifier field to locate the matching source traceability information in the source traceability information set, and write the verification input into the verification input record.

[0152] The verification input record includes the verification task identifier field, verification item identifier field, verification action type field, field value before verification, field value after verification, operator identifier field, and verification occurrence time field. The verification occurrence time field is expressed as a UTC millisecond timestamp.

[0153] The initial draft of the data collection configuration description adds a version number field with an initial value of 1. When the verification action type field, which represents the data collection configuration information, needs to be adjusted, the corresponding field in the initial draft of the data collection configuration description is updated according to the field value after verification, and the version number field is incremented by 1.

[0154] When the verification action type field represents the collection configuration information and remains unchanged, the original field value is retained. After all items to be verified are verified and updated, the collection configuration description that has passed the verification is output.

[0155] S3.5: Backfill the verified collection configuration descriptions into the cross-stage event set, output the event set with collection configuration descriptions, and aggregate the event set with collection configuration descriptions and collection configuration information to generate the basic configuration profile.

[0156] Based on the event type identifier and stage identifier in the verified collection configuration description, locate the event record in the cross-stage event set whose batch identifier, stage identifier and event type identifier match, backfill the event-related collection configuration information in the verified collection configuration description to the matching event record, and collect and output all event records with collection configuration description as an event set with collection configuration description.

[0157] Based on the event set with collection configuration description and the collection configuration information in the verified collection configuration description, the distribution of source identifiers, quality markers and event occurrence time coverage in the event set with collection configuration description are aggregated according to stage identifier and event type identifier. The aggregation results are combined with the corresponding collection configuration information to output the basic configuration profile.

[0158] S4: For each fusion query, extract the configuration profile from the source side and the configuration profile from the target side based on the configuration profile, calculate the configuration difference, and generate transferable gating control information and gating reason explanation.

[0159] S4.1: For each fusion query, read the target application scenario and output scope, and associate them with the configuration profile base to generate a profile extraction list containing source-side scope items and target-side scope items.

[0160] For each fusion query, based on the target application scenario, output scope, and fusion query-time range determination results in the fusion query processing context, the source identifier set is extracted from the multi-source access list, and the stage identifier value set and event type identifier value set are extracted from the output scope field values.

[0161] Write the target application scenario, output caliber, start time, end time, source identifier set, stage identifier value set, and event type identifier value set into the fusion query list reference record in the fusion query processing context.

[0162] The event records that fall within the start and end time ranges from the cross-stage event set are filtered and collected, and the output is a time range event set.

[0163] Extract the source identifier set from the time range event set, compare it with the source identifier set in the multi-source access list, select the source identifiers that exist in both, and generate the target side source identifier set.

[0164] For each stage identifier in the stage identifier value set and each event type identifier in the event type identifier value set, generate a target-side scope entry.

[0165] In the configuration profile base, retrieve configuration profile base entries that match the phase identifier, match the event type identifier, and whose source identifier falls into the target side source identifier set;

[0166] For the retrieved configuration profile basic entries, the event occurrence time coverage and quality label distribution are aggregated based on the source identifier to generate a set of source candidate profile entries.

[0167] The effective percentage of the source candidate image entries is calculated based on the quality label distribution, and the coverage length is calculated based on the event occurrence time coverage.

[0168] Sort the effective percentage and coverage length in descending order, and extract the source identifier set at the beginning of the sorting based on the number of source identifier sets on the target side to generate the source identifier set on the source side.

[0169] For each stage identifier in the stage identifier value set and each event type identifier in the event type identifier value set, generate a source-side scope entry.

[0170] For each stage identifier and event type identifier combination, pair one source-side scope entry and one target-side scope entry to generate a portrait extraction list entry. Collect all portrait extraction list entries and output them as a portrait extraction list.

[0171] S4.2: Based on the source-side scope entries and target-side scope entries in the image extraction list, extract the source-side configuration image and target-side configuration image from the configuration image base to generate a dual-side configuration image.

[0172] For each entry in the image extraction list, construct search criteria using the stage identifier, event type identifier, start time, end time, source-side source identifier set, and target-side source identifier set.

[0173] In the configuration profile base, the source side source identifier set and the target side source identifier set are used as source identifier retrieval conditions. The basic configuration profile entries that match both the stage identifier and the event type identifier, whose source identifier falls into the corresponding source identifier set, and whose event occurrence time coverage overlaps with the start time and end time are selected.

[0174] The rules for determining the overlap are limited to: the start time of the event occurrence time coverage interval of the basic profile entry must be less than or equal to the end time, and the end time of the coverage interval must be greater than or equal to the start time.

[0175] For each side of the search results, a field aggregation is performed. The field aggregation includes the collection configuration information, source identifier distribution, quality marker distribution, and event occurrence time coverage. The source side configuration profile and the target side configuration profile are output.

[0176] Pair the source-side configuration image and the target-side configuration image corresponding to the same image extraction list item to generate a dual-side configuration image, and collect all dual-side configuration images to output a dual-side configuration image set.

[0177] It should be noted that the field aggregation rules are limited to: collecting configuration information by selecting the value of the field corresponding to the latest verification input record after verification based on the verification occurrence time field; summarizing the source identifier distribution by counting the source identifier; summarizing the quality mark distribution by counting the quality mark; and generating the coverage range for event occurrence time based on the earliest and latest event occurrence times.

[0178] S4.3: Perform item-level comparison on the two-sided configuration profiles to generate a set of difference items. Associate the set of difference items with the stage identifier and event type identifier of the cross-stage event set to generate a description of the impact of the difference.

[0179] For each bilateral configuration portrait, generate difference entries according to the following four categories of control items:

[0180] Collection configuration information comparison item: Compare the values ​​of the collection configuration information fields of the source side configuration profile with the values ​​of the collection configuration information fields of the target side configuration profile field by field;

[0181] Source identifier distribution comparison item: Compare the source identifier distribution of the source-side configuration image with the source identifier distribution of the target-side configuration image;

[0182] Quality Marker Distribution Comparison: Compare the quality marker distribution of the source-side configuration image with the quality marker distribution of the target-side configuration image;

[0183] Event occurrence time coverage comparison item: Compare the event occurrence time coverage of the source side configuration profile with the event occurrence time coverage of the target side configuration profile.

[0184] For each bilateral configuration profile, comparison results are generated item by item based on the collected configuration information comparison items, source identifier distribution comparison items, quality mark distribution comparison items, and event occurrence time coverage comparison items. Inconsistency marks are determined based on the comparison results, and the values ​​of the inconsistency marks are limited to consistent and inconsistent.

[0185] Write the stage identifier, event type identifier, comparison item name, source side value summary, target side value summary, and inconsistency flag into the difference entries, and then aggregate all difference entries into a difference entry set.

[0186] Use cross-stage event sets and fusion query - time range to determine the start and end times in the results. For each difference entry, filter the event records in the cross-stage event set by stage identifier, event type identifier and event occurrence time range to generate a subset of difference entry events.

[0187] Using the batch number thread index, extract the batch identifier set from the subset of difference entry events, and perform chain extraction for each batch identifier in the batch identifier set:

[0188] Based on the batch number thread index, the event records corresponding to the batch identifier are linked in the cross-stage event set, and the output is the event chain corresponding to the batch identifier.

[0189] Locate the event record position in the event chain where the stage identifier matches the event type identifier and the event occurrence time falls within the range of the start and end times;

[0190] Starting from the location of the event record, the system jumps sequentially along the index mapping relationship of the batch number thread index to extract subsequent event records until the index mapping relationship no longer contains the location information of subsequent event records.

[0191] Count the number of successive jumps, where the jump count is the count of the number of event jump relationships, and output the event link propagation length count corresponding to the batch identifier.

[0192] Collect event link propagation length counts for the batch identifier set and output the event link propagation length summary.

[0193] For each batch identifier in the batch identifier set, based on the index mapping relationship of the batch number thread index, jump from the starting location information to the subsequent event record location information until the index mapping relationship no longer exists. Count the number of jumps and output the event link length count.

[0194] Collect event link length counts for the batch identifier set and output the event link length summary.

[0195] For the subset of events with discrepancies, compile a set of batch identifiers, a set of source identifiers, and the time range of the event occurrence.

[0196] The batch identifier set, source identifier set, event occurrence time range, and event propagation length are summarized and written into the difference impact description;

[0197] Write the difference impact description into the corresponding difference entry in the difference entry set to generate a difference entry set with difference impact description.

[0198] S4.4: Utilize the difference impact description to perform gating determination on the transferability of differences, generate transferable gating control information, and summarize and organize it with the difference impact description to generate gating cause explanations.

[0199] The ratio calculation method is used to calculate the ratio between the set size of the batch identifier set corresponding to the subset of difference item events and the set size of the batch identifier set corresponding to the same stage identifier and the same event type identifier in the cross-stage event set, and output the event batch influence ratio.

[0200] The percentage of events affected by a particular batch is expressed as follows:

[0201] ;

[0202] In the formula, This is the set of batch identifiers corresponding to the subset of difference item events. Size of the batch identifier set; This refers to the set of batch identifiers corresponding to the same stage identifier and the same event type identifier in the cross-stage event set. The size of the batch identifier set (take the cardinality of the batch identifier set as the count value). The proportion of events affected by batches

[0203] The ratio calculation method is used to calculate the ratio between the sum of the event link propagation lengths corresponding to the difference items and the sum of the event link lengths corresponding to the same batch of identifier sets, and the event link propagation ratio is output.

[0204] The event propagation ratio is expressed as follows:

[0205] ;

[0206] in, Summarize the event chain propagation lengths corresponding to the difference entries. This is a summary of the event link lengths corresponding to the same batch of identifier sets. The proportion of event propagation along the chain. The minimum value between the sum of event propagation length and the sum of event link length is taken. To prevent division by zero errors when the total event link length is zero, The value is not greater than ,and The value of is not less than Event chain propagation ratio The value range of is limited to the interval (0 to 1).

[0207] Perform a combined calculation on the batch impact ratio and the link propagation ratio, and output the joint impact ratio.

[0208] The proportion of joint impact is expressed as follows:

[0209] ;

[0210] In the formula, This represents the proportion of combined influence.

[0211] The difference impact score is determined based on inconsistency markers and the proportion of joint impact:

[0212] When inconsistency is marked as consistency, the difference affects the score and is written to zero.

[0213] When inconsistency is marked as inconsistency, the difference impact score is written into the joint impact ratio.

[0214] The difference affects the score, expressed as follows:

[0215] ;

[0216] In the formula, Inconsistency marker, Mark inconsistencies as inconsistent. Mark inconsistencies as consistent; This is an indicator function.

[0217] Write the difference entry based on the difference impact score, and output the difference entry set with the difference impact score.

[0218] Merge the difference item set with difference impact score according to stage identifier and event type identifier, summarize the difference impact score for each merge result, and output the gated score.

[0219] Among them, the gating score is obtained by evaluating the impact of the aggregated differences on the merged results, and is expressed as follows:

[0220] ;

[0221] In the formula, To score the gate control, For the first The differences in the items affect the score. This represents the number of difference entries after merging.

[0222] Read the quality label distribution of the target side configuration image and calculate the effective proportion of the target side. Determine the effective proportion of the target side as the gating threshold (example value: 0~1).

[0223] The effective proportion of the target side is defined as the gating threshold, expressed as:

[0224] ;

[0225] In the formula, Configure the effective counts in the image quality marker distribution for the target side. Configure anomaly counts in the image quality label distribution for the target side. The threshold value is used for gate control. Configure the target side to distribute the effective count and the anomaly count of the image quality label distribution.

[0226] The type of gate action is determined based on the number of merged difference entries, gate score, and gate threshold.

[0227] The gating action type is represented as follows:

[0228] ;

[0229] In the formula, This refers to the gate action type in the transferable gate control information.

[0230] when At this time, writing the gate action type is prohibited;

[0231] when At that time, writing the gate action type is allowed;

[0232] when Gating action type write is allowed;

[0233] when At that time, writing the gate action type is prohibited.

[0234] The target application scenario, output caliber, stage identifier, event type identifier, and gating action type are aggregated and output as transferable gating control information.

[0235] For each piece of transferable gating control information, retrieve the difference entries that match the stage identifier and event type identifier, sort them in descending order by difference impact score, and extract the first few difference entries (a fixed number of entries is 5) as the cause entries.

[0236] The output includes the name of the control item, a summary of the source-side value, a summary of the target-side value, a description of the difference in impact, and a score for the difference in impact. This output is then combined with the gating score, gating threshold, and gating action type to form an explanation of the gating cause.

[0237] S5: Based on the constraints of transferable gating control information, filter the event subset that matches the target application scenario and construct the target reference and source comparison dataset to generate stable correction weights.

[0238] S5.1: Utilize transferable gating control information to perform gating filtering on cross-stage event sets, generate event subsets that match the target application scenario, and perform field alignment and time aggregation on the event subsets according to the output criteria to generate the target reference dataset.

[0239] Using transferable gating control information, records with allowed gating action types are filtered out, and the combination of stage identifier and event type identifier is extracted, deduplicated, and aggregated to generate a set of allowed combinations.

[0240] The system retrieves combinations of stage identifiers and event type identifiers that match within the allowed combination set; when a retrieval is successful, the event records are aggregated into an event subset; when a retrieval fails, the event records are skipped, and the aggregated results are output as an event subset that matches the target application scenario.

[0241] Perform field alignment on each event record in the event subset, preserving the batch identifier, stage identifier, event type identifier, event occurrence time, and source identifier;

[0242] The verification phase identifier value belongs to the phase identifier value set, and the verification event type identifier value belongs to the event type identifier value set. The event records that pass the verification are collected and output as an aligned event subset.

[0243] Group the aligned event subsets by batch identifier, and sort them in ascending order by event occurrence time within each batch identifier group.

[0244] Within each batch identifier group, extract the preceding stage identifier and preceding event type identifier, and the subsequent stage identifier and subsequent event type identifier from adjacent event records, generate event jump entries, and count the number of event jump entries; at the same time, count the number of event count entries by grouping by stage identifier, event type identifier, and source identifier.

[0245] Write the event count entries and event jump entry counts into a lookup key count table, and output the lookup key count table as the target reference dataset.

[0246] S5.2: Based on the output criteria, perform field alignment and time aggregation on event records in the cross-stage event set that have not entered the event subset, and generate the source comparison dataset.

[0247] For each event record set that is not included in the event subset, retain the batch identifier, stage identifier, event type identifier, event occurrence time and source identifier, and verify that the stage identifier value belongs to the stage identifier value set and the event type identifier value belongs to the event type identifier value set, and compile them to generate an aligned comparison event set.

[0248] The alignment comparison event set is grouped by batch identifier and sorted in ascending order by event occurrence time. The event jump entries formed by adjacent event records are counted. The event count entries are grouped and counted by stage identifier, event type identifier and source identifier.

[0249] Write the event jump entry count and event count entries into the reference key count table and output the reference key count table as the source reference dataset.

[0250] Specifically, the reference key is expressed using a structured key, and a key type marker field is added to the reference key; the reference key for event count entries is generated by combining the stage identifier, event type identifier, and source identifier, and the reference key for event jump entry count is generated by combining the preceding stage identifier, preceding event type identifier, subsequent stage identifier, and subsequent event type identifier; the reference key and corresponding count are written into the reference key count table and the reference key count table is output.

[0251] S5.3: Perform stratified comparison on the target reference dataset and the source control dataset to generate a set of correction factors and output the initial correction weights.

[0252] Using the reference key as the stratification key, alignment is performed on the target reference dataset and the source reference dataset to generate a stratified alignment table; the stratified alignment table writes the target side count and the source side count for each reference key.

[0253] Calculate the correction factor for each control key in the hierarchical alignment table and write it into the correction factor set. Organize the correction factor set according to the control key and output it as the initial correction weight.

[0254] The initial correction weight is expressed as:

[0255] ;

[0256] In the formula, For comparison keys; For the target reference dataset, refer to the key. Corresponding count, For the reference key in the source reference dataset Corresponding count, This is the smoothing constant (the example value is 1). Indicates the reference key Corresponding to the initial correction weights.

[0257] S5.4: Perform weight stabilization processing on the initial correction weights to generate weight stabilization information and output the stabilized correction weights.

[0258] By retrieving the previous running version identifier from the fusion running record that matches the target application scenario and output caliber and the reference key, the stabilization correction weight corresponding to the previous running version identifier is read and recorded as the historical stabilization correction weight.

[0259] When no matching record is found, the initial correction weight is recorded as the historical stabilization correction weight.

[0260] Based on the comparison key, stage identifiers and event type identifiers are extracted. The gating thresholds that match the stage identifiers and event type identifiers are retrieved. The smoothing coefficient is determined based on the gating thresholds. The smoothing coefficient is limited to a range of 0 to 1. When the gating thresholds fail to match, the smoothing coefficient is set to 0.

[0261] Based on the comparison key, stage identifiers and event type identifiers are extracted, and a gate threshold for matching stage identifiers and event type identifiers is retrieved. The smoothing coefficient is determined using a piecewise mapping method. ;

[0262] Example values ​​for the smoothing coefficient: when hour, Write 0.2; when hour, Write 0.5; when hour, Write 0.8;

[0263] Smoothing coefficient The value range is limited to 0 to 1, when the gate threshold Smoothing coefficient when matching fails The value is 0.

[0264] The corresponding stabilization correction weight is calculated based on the historical stabilization correction weight, the initial correction weight, and the smoothing coefficient.

[0265] The stabilization correction weight is expressed as:

[0266] ;

[0267] In the formula, For comparison key Corresponding stabilization correction weights, For comparison key Corresponding to historical stabilization correction weights, This is the smoothing coefficient.

[0268] The stabilization correction weights corresponding to all reference keys are collected and output as stabilization correction weights.

[0269] S6: Utilize stabilization correction weights to perform real-time reweighted fusion on the batch digital thread index of the event set across stages, output the fusion result, and store it together with the collection configuration description, gating reason explanation and weight stabilization information in a versioned manner to generate a replayable fusion operation record.

[0270] S6.1: Utilize the output scope of the fusion query binding to determine the summary granularity and field set, and output summary constraints.

[0271] The output caliber field value is obtained from the fusion query-output caliber binding relationship record in the fusion query processing context, and the start time and end time are obtained from the fusion query-time range determination result record.

[0272] Based on the values ​​of the output caliber field, the list of caliber fields is read, and the summary granularity fields are determined to be stage identifier and event type identifier. The output field set is determined to be stage identifier, event type identifier, and event count.

[0273] The summary granularity field, the set of output fields, the start time, and the end time are aggregated and written into the summary constraint, and the summary constraint is output.

[0274] S6.2: Based on the batch number thread index, locate the event link corresponding to the batch identifier in the cross-stage event set, generate link extraction conditions, extract event records along the event jump relationship, sort them by the event occurrence time, and output the batch stage event sequence.

[0275] Based on each batch identifier in the batch number thread index, read the starting location information corresponding to the batch identifier, and write the batch identifier and the starting location information into the link extraction conditions.

[0276] For each link, extract the conditions and search the cross-stage event set for event records that match the start location information and whose event occurrence time falls within the range of start and end times, and determine the start event record of the batch.

[0277] The batch start event record is used as the current event record, and the location information of the subsequent event record corresponding to the current event record is retrieved in the index mapping relationship of the batch number thread index.

[0278] Search the cross-stage event set for event records that match the location information of subsequent event records and whose event occurrence time falls within the range of start and end times. Update the current event record and record the event record. Stop the extraction when the location information of subsequent event records is empty or the event record is empty.

[0279] Sort the extracted event records in ascending order by the event occurrence time, and output the event sequence of the batch corresponding to the batch identifier.

[0280] All batch stage event sequences corresponding to all batch identifiers are collected and output as a batch stage event sequence set.

[0281] S6.3: Establish a weight mapping relationship using stabilization correction weights, stage identifiers and event type identifiers, output weight assignment rules, write stabilization correction weights into each batch stage event sequence, and form a weighted event sequence.

[0282] The stabilization correction weights are merged according to the combination of stage identifier and event type identifier, and the corresponding stabilization correction weight values ​​are collected for each combination of stage identifier and event type identifier.

[0283] For each combination of stage identifier and event type identifier, read all the collected stabilization correction weight values, calculate the arithmetic mean of the stabilization correction weight values, and output the target stabilization correction weight corresponding to the combination of stage identifier and event type identifier.

[0284] Write the stage identifier, event type identifier, and target stabilization correction weight into the weight mapping record, and then aggregate all weight mapping records to generate a weight mapping relationship set.

[0285] The set of weight mapping relationships is written into the weight assignment rules. Each rule in the weight assignment rules includes a stage identifier field, an event type identifier field, and a stabilization correction weight field. The combination of the values ​​of the stage identifier field and the event type identifier field is used as the search condition to determine the stabilization correction weight value.

[0286] For each event record in the batch stage event sequence set, read the stage identifier and event type identifier, search the weight assignment rules for rule records that match both the stage identifier and event type identifier, and take the corresponding stabilization correction weight value.

[0287] The stabilization correction weight values ​​are appended to the event records and written to the weight field. All event records with the appended weight field are then aggregated and output as a weighted event sequence set.

[0288] S6.4: Perform real-time incremental aggregation on the weighted event sequence using aggregation constraints to generate fusion results.

[0289] Using the summary granularity field and output field set in the summary constraint, the weighted event sequence set is merged according to the stage identifier and event type identifier, and the stage identifier merge result and event type identifier are output.

[0290] The initial value of the event count is zero for the combination of stage identifier and event type identifier. Event records are read one by one in ascending order of event occurrence time and the stabilization correction weight value is accumulated to generate the event count corresponding to the combination of stage identifier and event type identifier.

[0291] The stage identifier, event type identifier, and event count are organized and output as a fusion result according to the set of output fields, and the fusion result is output to the output scope bound to the fusion query.

[0292] S7: Store the fusion results, acquisition configuration description, gating reason explanation and weight stabilization information in a versioned manner, and generate a fusion operation record that can be replayed.

[0293] The runtime version identifier is generated by combining the UTC millisecond timestamp of the fusion query reception time with the incrementing sequence number. The incrementing sequence number starts from 1 and increments within the same fusion query reception time.

[0294] The runtime version identifier is associated with the target application scenario, output caliber, fusion result, collection configuration description, gating reason explanation and weight stabilization information, respectively. The runtime version identifier is written into each associated record and used as the index information for retrieving the above associated records.

[0295] For each event record in the weighted event sequence set, record the batch identifier, stage identifier, event type identifier, event occurrence time, source identifier, and stabilization correction weight value, and write the running version identifier to generate a real-time reweighted fusion record. The running version identifier in the real-time reweighted fusion record is also used as index information.

[0296] All associated records and real-time reweighted fusion records are collected and stored together according to the running version identifier, generating a fusion running record set with the running version identifier as the index key, and outputting fusion running records that can be replayed.

[0297] In summary, this invention achieves unified field definitions and temporal semantics for multi-source real-time data by generating a unified event stream throughout the entire lifecycle and constructing a batch digital thread index to connect and associate cross-stage event records within the same batch in the context of electronic digital data processing. This forms a continuous and consistent batch cross-stage event chain, reduces broken links and missing associations, and makes it easier to verify and locate stage identifiers, event type identifiers, and event occurrence times under the same output definition. This reduces fluctuations in fusion results and improves interpretability, traceability, and replayability.

[0298] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A data fusion processing method for the entire vaccine engineering lifecycle, characterized in that: include, It receives fusion queries for the entire vaccine engineering lifecycle and binds them to target application scenarios and output standards. Based on the fusion queries, it accesses real-time data from multiple sources and performs unified preprocessing and encapsulation to generate a unified event stream for the entire lifecycle. A batch digital thread index is constructed based on a unified event stream throughout the entire lifecycle, and the events throughout the entire lifecycle of the batch are linked together to generate a cross-stage event set. Inject collection configuration descriptions into cross-stage event sets, complete the verification and correction of collection configuration information through an interactive wizard, and generate an event set and configuration profile base with collection configuration descriptions. For each fusion query, the configuration difference between the source-side configuration profile and the target-side configuration profile is calculated based on the configuration profile, and transferable gating control information and gating reason explanation are generated. Based on the constraints of transferable gating control information, a subset of events matching the target application scenario is selected and a target reference and source comparison dataset is constructed to generate stable correction weights. Real-time reweighted fusion of cross-stage event sets is performed on the batch digital thread index using stabilization correction weights, and the fusion result is output. The fusion results, acquisition configuration descriptions, gating reason explanations, and weight stabilization information are versioned and stored to generate a replayable fusion operation record.

2. The data fusion processing method for the entire vaccine engineering lifecycle as described in claim 1, characterized in that: The specific steps for generating a unified event stream throughout its entire lifecycle are as follows. Based on the fusion query, the target application scenario and output scope are analyzed to determine the source set and time range of multi-source real-time data, a multi-source access list is generated, multi-source real-time data is accessed according to the multi-source access list and the source identifier and collection time metadata are written, and data with metadata is output. The data with metadata is standardized by performing field caliber consistency and time semantic consistency to generate standardized data. Then, a unified preprocessing encapsulation is performed through a unified event field structure to output a unified event stream with full lifecycle, including batch identifier, stage identifier, event type identifier, event occurrence time and source identifier.

3. The data fusion processing method for the entire vaccine engineering lifecycle as described in claim 1, characterized in that: The specific steps for constructing a batch digital thread index and generating a cross-stage event set based on a unified event stream throughout the entire lifecycle are as follows. Based on the unified event flow throughout the entire lifecycle, batch identifiers and stage identifiers are extracted, an index key set is generated, and the event records are sorted according to the event occurrence time to output the batch stage event sequence. For the event sequence in the batch stage, use the event type identifier to establish the event jump relationship and generate the batch number thread index; By using batch number thread indexes to connect and associate batch identifiers with full-cycle event records in the full-cycle unified event stream, a cross-stage event set is generated.

4. The data fusion processing method for the entire vaccine engineering lifecycle as described in claim 1, characterized in that: The specific steps for generating the event set and configuration profile with collection configuration description are as follows. Based on the cross-stage event set, basic event entries are constructed, source traceability information and quality tags are generated, and the basic event entry set is organized into event source entries and collection context entries. The event source entries and collection context entries are merged using event type identifiers and stage identifiers to generate a draft collection configuration description. By collecting the initial draft of the configuration description, the items to be verified for the configuration information are output, and the interactive wizard is used to generate the verification task. Receive verification input for the verification task, associate the verification input with the source traceability information, record and update the initial draft of the collection configuration description, and generate a collection configuration description that has passed verification. The verified collection configuration descriptions are backfilled into the cross-stage event set, the event set with collection configuration descriptions is output, and the event set with collection configuration descriptions and collection configuration information are aggregated to generate the basic configuration profile.

5. The data fusion processing method for the entire vaccine engineering lifecycle as described in claim 1, characterized in that: The specific steps for generating transferable gating control information and gating reason explanations are as follows: For each fusion query, the target application scenario and output scope are read, and the profile extraction list containing source-side scope items and target-side scope items is generated by associating it with the configuration profile base. Based on the source-side scope entries and target-side scope entries in the image extraction list, extract the source-side configuration image and target-side configuration image from the configuration image base to generate a dual-side configuration image. Perform item-level comparison on the two-sided configuration profiles to generate a set of difference items. Associate the set of difference items with the stage identifier and event type identifier of the cross-stage event set to generate a description of the impact of the difference. The difference impact description is used to perform gating determination on the transferability of differences, generate transferable gating control information, and summarize and organize it with the difference impact description to generate gating cause explanations.

6. The data fusion processing method for the entire vaccine engineering lifecycle as described in claim 1, characterized in that: The specific steps for filtering and matching the event subset of the target application scenario based on the transferable gating control information constraints and constructing the target reference and source comparison dataset are as follows. By utilizing transferable gating control information, gating and filtering are performed on the cross-stage event set to generate an event subset that matches the target application scenario. Then, field alignment and time aggregation are performed on the event subset according to the output caliber to generate the target reference dataset. Based on the output criteria, perform field alignment and time aggregation on event records in the cross-stage event set that did not enter the event subset to generate a source comparison dataset; Perform stratified alignment of the target reference dataset and the source control dataset by the control key, generate a set of correction factors and output the initial correction weights; Perform weight stabilization processing on the initial correction weights to generate weight stabilization information, and output the stabilized correction weights.

7. The data fusion processing method for the entire vaccine engineering lifecycle as described in claim 1, characterized in that: The gating cause explanation refers to the process of summarizing configuration differences by using stage identifiers and event type identifiers to generate entries, and associating each configuration difference entry with a description of the difference's impact and the corresponding transferable gating control information.

8. The data fusion processing method for the entire vaccine engineering lifecycle as described in claim 1, characterized in that: The specific steps for performing real-time reweighted fusion of cross-stage event sets on the batch digital thread index using stabilization correction weights and outputting the fusion result are as follows. By utilizing the output scope of the fusion query binding, the summary granularity and field set are determined, and the summary constraints are output. Based on the batch digital thread index, locate the event link corresponding to the batch identifier in the cross-stage event set, generate link extraction conditions, extract event records along the event jump relationship, and output the batch stage event sequence in sorted by the event occurrence time. By establishing a weight mapping relationship using stabilization correction weights, stage identifiers, and event type identifiers, and generating weight assignment rules, the stabilization correction weights are written into each event sequence of the batch stage to generate a weighted event sequence. Real-time incremental aggregation is performed on the weighted event sequence using aggregation constraints to generate fusion results.

9. The data fusion processing method for the entire vaccine engineering lifecycle as described in claim 1, characterized in that: The specific steps for generating a replayable fusion operation record are as follows: Generate runtime version identifiers from fusion queries; Based on the running version identifier, the target application scenario, output caliber and fusion result are associated and stored, and index information is generated for the collection configuration description, gating reason explanation, weight stabilization information and the weighted event sequence corresponding to the batch digital thread index. The index information is associated with the running version identifier to perform versioned storage, generating a replayable fusion running record.

10. The data fusion processing method for the entire vaccine engineering lifecycle as described in claim 1, characterized in that: The gating determination refers to generating a gating score based on the difference impact description, using the effective proportion of the target side configuration profile as the gating threshold, and writing "allow" in the transferable gating control information when the gating score is not greater than the gating threshold, and writing "prohibit" in the transferable gating control information when the gating score is greater than the gating threshold.

Images