Intelligent acquisition and fusion management method for multi-source heterogeneous data

Abstract

The application belongs to the technical field of data fusion processing, and particularly relates to an intelligent acquisition and fusion management method and system for multi-source heterogeneous data. The method comprises the following steps: classifying and sub-class constructing the multi-source heterogeneous data, identifying and distinguishing the structural and attribute differences among the data; taking the data of different sub-classes in the same category as convergence centers respectively, and matching and evaluating the convergence centers, calculating the matching probability and matching interval, taking the matching probability and matching interval as the judgment basis, and generating the unique fusion node under the category; and through accuracy verification on the fusion result, the fusion result is monitored and self-adaptively corrected in real time and intelligently. The application generates the fusion node based on the matching probability and fluctuation range of the data characteristics, improves the generation efficiency and accuracy of the data fusion node, avoids the deviation of the fusion result caused by the data quality problem, and improves the accuracy, reliability and stability of the data fusion result.

Description

Technical Field

[0001] This invention belongs to the field of data fusion and processing technology, specifically relating to an intelligent acquisition and fusion management method and system for multi-source heterogeneous data. Background Technology

[0002] With the application of artificial intelligence technology in the smart manufacturing industry, data is continuously generated in the smart manufacturing process. By effectively utilizing the diverse data generated in the production line, intelligent decision-making can be achieved, production processes can be optimized, and core competitiveness can be enhanced. Therefore, how to efficiently and accurately collect, integrate, and manage heterogeneous data from different sources, formats, and semantics has become a challenge faced in the current smart manufacturing process.

[0003] Existing technologies still have limitations in the intelligent acquisition and fusion governance of multi-source heterogeneous data. Common data fusion methods are mostly based on simulation weighting or preset rules, and usually rely on empirical parameters or static models. They are difficult to adapt to the dynamic changes and complexity of data sources, and cannot guarantee fusion accuracy. This leads to inherent biases in the fusion results, affecting the reliability of data analysis and decision-making. When data patterns drift, outliers appear, or new data sources are added, existing technologies are difficult to adjust the fusion strategy. Model reconstruction or parameter tuning is required, which not only increases operation and maintenance costs, but also limits the real-time response capability and robustness of data processing.

[0004] Furthermore, in the face of the explosive growth in data volume in scenarios such as Industrial Internet of Things and smart manufacturing, the centralized processing mode is unable to meet the real-time requirements due to data transmission latency, bandwidth pressure and edge computing resource limitations. This results in data not being able to be transformed into valuable information in a timely and effective manner, affecting the timeliness and accuracy of applications such as fault diagnosis and predictive maintenance, and easily causing production interruptions or safety hazards.

[0005] To address the aforementioned issues, this application provides an intelligent acquisition and fusion governance method and system for multi-source heterogeneous data. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent acquisition and fusion management method and system for multi-source heterogeneous data, in order to solve the problem in the prior art where the accuracy and reliability of data correction results decrease because the correspondence between data is not considered during the data correction process after data structure standardization.

[0007] To achieve the above-mentioned objectives, the technical solution adopted by the present invention is as follows:

[0008] A method for intelligent acquisition and fusion governance of multi-source heterogeneous data includes the following steps:

[0009] When multi-source heterogeneous data is detected, the multi-source heterogeneous data is classified and organized to obtain subclasses;

[0010] Based on subclasses, similarity is determined, convergence centers are generated, and fusion nodes are identified.

[0011] Based on the fusion node, data fusion processing is performed according to the fusion rules to obtain the fusion result;

[0012] The fusion results are verified and updated based on a preset verification sample set, and a replaceable fusion result is output.

[0013] The process of verifying and updating the fusion result based on a preset verification sample set and outputting a replaceable fusion result includes: calculating the total verification score of the fusion result and the preset verification sample set; comparing the total verification score with a score threshold; if the total verification score is lower than the score threshold, revising the fusion result to obtain a replaceable fusion result; and outputting the replaceable fusion result.

[0014] Among them, classifying and organizing multi-source heterogeneous data to obtain subclasses includes: performing structural classification and attribute classification based on the structural and attribute features of multi-source heterogeneous data, thereby improving classification accuracy and recognition accuracy.

[0015] Preferably, the step of performing similarity determination based on subclasses, generating convergence centers, and determining fusion nodes includes:

[0016] Using similarity analysis tools, subclasses are compared to determine the matching relationships between them, and convergence centers are generated based on these matching relationships.

[0017] Preferably, the step of performing similarity determination based on subclasses, generating convergence centers, and determining fusion nodes further includes:

[0018] Based on the importance weights and differences of attributes, a weighted summation is performed to obtain the total similarity between the fusion result and the verification sample set, and the fusion node is determined based on the total similarity.

[0019] Preferably, the total verification score of the calculation fusion result and the preset verification sample set includes:

[0020] The attribute similarity score and attribute importance weight of each data in the preset verification sample set are weighted and summed to obtain the score of a single sample. The scores of all single samples are then normalized to obtain the total verification score.

[0021] Preferably, comparing the total verification score with a scoring threshold, and revising the fusion result to obtain a replaceable fusion result if the total verification score is lower than the scoring threshold, includes:

[0022] Identify the evaluation revision areas in the fusion results that are in a pending confirmation state, and replace the data in the evaluation revision areas based on a preset verification sample set to generate replaceable fusion results.

[0023] An intelligent acquisition and fusion governance system for multi-source heterogeneous data includes the following modules:

[0024] The data acquisition and classification module is used to classify and organize multi-source heterogeneous data and obtain subcategories when multi-source heterogeneous data is detected.

[0025] The subclass identification and node generation module is used to perform similarity determination based on subclasses, generate convergence centers, and determine fusion nodes;

[0026] The data fusion execution module is used to perform data fusion processing based on fusion nodes and fusion rules, and to obtain fusion results.

[0027] The result verification and update module is used to verify and update the fusion results based on a preset verification sample set, and output replaceable fusion results.

[0028] Preferably, the step of verifying and updating the fusion result based on a preset verification sample set and outputting a replaceable fusion result includes: calculating the total verification score of the fusion result and the preset verification sample set; comparing the total verification score with a scoring threshold; if the total verification score is lower than the scoring threshold, revising the fusion result to obtain a replaceable fusion result; and outputting the replaceable fusion result.

[0029] Preferably, the step of classifying and organizing multi-source heterogeneous data to obtain subclasses includes: performing structural classification and attribute classification based on the structural and attribute features of the multi-source heterogeneous data, thereby improving classification accuracy and recognition accuracy.

[0030] Preferably, the step of performing similarity determination based on subclasses, generating convergence centers, and determining fusion nodes includes:

[0031] Using similarity analysis tools, subclasses are compared to determine the matching relationships between them, and convergence centers are generated based on these matching relationships.

[0032] Preferably, the step of performing similarity determination based on subclasses, generating convergence centers, and determining fusion nodes further includes:

[0033] Based on the importance weights and differences of attributes, a weighted summation is performed to obtain the total similarity between the fusion result and the verification sample set, and the fusion node is determined based on the total similarity.

[0034] Beneficial effects

[0035] This invention classifies and organizes multi-source heterogeneous data, extracts data structure and attributes, filters each type of data based on a subclass identification model, calculates similarity values ​​and compares them with a preset similarity threshold to determine whether the filtering results are comparable and belong to the same subclass or are retained as independent objects. This process classifies and constructs subclasses for multi-source heterogeneous data, identifies and distinguishes the structural and attribute differences between data, and improves the accuracy and precision of data classification.

[0036] This invention uses data from different subcategories within the same category as aggregation centers, performs matching evaluations between aggregation centers, calculates matching probabilities and matching intervals, and uses matching probabilities and matching intervals as the basis for judgment to generate unique fusion nodes under each category. The fusion nodes are generated based on the matching probabilities and fluctuation ranges of data features, solving the problem of matching difficulties caused by large data differences during the fusion process of multi-source heterogeneous data, and improving the generation efficiency and accuracy of data fusion nodes.

[0037] This invention verifies the accuracy of the fusion results by calculating the similarity between the fusion results and a preset verification sample set. If the similarity is lower than the scoring threshold, the evaluation revision area is marked, and a replacement fusion result is regenerated in the area and the original fusion result is updated. This provides real-time and intelligent quality monitoring and adaptive correction of the fusion results, avoiding deviations in the fusion results caused by data quality issues and improving the accuracy, reliability, and stability of the data fusion results. Attached Figure Description

[0038] Figure 1 This is a flowchart of the method of the present invention;

[0039] Figure 2 This is a system module diagram of the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the scope of protection of the invention.

[0041] Example 1

[0042] Please refer to Figure 1 This embodiment provides an intelligent acquisition and fusion governance method for multi-source heterogeneous data, including the following steps:

[0043] S1. Data Acquisition Classification: Extract the data content to be processed from multiple independent data acquisition sources such as sensors, external interfaces, and third-party platforms. The collected data includes structured data, semi-structured data, and unstructured data. Among them, structured data includes tables, logs, etc., semi-structured data includes XML, JSON files, etc., and unstructured data includes text, images, etc.

[0044] When multi-source heterogeneous data is detected, the multi-source heterogeneous data is collected. After the collection is completed, the data access mechanism is activated to perform unified data structure parsing and data attribute extraction on all data content.

[0045] By parsing data structures, structural differences such as field order, nesting level, and naming style between different data can be identified. Specifically, the original data can be scanned and pattern matched using a set of pre-defined parsing rules, such as parsers for different data formats, pattern matching rules, and data field mapping specifications, in order to identify and standardize its internal organization.

[0046] Data attribute extraction is used to identify, extract and mark key information such as data generation time, unit, and data source identifier from data. It uses techniques such as keyword matching, regular expressions or semantic analysis to identify data attribute values ​​with specific meanings from the data content.

[0047] Based on the differences in data structure reflected in the data structure parsing results, such as data organization, field definition, nesting level, data type, or naming convention, and the differences in data attribute reflected in the data attribute extraction results, such as the values, formats, or semantics of key attributes like data generation time, unit, and data source identifier, all data are divided into multiple data classification sets. Specifically, data with the same data structure and data attribute characteristics are grouped into the same data classification set, while data with significant differences in data structure or data attributes are separated into different data classification sets.

[0048] S2, Subclass Identification and Node Generation: For each data category set, a filtering process is performed to construct more refined subclasses;

[0049] For each independent data object in the data classification set, a pre-set subclass identification model and feature comparison mechanism are used to conduct in-depth analysis of its data structure and data attributes, evaluate the subtle features of the data object at the structural and attribute levels, and classify it into a specific subclass based on these features;

[0050] The pre-defined subclass identification model and feature comparison mechanism specifically include: for structured data, it can compare its field types, data range, constraints, etc.; for unstructured data, it can analyze its content theme, keyword distribution, contextual relationships, etc.; generate a subclass identification result for each data object indicating the subclass to which the data object is most likely to belong; calculate the similarity value between the data object and other members in the subclass in terms of structure and attributes, and compare the calculated similarity value with a pre-defined similarity threshold to determine the subclass to which the data object belongs;

[0051] The similarity value can be calculated using cosine similarity, which is obtained by calculating the cosine similarity between the feature vectors of the data object and other members in the subclass; the preset similarity threshold can be set to 0.8.

[0052] If the calculated similarity value is higher than the preset similarity threshold, the data object is determined to have a sufficiently high similarity in terms of structure and attribute features, and is comparable to the subclass, and is included in the subclass.

[0053] If the similarity value is lower than the preset similarity threshold, the data object will be regarded as an independent object, temporarily not belonging to any existing subclass, and marked as pending further analysis.

[0054] S3. Generation and analysis of the aggregation center: For each subclass that has been constructed, the aggregation center representing the core characteristics of the subclass is generated by aggregating the structural and data attribute characteristics of all data objects in the subclass.

[0055] The aggregation center is an abstract data structure template formed by aggregating the structural and attribute characteristics of all data objects within a specific subclass. It includes common fields, data types, value ranges, and statistical characteristics of key attributes such as average, mode, and common units shared by all data objects in that subclass.

[0056] The data marked as independent objects are processed. For each independent object, it is compared with all the generated aggregation centers one by one. Based on the similarity between the independent object and each aggregation center in terms of data structure and data attributes, the matching probability between the independent object and each aggregation center is calculated. The matching probability represents the possibility that the independent object belongs to that aggregation center.

[0057] Specifically, the matching probability is calculated by comparing field names, data type compatibility, and overlap of data attribute value distributions. The calculated comparisons of field names, data type compatibility, and overlap of data attribute value distributions are weighted. If the matching probability of an independent object with a certain aggregation center falls within a preset matching range, the independent object is considered to have a potential association with that aggregation center and is temporarily classified into the subclass corresponding to that aggregation center, pending further confirmation. The preset matching range can be set to 0.7-0.9.

[0058] If the matching probability is lower than the preset matching range, the independent object will remain in an independent state and will be recorded for manual review or subsequent iterative processing.

[0059] S4. Output fusion node: Integrate all data that has formed subcategories, including those that are directly categorized and those that are temporarily categorized, to generate a fusion node. The fusion node is a logically unified data entity. The fusion node is a common data structure, key attributes, and complete attribute set of the subcategories that have formed subcategories but are from different data sources but have high similarity.

[0060] The process of generating a fusion node includes: creating a unique identifier for each subclass as the name of the fusion node; using the common data structure and key data attributes of all data objects in the subclass as the basic structure and data attribute definition of the fusion node; and selectively retaining or merging non-key data attributes that differ among subclasses based on the majority principle to form a complete set of data attributes for the fusion node.

[0061] A backup fusion scheme is generated for each fusion node. The backup fusion scheme records the data segments that were discarded or not fully matched during the fusion process and their original sources, so that they can be backtracked or replaced in case of subsequent verification failure.

[0062] S5. Perform data fusion: Based on the generated fusion nodes, perform data fusion operations, apply preset fusion rules to process the data within the fusion nodes, and integrate heterogeneous data into a unified, standardized and high-quality fusion result.

[0063] The fusion rules include: data deduplication rules, data cleaning rules, data transformation rules, and data aggregation rules. Through the processing of these fusion rules, heterogeneous data from different data sources are integrated into a unified, standardized, and high-quality fusion result.

[0064] Among them, the data deduplication rule identifies and removes duplicate records based on the primary key or unique identifier; the data cleaning rule handles missing values, outliers, inconsistent formats, etc.; the data transformation rule unifies data from different units into standard units and unifies data from different encoding formats; and the data aggregation rule sums, averages, or counts data with the same data attributes.

[0065] S6. Result verification and update: After the fusion result is generated, the result will be strictly verified to ensure its accuracy and consistency.

[0066] The verification process includes: comparing the fusion result with a preset verification sample set, which contains known correct data patterns and business logic to evaluate the accuracy and quality of the fusion result; and calculating the total verification score between the fusion result and the preset verification sample set, which quantifies the degree of deviation between the fusion result and the standard used as the preset verification sample set.

[0067] The calculation of the total verification score includes: calculating the attribute similarity score between each data point in the fusion result and each single sample in the preset verification sample set; assigning attribute importance weights to each attribute similarity score based on experience; weighting and summing the attribute similarity scores and attribute importance weights to obtain the score of a single sample; and normalizing the scores of all single samples to obtain the total verification score, where the total verification score = (sum of scores of all single samples) / number of samples.

[0068] The attribute similarity score calculation includes: for each data record in the fusion result, finding the corresponding standard data record in the preset verification sample set, comparing each attribute value of the two records one by one, and calculating the attribute similarity score. For numerical attributes, the attribute similarity score = 1 - |fusion value - verification value| / verification value range; for enumerated or textual attributes, if the value matches completely, the attribute similarity score is 0, and if the value does not match, the attribute similarity score is 1.

[0069] A preset scoring threshold is set, which in this embodiment can be set to 0.7. If the total verification score does not exceed the preset scoring threshold, it indicates that there is a problem with the fusion result and it needs to be revised.

[0070] If the total verification score exceeds the preset score threshold, it indicates that there is a problem with the fusion result and it needs to be revised. Analyze the specific data records and data attributes that caused the low score and mark them as evaluation and revision areas in the fusion result that are in a state of pending confirmation. Specifically, it can be a data attribute value of a data record or a set of multiple data records. Specifically, specific data records and data attributes with similar attributes and scores lower than the preset score threshold can be found as evaluation and revision areas.

[0071] Based on the overall verification score, the evaluation revision area in the fusion result is identified, and an attempt is made to call the backup fusion scheme generated for that fusion node to replace it; if the backup fusion scheme can reduce the overall verification score to below the score threshold, then the backup scheme is adopted as the replaceable fusion result;

[0072] If the alternative solution cannot solve the problem, or if there is no available alternative solution, the fusion result will be marked as pending manual review, and a detailed verification report will be recorded so that manual intervention can be performed for analysis and correction.

[0073] Example 2

[0074] Please refer to Figure 2 This embodiment provides an intelligent acquisition and fusion governance system for multi-source heterogeneous data. It can be deployed in computing environments including server clusters, cloud computing platforms, or distributed data processing systems. It communicates and exchanges data with various data sources such as databases, API interfaces, file systems, and streaming data platforms through standard data interfaces. The system includes the following modules:

[0075] The data acquisition and classification module is used to classify and organize multi-source heterogeneous data when it is detected, in order to obtain subclasses; when multi-source heterogeneous data with different sources, different formats and semantic features are detected to flow in, the data access mechanism is activated to introduce these data into the system; based on the data structure and data attributes of multi-source heterogeneous data, data structure parsing and data attribute extraction are performed respectively.

[0076] For data structure, identify whether the data is a relational database table, JSON document, XML file, or text log, etc.; for data attributes, analyze the field names, data types, value ranges, semantic tags, etc. of the data.

[0077] By employing a dual classification mechanism, the accuracy of classification and recognition is improved. The original heterogeneous data is effectively organized into subcategories with similar characteristics. All data describing customer information but from different sources are grouped into one subcategory, and all data describing product sales records but in different formats are grouped into another subcategory.

[0078] The subclass identification and node generation module is used to respond to the classification results of the data acquisition and classification module, perform similarity determination based on subclasses, generate convergence centers and determine fusion nodes;

[0079] The subclass identification and node generation module includes the following units:

[0080] The subclass identification unit is used to identify subclasses based on a preset similarity threshold; it performs preliminary analysis on each subclass output by the data acquisition and classification module to identify which subclasses may have potential associations or similarities, and provides a candidate set for similarity determination;

[0081] The classification judgment unit is used to judge the comparability of subclasses; before performing similarity judgment, it assesses whether there are comparable dimensions or attributes between different subclasses to avoid invalid similarity calculations for subclasses that are not comparable.

[0082] The similarity comparison unit is used to compare subclasses using a feature comparison mechanism, determine the matching relationship between subclasses, and generate convergence centers based on the matching relationship. It uses multiple algorithms such as text similarity, structural similarity, or semantic similarity to compare data entities or attributes in different subclasses. When highly similar or matching subclasses are identified, it is found that customer ID and customer number refer to the same concept in different subclasses. Based on these matching relationships, subclasses with common semantics or logical connections are aggregated to generate convergence centers that logically represent the core theme or entity of the data to be merged.

[0083] The attribute difference analysis unit is used to analyze the degree of difference between subclasses based on the importance weight of attributes. In the process of determining matching relationships and convergence centers, the importance of different attributes is considered. Higher weights are given to the differences of key attributes such as primary keys and core business attributes, while lower weights are given to the differences of secondary attributes, so as to more accurately assess the differences between subclasses.

[0084] Based on the importance weight and the degree of difference of attributes, the total similarity between the fusion result and the preset verification sample set is obtained by weighted summation, and the fusion node is determined based on the total similarity. When determining which data entities or attributes should be fused, the similarity between subclasses, attribute differences and preset fusion rules are comprehensively considered to determine the specific fusion node that indicates the specific goal and method of data fusion operation.

[0085] The data fusion execution module is used to respond to the determination results of the subclass identification and node generation module, perform data fusion processing based on the fusion nodes and fusion rules, and obtain the fusion results. When the subclass identification and node generation module determines the fusion nodes and the aggregation center, the data fusion execution module performs actual fusion operations on the data from different subclasses according to the preset fusion rules.

[0086] The fusion rules include: data deduplication rules, conflict resolution rules, data completion rules, and data transformation rules. For customer name attributes that exist in both subclasses, if the fusion rules stipulate that the latest updated data shall prevail, the corresponding update operation shall be performed. By performing the fusion process, heterogeneous data is transformed into a consistent and high-quality fusion result.

[0087] The result verification and update module is used to respond to the fusion result of the data fusion execution module, verify and update the fusion result based on the preset verification sample set, and output a replaceable fusion result.

[0088] After receiving the fusion result output by the data fusion execution module, the result verification and update module will start the verification and update process; calculate the total verification score of the fusion result and the preset verification sample set; where the preset verification sample set is a high-quality reference dataset that has been verified by humans or authoritative systems and is used to evaluate the accuracy and completeness of the fusion result.

[0089] The calculation of the total verification score includes: weighting and summing the attribute similarity score and attribute importance weight of each data in the preset verification sample set to obtain the score of a single sample, and normalizing the scores of all single samples to obtain the total verification score.

[0090] For a customer data set in the validation sample set, it is compared with the corresponding customer data in the fusion result. The similarity of attributes such as name, address, and phone number is calculated, and different weights are assigned according to the importance of these attributes to finally obtain the score of the customer data.

[0091] The total verification score is compared with the scoring threshold. If the total verification score is lower than the scoring threshold, it indicates that the quality of the fusion result does not meet the expected standard and needs to be revised.

[0092] Revising the fusion results includes: identifying evaluation revision areas in the fusion results that are in a pending confirmation state, and replacing the data in the evaluation revision areas based on a preset verification sample set to generate replaceable fusion results; if the address information of a customer in the fusion results does not match the verification sample set, and the area is marked as an evaluation revision area, then the correct address information in the fusion results is used to replace the erroneous information in the fusion results, thereby generating a revised and higher-quality replaceable fusion result.

[0093] The above description is merely a preferred embodiment of this application and is not intended to limit this application. For those skilled in the art, this application can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for intelligent acquisition and fusion governance of multi-source heterogeneous data, characterized in that, Includes the following steps: When multi-source heterogeneous data is detected, the multi-source heterogeneous data is classified and organized to obtain subclasses; Based on subclasses, similarity is determined, convergence centers are generated, and fusion nodes are identified. Based on the fusion node, data fusion processing is performed according to the fusion rules to obtain the fusion result; The fusion results are verified and updated based on a preset verification sample set, and a replaceable fusion result is output. The process of verifying and updating the fusion result based on a preset verification sample set and outputting a replaceable fusion result includes: calculating the total verification score of the fusion result and the preset verification sample set; comparing the total verification score with a score threshold; if the total verification score is lower than the score threshold, revising the fusion result to obtain a replaceable fusion result; and outputting the replaceable fusion result. Among them, classifying and organizing multi-source heterogeneous data to obtain subclasses includes: performing structural classification and attribute classification based on the structural and attribute features of multi-source heterogeneous data, thereby improving classification accuracy and recognition accuracy.

2. The intelligent acquisition and fusion governance method for multi-source heterogeneous data according to claim 1, characterized in that, The process of determining similarity based on subclasses, generating convergence centers, and identifying fusion nodes includes: Using similarity analysis tools, subclasses are compared to determine the matching relationships between them, and convergence centers are generated based on these matching relationships.

3. The intelligent acquisition and fusion management method for multi-source heterogeneous data according to claim 2, characterized in that, The process of determining similarity based on subclasses, generating convergence centers, and identifying fusion nodes also includes: Based on the importance weights and differences of attributes, a weighted summation is performed to obtain the total similarity between the fusion result and the verification sample set, and the fusion node is determined based on the total similarity.

4. The intelligent acquisition and fusion management method for multi-source heterogeneous data according to claim 1, characterized in that, The total verification score of the calculation fusion result and the preset verification sample set includes: The attribute similarity score and attribute importance weight of each data in the preset verification sample set are weighted and summed to obtain the score of a single sample. The scores of all single samples are then normalized to obtain the total verification score.

5. The intelligent acquisition and fusion management method for multi-source heterogeneous data according to claim 1, characterized in that, The step of comparing the total verification score with a scoring threshold, and revising the fusion result if the total verification score is lower than the scoring threshold, to obtain a replaceable fusion result includes: Identify the evaluation revision areas in the fusion results that are in a pending confirmation state, and replace the data in the evaluation revision areas based on a preset verification sample set to generate replaceable fusion results.

6. A smart acquisition and fusion management system for multi-source heterogeneous data, characterized in that, Includes the following modules: The data acquisition and classification module is used to classify and organize multi-source heterogeneous data and obtain subcategories when multi-source heterogeneous data is detected. The subclass identification and node generation module is used to perform similarity determination based on subclasses, generate convergence centers, and determine fusion nodes; The data fusion execution module is used to perform data fusion processing based on fusion nodes and fusion rules, and to obtain fusion results. The result verification and update module is used to verify and update the fusion results based on a preset verification sample set, and output replaceable fusion results.

7. The intelligent acquisition and fusion management system for multi-source heterogeneous data according to claim 6, characterized in that, The process of verifying and updating the fusion result based on a preset verification sample set and outputting a replaceable fusion result includes: calculating the total verification score of the fusion result and the preset verification sample set; comparing the total verification score with a scoring threshold; if the total verification score is lower than the scoring threshold, revising the fusion result to obtain a replaceable fusion result; and outputting the replaceable fusion result.

8. The intelligent acquisition and fusion management system for multi-source heterogeneous data according to claim 6, characterized in that, The process of classifying and organizing multi-source heterogeneous data to obtain subclasses includes: performing structural classification and attribute classification based on the structural and attribute features of the multi-source heterogeneous data, thereby improving classification accuracy and recognition accuracy.

9. The intelligent acquisition and fusion management system for multi-source heterogeneous data according to claim 6, characterized in that, The process of determining similarity based on subclasses, generating convergence centers, and identifying fusion nodes includes: Using similarity analysis tools, subclasses are compared to determine the matching relationships between them, and convergence centers are generated based on these matching relationships.

10. The intelligent acquisition and fusion management system for multi-source heterogeneous data according to claim 9, characterized in that, The process of determining similarity based on subclasses, generating convergence centers, and identifying fusion nodes also includes: Based on the importance weights and differences of attributes, a weighted summation is performed to obtain the total similarity between the fusion result and the verification sample set, and the fusion node is determined based on the total similarity.

Images