Classification method for digital model generation data

By constructing multi-dimensional data space and feature extraction technology in digital models, the problem of inefficient data classification of multi-source points is solved, and more refined and related data classification results are achieved.

CN119004187AInactive Publication Date: 2024-11-22安徽博盾信息科技有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411257949.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-09
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

In the process of massive data classification at multi-source points, due to the complexity of the data content, it is difficult for the existing technology to achieve reasonable multi-dimensional data classification, resulting in inefficient classification.

Method used

By constructing a multi-dimensional data space in a digital model, the data is classified first-level, and through feature extraction and orthogonal vector mapping, a covariance matrix is ​​generated for dimensionality reduction, further refine data classification and improve the correlation of classification.

Benefits of technology

The refinement classification of multi-source point data is realized, which improves the correlation between data classification efficiency and results, so that data within the same orthogonal vector has stronger correlation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004187A_ABST
    Figure CN119004187A_ABST
Patent Text Reader

Abstract

The invention discloses a classification method for digital model generation data, and relates to the technical field of data classification, and the method comprises the steps: building a digital model, setting a multi-dimensional data space in the digital model, and carrying out the primary classification of data from multiple source points according to the multi-dimensional data space; feature extraction is carried out on the data, the extracted features are mapped to orthogonal vectors corresponding to all dimensions in a multi-dimensional data space, and a covariance matrix corresponding to each orthogonal vector is obtained according to the features contained in each orthogonal vector; and according to the characteristic vector matrix corresponding to the obtained covariance matrix, obtaining the dimension-reduced value corresponding to the collaborative variance matrix, thereby completing further classification of the data input by each data source point, enabling the classification of the multi-source point data to be more detailed, enabling the classification results of the data to be more correlated, and improving the classification efficiency. The data is the relevance between the data in the same orthogonal vector.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data classification, and in particular to a classification method for data generated by a digital model. Background Art

[0002] With the development of computer science and information science, various enterprises and units pay more and more attention to the construction of information systems. Various information systems are gradually improved and generate massive amounts of data in daily operation. The data generated by multiple different information systems have different organizational forms and structures, and even generate tables with unclear meanings or temporary tables.

[0003] In the process of classifying massive data from multiple sources, it may be impossible to classify the data reasonably due to the complexity of the data content. How to establish a multi-dimensional data classification method based on the data content to improve the efficiency of data classification is a problem we need to solve. To this end, a classification method for data generated by a digital model is now provided. Summary of the invention

[0004] The purpose of the present invention is to provide a classification method for digital model generation data.

[0005] The object of the present invention can be achieved by the following technical solution: A classification method for digital model generation data, comprising the following steps:

[0006] Step 1: authenticate the data source point and aggregate the data uploaded by the data source point to obtain a sample data set;

[0007] Step 2: Input the sample data set into the digital model, and complete the classification of the data in the sample data set through the digital model;

[0008] Step 3: Import the classified data to a designated location for storage.

[0009] Furthermore, the process of authenticating the data source point includes:

[0010] Setting an authority verification unit to verify the identity of the data source point through the authority verification unit;

[0011] Each data source point is set with a corresponding identity recognition sequence; after completing the personal information registration in the data source point, you can upload data in the data source point class, establish a user information sub-library in the data source point, and import the personal basic information of the registered basic users into the user information sub-library, and upload the identity recognition sequence corresponding to the data source point to the authority verification unit for filing. The data source point that has been filed will obtain the authority to upload the source data.

[0012] Furthermore, the process of completing the registration of personal information at the data source point includes:

[0013] The data source point is provided with a registration unit and a login unit for basic users to register their identities and enter the data source point;

[0014] Inputting the basic user's personal basic information through the registration unit, and the data source point auditing the inputted personal basic information;

[0015] After the review is passed, a login account and password will be generated based on the mobile phone number in the personal basic information entered, and the generated login account and password will be sent to the corresponding mobile terminal;

[0016] The basic user obtains the authority to upload data into the data source point by inputting the obtained login account and login password into the login unit.

[0017] Furthermore, the process of obtaining the sample data set includes:

[0018] The data uploaded by each data source point is marked and recorded as the original data; the obtained original data is bound to the identity identification sequence of the corresponding data source point to form the source data; the source data is converted into the corresponding data stream, and the data streams obtained from each data source point are summarized to form a sample data set.

[0019] Furthermore, the process of classifying the data in the sample data set by the digital model includes:

[0020] The obtained sample data set is input into the digital model as input, and all data streams in the sample data set are read;

[0021] An N-dimensional data space is provided in the digital model, and a primary data label is provided corresponding to each dimension of the data space;

[0022] Based on each primary data label, an M-dimensional orthogonal vector associated with the corresponding primary data label is established;

[0023] According to the data source points corresponding to the data streams in the sample data set, the data streams in the sample data set are mapped to the corresponding dimensions in the data space, and a primary classification result is generated;

[0024] Output the first-level classification results, and establish a first-level storage space based on the first-level classification results;

[0025] Based on the first-level classification results, the data streams in each dimension in the data space are marked, and data features are extracted from the data streams in each dimension to obtain feature vectors of the data streams;

[0026] Mapping the obtained characteristic vectors of the data stream to M-dimensional orthogonal vectors associated with the corresponding dimensions of the data space; obtaining the mean vector of the characteristic vectors contained in each orthogonal vector;

[0027] Performing a zero-mean transformation on the mean vectors in each of the obtained orthogonal vectors to obtain a covariance matrix after the zero-mean transformation; obtaining all eigenvalues ​​and corresponding eigenvectors of the covariance matrix;

[0028] Set the eigenvalue threshold, delete the eigenvalues ​​and their corresponding eigenvectors that are smaller than the eigenvalue threshold, and form the remaining eigenvectors into a vector matrix;

[0029] Perform dimension reduction on the obtained vector matrix and obtain the value after dimension reduction;

[0030] The obtained reduced-dimensional value is bound and associated with the corresponding orthogonal vector, and a corresponding secondary storage space is established.

[0031] Furthermore, the process of storing the classified data includes:

[0032] Output the data stream processed by the digital model, and import the output data stream into the corresponding first-level storage space according to the first-level classification result;

[0033] According to the established secondary storage space, the data in each primary storage space will be aggregated and imported into the corresponding secondary storage space to complete the data storage process.

[0034] Compared with the prior art, the beneficial effects of the present invention are as follows: by constructing a digital model and setting a multi-dimensional data space in the digital model, data from multiple source points are classified at a primary level according to the multi-dimensional data space; then, by extracting features from the data, the extracted features are mapped to orthogonal vectors corresponding to each dimension in the multi-dimensional data space, and a covariance matrix corresponding to each orthogonal vector is obtained according to the features contained in each orthogonal vector, and then a reduced-dimensional value corresponding to the covariance matrix is ​​obtained according to a characteristic vector matrix corresponding to the obtained covariance matrix, thereby completing further classification of the data input from each data source point, making the classification of multi-source point data more refined, and the classification results between the data are more correlated, that is, the correlation between the data in the same orthogonal vector. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a schematic diagram of the present invention. DETAILED DESCRIPTION

[0036] like Figure 1 As shown, a classification method for digital model generation data includes the following steps:

[0037] Step 1: authenticate the data source point and aggregate the data uploaded by the data source point to obtain a sample data set;

[0038] It should be further explained that, in the specific implementation process, the process of authenticating the data source point includes:

[0039] Setting an authority verification unit to verify the identity of the data source point through the authority verification unit;

[0040] Specifically, each data source point is provided with a corresponding identity recognition sequence and a basic user who can upload data through the data source point. After the basic user completes the personal information registration in the data source point, the basic user can upload data in the data source point category;

[0041] In the specific implementation process, the process of basic users completing personal information registration at the data source point includes:

[0042] The data source point is provided with a registration unit and a login unit for basic users to register their identities and enter the data source point;

[0043] The basic user's personal basic information is input through the registration unit, and the data source point verifies the input personal basic information; the personal basic information includes name, gender, age and real-name authenticated mobile phone number;

[0044] After the review is passed, a login account and password will be generated based on the mobile phone number in the personal basic information entered, and the generated login account and password will be sent to the corresponding mobile terminal;

[0045] The basic user obtains the permission to upload data in the data source point by inputting the obtained login account and login password into the login unit;

[0046] A user information sub-database is established in the data source point, and the basic personal information of the registered basic users is imported into the user information sub-database; it should be further explained that, in the specific implementation process, the user information sub-database is bound and associated with the identity recognition sequence corresponding to the data source point, and the basic personal information of all basic users in the user information sub-database is assigned the identity recognition sequence;

[0047] The identity recognition sequence corresponding to the data source point is uploaded to the authority verification unit for filing, and the data source point that has passed the filing obtains the authority to upload the source data.

[0048] It should be further explained that, in the specific implementation process, the process of obtaining the sample data set includes:

[0049] Mark the data uploaded by each data source point and record the uploaded data as original data;

[0050] Binding the obtained raw data with the identity identification sequence of the corresponding data source point to form source data;

[0051] The source data is converted into corresponding data streams, and the data streams obtained from various data source points are aggregated to form a sample data set.

[0052] Step 2: Input the sample data set into the digital model, and complete the classification of the data in the sample data set through the digital model;

[0053] Specifically, the obtained sample data set is input into the digital model as input, and all data streams in the sample data set are read;

[0054] An N-dimensional data space is provided in the digital model, and a primary data label is provided in each dimension of the data space; it should be further explained that, in a specific implementation process, each data label corresponds to a data category, wherein the number of dimensions N of the data space corresponds to the number of data source points; it should be further explained that, in a specific implementation process, the primary data label corresponds to the identity identification sequence of the data source point one by one;

[0055] Based on each primary data label, an M-dimensional orthogonal vector associated with the corresponding primary data label is established, where M < N;

[0056] According to the data source points corresponding to the data streams in the sample data set, the data streams in the sample data set are mapped to the corresponding dimensions in the data space, and a primary classification result is generated;

[0057] Output the first-level classification results, and establish a first-level storage space based on the first-level classification results;

[0058] Based on the first-level classification results, the data streams of each dimension in the data space are marked, and data features are extracted from the data streams in each dimension to obtain feature vectors of the data streams. The obtained feature vectors are then deduplicated, that is, when there are duplicate feature vectors, only unique feature vectors are retained.

[0059] The obtained feature vector of the data stream is mapped to an M-dimensional orthogonal vector associated with the corresponding dimension of the data space; it should be further explained that in the specific implementation process, when the number of extracted feature vectors is more than M, the excess part is cyclically mapped;

[0060] For example:

[0061] Assume that the number of orthogonal vectors associated with each dimension of the data space is M=5, and the orthogonal vectors are marked as k1, k2, k3, k4, k5 respectively;

[0062] The feature vectors extracted from the data stream in a certain dimension in the data space are n1, n2, ..., n7;

[0063] The result of mapping the feature vector extracted from the data stream to the corresponding orthogonal vector is k1→n1, n6; k2→n2, n7, k3→n3, k4→n4, k5→n5;

[0064] It should be further explained that, in the specific implementation process, when the digital model extracts features from the data stream in the new sample data set, a new feature vector that has never appeared before is obtained, and the new feature vector is mapped in sequence;

[0065] Example

[0066] The orthogonal vectors are marked as k1, k2, k3, k4, k5, and the feature vectors extracted from the old data stream are n1, n2, ..., n7;

[0067] Then the result of mapping the eigenvector to the corresponding orthogonal vector is k1→n1, n6; k2→n2, n7, k3→n3, k4→n4, k5→n5;

[0068] The new eigenvectors obtained are n1, n,5, n8, n9;

[0069] Then the result of mapping the new eigenvector to the corresponding orthogonal vector is k1→n1, n6, k5→n5, n7, k3→n3, n8, k4→n4, n9;

[0070] Obtain the mean vector of the eigenvectors contained in each orthogonal vector, and denote the mean vector as p;

[0071] Each eigenvector in the orthogonal vector is labeled as i, where i = 1, 2, ..., a, where a is an integer, and each eigenvector is labeled as X i ;

[0072] but

[0073] Performing a zero-mean transformation on the mean vectors in each of the obtained orthogonal vectors to obtain a covariance matrix after the zero-mean transformation, and recording the covariance matrix after the zero-mean transformation as C;

[0074] in

[0075] Then all eigenvalues ​​and corresponding eigenvectors of the covariance matrix C are obtained;

[0076] Set an eigenvalue threshold, delete the eigenvalues ​​and their corresponding eigenvectors that are smaller than the eigenvalue threshold, and form the remaining eigenvectors into a vector matrix, which is recorded as V;

[0077] Reduce the dimension of the obtained vector matrix V and obtain the value after dimension reduction. The value after dimension reduction is recorded as y, then y = V T (X i -p);

[0078] The obtained reduced-dimensional value y is bound and associated with the corresponding orthogonal vector, and a corresponding secondary storage space is established.

[0079] Step 3: Import the classified data to a designated location for storage.

[0080] Output the data stream processed by the digital model, and import the output data stream into the corresponding first-level storage space according to the first-level classification result;

[0081] According to the established secondary storage space, the data in each primary storage space will be aggregated and imported into the corresponding secondary storage space to complete the data storage process.

[0082] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A classification method for digital model generated data, It is characterized in that The following steps are involved: Step 1: authenticate the data source point and aggregate the data uploaded by the data source point to obtain a sample data set; Step 2: Input the sample data set into the digital model, and complete the classification of the data in the sample data set through the digital model; Step 3: Import the classified data to a designated location for storage.

2. A classification method for digital model generation data according to claim 1, It is characterized in that The process of authenticating a data source point includes: Setting an authority verification unit to verify the identity of the data source point through the authority verification unit; Each data source point is set with a corresponding identity recognition sequence; after completing the personal information registration in the data source point, you can upload data in the data source point class, establish a user information sub-library in the data source point, and import the personal basic information of the registered basic users into the user information sub-library, and upload the identity recognition sequence corresponding to the data source point to the authority verification unit for filing. The data source point that has been filed will obtain the authority to upload the source data.

3. A classification method for digital model generation data according to claim 2, It is characterized in that The process of completing personal information registration at the data source point includes: The data source point is provided with a registration unit and a login unit for basic users to register their identities and enter the data source point; Inputting the basic user's personal basic information through the registration unit, and the data source point auditing the inputted personal basic information; After the review is passed, a login account and password will be generated based on the mobile phone number in the personal basic information entered, and the generated login account and password will be sent to the corresponding mobile terminal; The basic user obtains the authority to upload data into the data source point by inputting the obtained login account and login password into the login unit.

4. A classification method for digital model generation data according to claim 3, It is characterized in that The process of obtaining the sample data set includes: The data uploaded by each data source point is marked and recorded as the original data; the obtained original data is bound to the identity identification sequence of the corresponding data source point to form the source data; the source data is converted into the corresponding data stream, and the data streams obtained from each data source point are summarized to form a sample data set.

5. A classification method for digital model generation data according to claim 4, It is characterized in that The process of classifying the data in the sample dataset through the digital model includes: The obtained sample data set is input into the digital model as input, and all data streams in the sample data set are read; An N-dimensional data space is provided in the digital model, and a primary data label is provided corresponding to each dimension of the data space; Based on each primary data label, an M-dimensional orthogonal vector associated with the corresponding primary data label is established; According to the data source points corresponding to the data streams in the sample data set, the data streams in the sample data set are mapped to the corresponding dimensions in the data space, and a primary classification result is generated; Output the first-level classification results, and establish a first-level storage space based on the first-level classification results; Based on the first-level classification results, the data streams in each dimension in the data space are marked, and data features are extracted from the data streams in each dimension to obtain feature vectors of the data streams; Mapping the obtained characteristic vectors of the data stream to M-dimensional orthogonal vectors associated with the corresponding dimensions of the data space; obtaining the mean vector of the characteristic vectors contained in each orthogonal vector; Performing a zero-mean transformation on the mean vectors in each of the obtained orthogonal vectors to obtain a covariance matrix after the zero-mean transformation; obtaining all eigenvalues ​​and corresponding eigenvectors of the covariance matrix; Set the eigenvalue threshold, delete the eigenvalues ​​and their corresponding eigenvectors that are smaller than the eigenvalue threshold, and form the remaining eigenvectors into a vector matrix; Perform dimension reduction on the obtained vector matrix and obtain the value after dimension reduction; The obtained reduced-dimensional value is bound and associated with the corresponding orthogonal vector, and a corresponding secondary storage space is established.

6. A method for classifying digital model generated data according to claim 5, It is characterized in that The process of storing the classified data includes: Output the data stream processed by the digital model, and import the output data stream into the corresponding first-level storage space according to the first-level classification result; According to the established secondary storage space, the data in each primary storage space will be aggregated and imported into the corresponding secondary storage space to complete the data storage process.