Project full-process tracking and supervising system based on data annotation

Abstract

The invention relates to a project full-process tracking and supervising system based on data annotation, which belongs to the technical field of data object identification, and comprises a plurality of data verification modules, a data unified verification module and a data annotation and distribution management system, the data unified verification module is used for receiving data sent by the plurality of data verification modules at the same time, one end of the data unified verification module is connected with the plurality of data verification modules, and the data annotation and distribution management system is used for annotating and distributing the data sent by the plurality of data verification modules respectively. The data annotation and distribution management system is connected with the other end of the data unified verification module; the system has the beneficial effects that the data convergence identification processing module of the data annotation and distribution management system and the data distribution processing module of the data annotation and distribution management system realize reasonable annotation and efficient distribution of data in electronic project text data.

Description

Technical Field

[0001] The present invention belongs to the technical field of data object identification, and in particular relates to a project full-process tracking and supervision system based on data annotation. Background Art

[0002] Data identification is a crucial link in the field of artificial intelligence (AI) and machine learning (ML). Its main purpose is to mark and classify raw data so that it can be recognized and processed by computer systems. Specifically, data identification is the process of processing raw voice, pictures, text, video and other data through data annotation to transform them into machine-recognizable information. Data identification is a key link for the effective operation of artificial intelligence algorithms. It helps machine learning models better understand and process data by providing structured training data, thereby improving the performance and accuracy of the models.

[0003] The data labeling process is usually completed manually by human labelers, and the content and standards of the labeling need to be determined according to the specific tasks and application scenarios. Common data labeling types include image labeling, speech labeling, text labeling, and 3D point cloud labeling. For example, image labeling involves classifying image data, labeling frames, and key points; while text labeling includes part-of-speech tagging, sentiment analysis, and text cleaning. Summary of the invention

[0004] The present invention provides a project full-process tracking and supervision system based on data labeling, which is used to solve the technical problem that data in electronic project text data in the prior art cannot be reasonably labeled and cannot be efficiently allocated. The present invention realizes reasonable labeling and efficient allocation of data in electronic project text data by adopting a category dispersion method in a data convergence identification processing module and a position transformation selection allocation method in a data allocation processing module.

[0005] In order to achieve the above object, the present invention is implemented by the following technical solutions:

[0006] The full-process project tracking and supervision system based on data annotation includes:

[0007] A plurality of data verification modules, wherein the plurality of data verification modules are used to respectively verify the data corresponding to the respective modules;

[0008] A unified data verification module, the unified data verification module is used to receive data sent simultaneously by multiple data verification modules, one end of the unified data verification module is connected to multiple data verification modules;

[0009] A data labeling and distribution management system, which is used to label and distribute the data sent by each of the multiple data verification modules, and the data labeling and distribution management system is connected to the other end of the unified data verification module;

[0010] After each of the data verification modules has simultaneously verified the corresponding data, each of the data verification modules will simultaneously send the verified data to the unified data verification module for simultaneous verification of the data. The data labeling and distribution management system will then label and distribute the verified data sent by the unified data verification module to the data labeling and distribution management system.

[0011] Optionally, each of the data verification modules is connected to its corresponding data platform respectively, and each of the data verification modules is started at the same time and verifies the data corresponding to each of the data verification modules at the same time.

[0012] Optionally, the unified data verification module is used for simultaneity verification of each verified data. After each data verification module simultaneously verifies the corresponding data, the unified data verification module is started, and the unified data verification module receives data sent to the unified data verification module simultaneously by each data verification module.

[0013] Optionally, the data labeling and distribution management system includes a data convergence identification processing module and a data distribution processing module;

[0014] The data convergence identification processing module is connected to the data unified verification module, and the data convergence identification processing module is used to mark the respectively verified data in a category-dispersed manner;

[0015] The data distribution processing module is connected to the data convergence identification processing module, and is used to distribute the data after verification and annotation by adopting a position transformation selection distribution method.

[0016] Optionally, the data convergence identification processing module includes a data concentration unit, a data dispersion unit and a data labeling unit; the data concentration unit receives and collects the respectively verified data sent by the data unified verification module, the data dispersion unit disperses the respectively verified data, and the data labeling unit labels the respectively verified data.

[0017] Optionally, the data aggregation identification processing module adopts a category dispersion method to mark, which is a working method using the following formula (1):

[0018] (1);

[0019] in, To collect and then distribute the verified data, is the serial number of the i-th verified data, To annotate the database; To call the annotated data in the annotated database, the called annotated data is matched to the i-th verified data; Represents data annotation operations;

[0020] To call the annotated data in the annotation database, the called annotated data is corresponded to the i-th verified data, and the i-th verified data is annotated.

[0021] Optionally, the data allocation processing module includes a data allocation unit and a data selection distributor;

[0022] The data distribution unit is connected to the data dispersion unit, and the data distribution unit randomly transforms the positions of the respectively verified and marked data;

[0023] The data selection distributor is connected to the data distribution unit, and the data selection distributor selects and distributes the data after random position transformation and respective verification and annotation in the data distribution unit through position transformation.

[0024] Optionally, the data allocation processing module adopts a position transformation selection allocation method, which adopts the working method of the following formula (2):

[0025] (2);

[0026] in, represents the operation of random position transformation, To verify that the position of the labeled data can be randomly changed, To select an operation, The data after the selected verification annotation is the jth serial number, and y is the operation corresponding to the jth serial number of the data after the selected verification annotation; To allocate the data with the jth verification mark, allocate the data with the jth verification mark;

[0027] A certain data among the verification and annotated data after random position transformation is selected, the selected verification and annotated data corresponds to the j-th serial number, and the verification and annotated data of the j-th serial number is allocated.

[0028] Optionally, a data monitoring module is connected to one end of the data labeling and distribution management system away from the unified data verification module, and the data monitoring module is used to monitor the distribution status of the data labeling and distribution management system.

[0029] Beneficial effects of the present invention:

[0030] 1. The data labeling and distribution management system of the present invention re-labels and distributes the respectively verified data sent to the data labeling and distribution management system by the data unified verification module. The data labeling and distribution management system includes a data convergence identification processing module and a data distribution processing module. The data convergence identification processing module adopts a category dispersion method to label the respectively verified data, and the data distribution processing module adopts a position transformation selection distribution method to distribute the respectively verified and labeled data. The reasonable labeling and efficient distribution of the data in the electronic project text data are achieved by the data convergence identification processing module adopting a category dispersion method and the data distribution processing module adopting a position transformation selection distribution method.

[0031] 2. The various data verification modules of the present invention are started simultaneously and verify the data corresponding to each data verification module simultaneously. After each data verification module verifies the corresponding data simultaneously, the data unified verification module is started, and the data unified verification module verifies the data simultaneously. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0033] Figure 1 It is a schematic diagram of the overall structure of the present invention;

[0034] Figure 2 It is a structural schematic diagram of a data aggregation identification processing module of the present invention;

[0035] Figure 3 It is a structural schematic diagram of the data allocation processing module of the present invention. DETAILED DESCRIPTION

[0036] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0037] Example 1

[0038] like Figure 1 As shown, this embodiment provides a project full-process tracking and supervision system based on data labeling, including: multiple data verification modules, a unified data verification module, a data labeling and distribution management system, and a data monitoring module.

[0039] The plurality of data verification modules are used to verify the data corresponding to them respectively;

[0040] The unified data verification module is used to receive data sent simultaneously by a plurality of the data verification modules, and one end of the unified data verification module is connected to the plurality of data verification modules;

[0041] The data labeling and distribution management system is used to label and distribute the data sent by multiple data verification modules respectively, and the data labeling and distribution management system is connected to the other end of the data unified verification module;

[0042] The end of the data labeling and distribution management system away from the data unified verification module is connected to a data monitoring module, which is used to monitor the distribution status of the data labeling and distribution management system, that is, the data monitoring module monitors or supervises whether the selected verified and labeled data is distributed.

[0043] After each data verification module has verified its corresponding data simultaneously, each data verification module will send the verified data to the unified data verification module for simultaneous verification of the data. The data labeling and distribution management system will then label and distribute the verified data sent by the unified data verification module to the data labeling and distribution management system.

[0044] Only after each data verification module verifies its corresponding data at the same time, and each data verification module sends the verified data to the unified data verification module for simultaneous verification, can the data labeling and distribution management system label and distribute the verified data sent by the unified data verification module to the data labeling and distribution management system.

[0045] Specifically, each data verification module is connected to its corresponding data platform respectively, and each data verification module is started at the same time and verifies the data corresponding to each data verification module at the same time.

[0046] The unified data verification module is used for simultaneous verification of each verified data. After each data verification module simultaneously verifies the corresponding data, the unified data verification module is started, and the unified data verification module receives the data sent to the unified data verification module simultaneously by each data verification module.

[0047] like Figure 1 As shown, from Class A data to Class H data (i.e., each data platform), the Class A data verification module verifies the Class A data, and similarly, the Class H data verification module verifies the Class H data. In this embodiment, each data verification module is started simultaneously and verifies the data corresponding to each data verification module simultaneously. After each data verification module verifies the data corresponding to each data verification module simultaneously, the data unified verification module is started, and the data unified verification module receives the data sent to the data unified verification module simultaneously by each data verification module.

[0048] Example 2

[0049] Based on Example 1, Figure 1 and Figure 2 As shown, the data labeling and distribution management system includes a data convergence identification processing module and a data distribution processing module;

[0050] The data convergence identification processing module is connected to the data unified verification module, and the data convergence identification processing module is used to mark the respectively verified data in a category dispersion manner;

[0051] The data convergence identification processing module includes a data concentration unit, a data dispersion unit and a data labeling unit; one end of the data concentration unit is connected to the data unified verification module (to collect the data sent by the data unified verification module), the data concentration unit receives and collects the respectively verified data sent by the data unified verification module, the other end of the data concentration unit is connected to the data dispersion unit, the data dispersion unit further disperses the respectively verified data (the respectively verified data are further dispersed for convenience of labeling), the data labeling is connected to the data dispersion unit, and the data labeling unit labels the respectively verified data.

[0052] The data dispersion unit of the data aggregation identification processing module combines with the data annotation unit to adopt the category dispersion method to annotate, which adopts the working method of the following formula (1):

[0053] (1);

[0054] in, The data collected and then dispersed after verification (i.e., Class A data to Class H data) is the serial number of the i-th verified data (i.e. the serial number of a certain type of data from type A to type H), To annotate the database; To call the annotated data in the annotated database, the called annotated data is matched to the i-th verified data; Represents data annotation operations;

[0055] To call the annotated data in the annotation database, the called annotated data is corresponded to the i-th verified data, and the i-th verified data is annotated.

[0056] Formula (1) is to disperse the data of category A to category H, the serial number of a certain type of data from the data of category A to category H is the i-th serial number, the annotation data of the corresponding i-th serial number in the annotation database is called, and the called i-th serial number annotation data is annotated to the verified data of the i-th serial number, thereby realizing the reasonable annotation of the data in the electronic project text data.

[0057] Example 3

[0058] Based on Example 2, Figure 1 and Figure 3 As shown, the data distribution processing module is connected to the data convergence identification processing module, and the data distribution processing module is used to distribute the data after verification and annotation by adopting the position transformation selection distribution method.

[0059] The data distribution processing module includes a data distribution unit and a data selection distributor;

[0060] The data distribution unit is connected to the data dispersion unit, and the data distribution unit randomly transforms the positions of the data after verification and annotation, that is, A R …B R …G R …H R The order of random transformation is B R …A R …H R …G R The order of A R …B R …G R …H R To verify the labeled data.

[0061] The data selection distributor is connected to the data distribution unit, and the data selection distributor selects and distributes the data after random position transformation and respective verification and annotation in the data distribution unit through position transformation.

[0062] The data allocation processing module adopts the position transformation selection allocation method, which uses the following formula (2) working method:

[0063] (2);

[0064] in, represents the operation of random position transformation, To verify that the position of the labeled data can be randomly changed, To select an operation, The data after the selected verification annotation is the jth serial number, and y is the operation corresponding to the jth serial number of the data after the selected verification annotation; To allocate the data after verification and annotation of the jth serial number, the data after verification and annotation of the jth serial number is allocated, and the allocated data is output through the data output module;

[0065] A certain data among the verification and annotated data after random position transformation is selected, the selected verification and annotated data corresponds to the j-th serial number, and the verification and annotated data of the j-th serial number is allocated.

[0066] Formula (2) is to select a data from each verification labeled data with random position transformation (i.e. select BR …A R …H R …G R ), set the serial number of the selected verification marked data to j, and then distribute the verification marked data with the jth serial number, thereby realizing the efficient distribution of the marked data in the electronic project text data.

[0067] Example 4

[0068] Based on Example 1, this embodiment provides the following details:

[0069] Each data verification module is connected to its corresponding data platform respectively, and each data platform sends its own data to its corresponding data verification module. Each data verification module is started at the same time and verifies the data corresponding to each data verification module at the same time. After each data verification module verifies the data corresponding to each data simultaneously, the data unified verification module is started, and the data unified verification module verifies the data simultaneously (that is, the data corresponding to each data verification module is sent to the data unified verification module at the same time, the data unified verification module is started, and the data unified verification module verifies the data simultaneously), and the data labeling and distribution management system then labels and distributes the verified data sent by the data unified verification module to the data labeling and distribution management system.

[0070] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope recorded in the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. The project full-process tracking and supervision system based on data annotation is characterized by: include: A plurality of data verification modules, wherein the plurality of data verification modules are used to respectively verify the data corresponding to the respective modules; A unified data verification module, the unified data verification module is used to receive data sent simultaneously by multiple data verification modules, one end of the unified data verification module is connected to multiple data verification modules; A data labeling and distribution management system, which is used to label and distribute the data sent by each of the multiple data verification modules, and the data labeling and distribution management system is connected to the other end of the unified data verification module; After each of the data verification modules has simultaneously verified the corresponding data, each of the data verification modules will simultaneously send the verified data to the unified data verification module for simultaneous verification of the data. The data labeling and distribution management system will then label and distribute the verified data sent by the unified data verification module to the data labeling and distribution management system.

2. The project full-process tracking and supervision system based on data annotation according to claim 1 is characterized in that: Each of the data verification modules is connected to its corresponding data platform respectively, and each of the data verification modules is started at the same time and verifies the data corresponding to each of the data verification modules at the same time.

3. The project full-process tracking and supervision system based on data annotation according to claim 1 is characterized in that: The unified data verification module is used for simultaneous verification of each verified data. After each data verification module simultaneously verifies the corresponding data, the unified data verification module is started, and the unified data verification module receives the data sent to the unified data verification module simultaneously by each data verification module.

4. The project full-process tracking and supervision system based on data annotation according to claim 1 is characterized in that: The data labeling and distribution management system includes a data convergence identification processing module and a data distribution processing module; The data convergence identification processing module is connected to the data unified verification module, and the data convergence identification processing module is used to mark the respectively verified data in a category-dispersed manner; The data distribution processing module is connected to the data convergence identification processing module, and is used to distribute the data after verification and marking by adopting a position transformation selection distribution method.

5. The project full-process tracking and supervision system based on data annotation according to claim 4 is characterized in that: The data convergence identification processing module includes a data concentration unit, a data dispersion unit and a data labeling unit; the data concentration unit receives and collects the respectively verified data sent by the data unified verification module, the data dispersion unit further disperses the respectively verified data, and the data labeling unit labels the respectively verified data.

6. The project full-process tracking and supervision system based on data annotation according to claim 5 is characterized in that: The data aggregation identification processing module adopts the category dispersion method to mark, and adopts the working method of the following formula (1): (1)； in, To collect and then distribute the verified data, is the serial number of the i-th verified data, To annotate the database; To call the annotated data in the annotated database, the called annotated data is matched to the i-th verified data; Represents data annotation operations; To call the annotated data in the annotation database, the called annotated data is corresponded to the i-th verified data, and the i-th verified data is annotated.

7. The project full-process tracking and supervision system based on data annotation according to claim 4 is characterized in that: The data distribution processing module includes a data distribution unit and a data selection distributor; The data distribution unit is connected to the data dispersion unit, and the data distribution unit randomly transforms the positions of the respectively verified and marked data; The data selection distributor is connected to the data distribution unit, and the data selection distributor selects and distributes the data after random position transformation and respective verification and annotation in the data distribution unit through position transformation.

8. The project full-process tracking and supervision system based on data annotation according to claim 7 is characterized in that: The data allocation processing module adopts the method of position transformation selection allocation, and adopts the working method of the following formula (2): (2)； in, represents the operation of random position transformation, To verify that the position of the labeled data can be randomly changed, To select an operation, The data after the selected verification annotation is the jth serial number, and y is the operation corresponding to the jth serial number of the data after the selected verification annotation; To allocate the data with the jth verification mark, allocate the data with the jth verification mark; A certain data among the verification and annotated data after random position transformation is selected, the selected verification and annotated data corresponds to the j-th serial number, and the verification and annotated data of the j-th serial number is allocated.

9. The project full-process tracking and supervision system based on data annotation according to claim 1 is characterized in that: The end of the data labeling and distribution management system away from the data unified verification module is connected to a data monitoring module, and the data monitoring module is used to monitor the distribution status of the data labeling and distribution management system.

Images