A dynamic task-based data classification method

Abstract

The present application relates to a kind of data classification methods based on dynamic task, belong to artificial intelligence technical field, solve the single and the problem that classification rule cannot be dynamically optimized in existing data classification method. Including: S1, according to task identification, obtain multiple tasks from rule base and construct as task tree;S2, adjust the task in task tree and transmit into classification big model, select root task as pending task and distribute to classification intelligent agent;Classification big model and classification intelligent agent are the big model application based on LangChain construction;S3, classification intelligent agent identifies whether pending task has been aligned data attribute, and calls data classification tool to execute pending task for the pending task of aligned data attribute;Classification big model selects new pending task from task tree according to execution result and distributes to classification intelligent agent, repeats step S3 until no new pending task, completes data classification. Flexible data classification and dynamic rule optimization are realized.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a data classification method based on dynamic tasks. Background Art

[0002] With the rapid development of information technology, the scale and complexity of data are growing exponentially, prompting the emergence and continuous evolution of data classification technology. With the advancement of technologies such as artificial intelligence and big data, data classification is also playing a key role in data mining, machine learning, and knowledge graph construction, providing a foundation for extracting valuable information from massive amounts of data and building intelligent applications.

[0003] Many existing data classification methods rely on manually predefined rules and standards. Once established, these rules are difficult to flexibly adjust to changing data or new application scenarios. For example, in some traditional industries, data classification standards are based on industry experience and historical data. When new business models emerge or data information changes, the original classification standards cannot effectively adapt. This results in classification results that cannot meet actual needs when processing dynamically changing data. Adding, modifying, and validating rules consumes a lot of time and manpower.

[0004] While some methods utilize deep learning networks or large language models for data classification, they fail to consider the differences in data characteristics and category distribution across different classification tasks. Using a single classification method fails to fully exploit and adapt to these differences, resulting in poor classification results. Furthermore, different classification tasks may have specific business needs and objectives, and using a single method may not be optimized for these specific requirements, making it unsuitable for real-world applications. Summary of the Invention

[0005] In view of the above analysis, an embodiment of the present invention aims to provide a data classification method based on dynamic tasks to solve the problem that existing data classification methods are single, resulting in poor classification results and inability to dynamically optimize classification rules.

[0006] An embodiment of the present invention provides a data classification method based on dynamic tasks, comprising the following steps:

[0007] S1. Acquire multiple tasks from the rule base according to the received task identifier, construct a task tree, and display it on the interface;

[0008] S2. After adjusting the tasks in the task tree on the interface, transfer them to the classification model and select the root task as the to-do task to be distributed to the classification agent. The classification model and classification agent are large-scale model applications built on LangChain.

[0009] S3. The classification agent identifies whether the attributes of the to-do tasks have been aligned, and calls the data classification tool to execute the to-do tasks for the to-do tasks that have been aligned, and feeds the execution results back to the classification model; the classification model selects new to-do tasks from the task tree based on the execution results and distributes them to the classification agent, repeating step S3 until there are no new to-do tasks and the data classification is completed.

[0010] Based on the further improvement of the above method, the method also includes: for to-do tasks with misaligned data attributes, the abnormal execution results are fed back to the classification model; the classification model saves the newly added and modified to-do tasks with normal execution results to the rule library.

[0011] Based on the further improvement of the above method, multiple tasks are obtained from the rule base according to the received task identifier, and the task corresponding to the task identifier is taken as the root task. According to the subtask identifier of the root task, all subtasks are traversed layer by layer from the rule base until the subtask identifier is empty.

[0012] Based on the further improvement of the above method, the to-do tasks include: task ID, parent task ID, child task ID, alignment ID, task name, data source, data description, task goal, task description, task requirements, task parameters, tool model and tool hyperparameters; task parameters include similarity threshold and sample sampling number.

[0013] Based on further improvements to the above method, the classification agent identifies whether the to-do tasks have aligned data attributes, including:

[0014] Identify the type of to-do task based on its data source, task ID, and alignment ID. If it is a newly added or modified structured data classification task, it is considered a task to be aligned; otherwise, the to-do task has its data attributes aligned.

[0015] Obtain metadata information for each field in the data table based on the data source of the task to be aligned; compare each attribute in the data description of the task to be aligned with the metadata information of each field, and filter out attributes that do not exist in the metadata information as attributes to be aligned;

[0016] A similarity threshold is obtained from the task parameters of the task to be aligned. Based on the similarity threshold, it is determined whether a field can be selected from each field. If so, the selected field is replaced with the attribute to be aligned, and the data attribute of the task to be aligned is aligned; otherwise, the data attribute of the task to be aligned is not aligned.

[0017] Further improvement based on the above method, whether a field can be selected from each field based on a similarity threshold, including: respectively acquiring semantic vectors of the to-be-aligned attribute and each field, if there is a field with a similarity to the semantic vector of the to-be-aligned attribute exceeding the similarity threshold, put it into the candidate set, select the field corresponding to the maximum similarity from the candidate set to replace the to-be-aligned attribute; otherwise, gradually reduce the similarity threshold in proportion but not less than the minimum similarity threshold, put the field greater than the reduced similarity threshold into the candidate set, select the field corresponding to the maximum similarity from the candidate set to replace the to-be-aligned attribute; when the maximum similarity is less than the minimum similarity threshold, the field cannot be selected.

[0018] Further improvement based on the above method, the to-do task of the aligned data attribute calls the data classification tool to execute the to-do task, including:

[0019] Using the data extraction tool bound to the second LLM model corresponding to the classification agent, the to-be-classified data is obtained from the data source of the to-do task;

[0020] When the tool model and the tool hyperparameter in the to-do task are not empty, the corresponding tool model is extracted from the data classification tool bound to the second LLM model, the hyperparameter setting is obtained from the tool hyperparameter and given to the tool model, the tool model is called by the second LLM model to execute the data classification task according to the task target in the to-do task, and the execution result is obtained; otherwise, the classification prompt word is constructed for the to-do task, and the second LLM model executes the data classification task according to the classification prompt word to obtain the execution result.

[0021] Further improvement based on the above method, the to-be-classified data is obtained from the data source of the to-do task, including: when the sample sampling number in the task parameter of the to-do task is not empty, the to-be-classified data is obtained according to the sample sampling number, and when the execution result is normal, all to-be-classified data is selected to be re-executed.

[0022] Further improvement based on the above method, when the execution result is abnormal and the number of fields in the candidate set is greater than 1, the field with the second largest similarity is selected from the candidate set to replace the to-be-aligned attribute, and the data classification tool is called again to execute the to-do task to obtain the final execution result.

[0023] Further improvement based on the above method, the classification large model selects a new to-do task from the task tree according to the execution result, including:

[0024] When the execution result is normal, the next task is taken out from the task tree as a new to-do task in a breadth-first manner;

[0025] When the execution result is timeout and the execution number of the current to-do task does not exceed the threshold, the current to-do task is taken as a new to-do task again;

[0026] When the execution result is abnormal, or the execution count of the current to-do task exceeds the threshold, the current to-do task and all subtasks with the current to-do task as the parent task will no longer be executed, and the next task will be taken from the task tree as the new to-do task in a breadth-first manner.

[0027] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0028] 1. Use the classification big model to distribute the to-do tasks submitted on the interface. During the iterative execution process, use the classification agent to call the appropriate data classification tool to complete the data classification, and feed back the execution results to the classification big model. The classification big model dynamically determines the distribution strategy based on the execution results. The two parties work together to complete it, which not only ensures the quality of data classification, but also saves time and labor costs and improves data processing efficiency.

[0029] 2. While classifying, it supports the addition of new tasks and the modification of existing tasks, realizes the automatic expansion and optimization of task rules in the rule library, and quickly adapts to data changes and new application scenarios.

[0030] In the present invention, the above-mentioned technical solutions can be combined with each other to achieve more preferred combinations. Other features and advantages of the present invention will be described in the following description, and some advantages will become apparent from the description or be learned through practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings are only used for the purpose of illustrating specific embodiments and are not to be considered as limiting the present invention. Like reference symbols denote like components throughout the accompanying drawings.

[0032] Figure 1 This is a flow chart of a data classification method based on dynamic tasks in an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The preferred embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, and are not used to limit the scope of the present invention.

[0034] A specific embodiment of the present invention discloses a data classification method based on dynamic tasks, which is the basis for enterprises to manage data assets and build intelligent applications. Figure 1 As shown, the following steps are included:

[0035] S1. Acquire multiple tasks from the rule base according to the received task identifier, construct a task tree, and display it on the interface;

[0036] S2. After adjusting the tasks in the task tree on the interface, transfer them to the classification model and select the root task as the to-do task to be distributed to the classification agent. The classification model and classification agent are large-scale model applications built on LangChain.

[0037] S3. The classification agent identifies whether the pending tasks have been aligned with data attributes, calls the data classification tool to execute the pending tasks for the pending tasks with aligned data attributes, and feeds back the execution results to the classification model; the classification model selects new pending tasks from the task tree based on the execution results and distributes them to the classification agent, repeating step S3 until there are no new pending tasks and the data classification is completed.

[0038] During implementation, the classification big model is used to distribute the to-do tasks submitted on the interface. During the iterative execution process, the classification intelligent agent is used to call the appropriate data classification tool to complete the data classification, and the execution results are fed back to the classification big model. The classification big model dynamically determines the distribution strategy based on the execution results. The two parties work together to complete the task, which not only ensures the quality of data classification, but also saves time and labor costs and improves data processing efficiency.

[0039] It should be noted that the rule base in step S1 includes multiple classification task rules, referred to as tasks; each task includes: task identifier, parent task identifier, child task identifier, alignment identifier, task name, data source, data description, task target, task description, task requirements, task parameters, tool model and tool hyperparameters; the task identifier is automatically generated when it is saved to the rule base, and the parent task identifier and child task identifier are automatically associated according to the hierarchical relationship of the tasks; the alignment identifier is used to indicate whether data attributes need to be aligned before performing classification; the data description is used to describe data information from multiple dimensions such as attribute name, attribute Chinese name, attribute type, length or precision; the task target is used to describe the basis and classification target of data classification; the task requirement is used to indicate the special processing requirements of the task, for example, the date format must be unified, or there are no special requirements; the task parameters are used to indicate the parameters required by the classification agent during the execution of the task, including but not limited to: similarity threshold and sample sampling number; the tool model is used to indicate the tool model used when processing the task, and the tool model comes from the tool library. The tool types in the tool library include: general tools and data classification tools; general tools include but are not limited to: data extraction tools, data vectorization tools and image processing tools; data classification tools include but are not limited to: text-CNN classification model, text-SVM classification model, text-decision tree classification model and image-CNN classification model; tool hyperparameters are the paths to the configuration files used to specify the hyperparameters required by the tool model.

[0040] For example, the task of “department employee classification” in the rule base is shown in Table 1.

[0041] Table 1 Tasks of “Department Employee Classification”

[0042]

[0043] It should be noted that before data classification, data information from different data sources should be collected first, including structured data and unstructured data. For example, structured data in the telecommunications industry includes master data, transaction data, and analytical data, while unstructured data includes text data and multimedia data.

[0044] Furthermore, the collected data is pre-processed, including data attribute extraction and data cleaning. Data attribute extraction involves obtaining the data attributes of each data table and the relationships between data tables from structured data. Data cleaning includes default value processing, outlier processing, and data format conversion.

[0045] Furthermore, the data attributes are vectorized and stored in a vector database. For example, the Word2Vec model is used to obtain the vector embedding representation of the data attributes and stored in the Chroma or Pinecone vector database to facilitate subsequent retrieval and analysis.

[0046] The task identifier received in step S1 can be transmitted through an internal interface or method, or can originate from the interface. For example, the tasks in the rule base can be displayed on the interface, or the corresponding task identifier can be obtained by searching for tasks in the rule base by entering a keyword representing the task name. Multiple tasks are retrieved from the rule base based on the received task identifiers. The task corresponding to the task identifier is used as the root task, and all subtasks are retrieved from the rule base layer by layer based on the subtask identifiers of the root task until the subtask identifiers are empty.

[0047] Furthermore, the constructed task tree is displayed on the interface, and detailed information is obtained from the rule base for each task node on the task tree according to its corresponding task identifier; adjusting the tasks in the task tree on the interface includes: adding new tasks and task details, modifying and deleting existing tasks / or detailed information of existing tasks.

[0048] Tasks without task identifiers are newly added tasks. If the data source of the newly added tasks and modified tasks is a structured database table, these tasks are structured data classification tasks. The alignment identifier of these tasks is set to 1, indicating that the data attributes need to be aligned before performing data classification to avoid inconsistencies with the data attributes in the data source.

[0049] It is understandable that if the database table is updated on a large scale, the alignment flags of all tasks in the task tree are set to 1.

[0050] In step S2, the adjusted task tree is transmitted into a classification large model as a to-be-processed task tree corresponding to the current operator, and a root task in the task tree is initially taken out as a to-be-done task and distributed to a classification agent.

[0051] It should be noted that the classification large model and the classification agent are large model applications based on LangChain and containing a series of processing processes, wherein the first LLM (Large Language Model) model is adopted in the classification large model, and the second LLM model is adopted in the classification agent. The first / second LLM model includes but is not limited to: GPT series model, LLaMA model and Bert model.

[0052] In step S3, the classification agent identifies whether the to-be-done task has been aligned with the data attribute, including:

[0053] According to the data source, task identifier and alignment identifier of the to-be-done task, the type of the to-be-done task is identified. If it is a structured data classification task of addition and modification, it is a to-be-aligned task; otherwise, the to-be-done task has been aligned with the data attribute.

[0054] According to the data source of the to-be-aligned task, the metadata information of each field in the data table is obtained; each attribute in the data description of the to-be-aligned task is compared with the metadata information of each field, and the attribute that does not exist in the metadata information is selected as the to-be-aligned attribute; for example, the attribute name is preliminarily screened.

[0055] The similarity threshold is obtained from the task parameters of the to-be-aligned task, and whether a field can be selected from each field based on the similarity threshold is identified. If yes, the selected field replaces the to-be-aligned attribute, and the to-be-done task has been aligned with the data attribute; otherwise, the to-be-done task has not been aligned with the data attribute.

[0056] Further, whether a field can be selected from each field based on the similarity threshold includes:

[0057] The semantic vectors of the to-be-aligned attribute and each field are obtained respectively. If there is a field with a similarity to the semantic vector of the to-be-aligned attribute exceeding the similarity threshold, the field is put into a candidate set, and the field corresponding to the maximum similarity is selected from the candidate set to replace the to-be-aligned attribute; otherwise, the similarity threshold is gradually reduced in proportion but not less than a minimum similarity threshold, and the field greater than the reduced similarity threshold is put into the candidate set, and the field corresponding to the maximum similarity is selected from the candidate set to replace the to-be-aligned attribute; when the maximum similarity is less than the minimum similarity threshold, the field cannot be selected.

[0058] For example, the semantic vector of the to-be-aligned attribute is obtained according to the attribute name and the attribute Chinese description, the semantic vector of the field is obtained according to the field name and the field description, and the cosine similarity between the two semantic vectors is calculated.

[0059] Furthermore, the classification agent feeds back abnormal execution results to the classification model for pending tasks with misaligned data attributes; and calls the data classification tool to execute pending tasks with aligned data attributes, including:

[0060] ① Use the data extraction tool bound to the second LLM model corresponding to the classification agent to obtain the data to be classified from the data source of the to-be-classified tasks.

[0061] It should be noted that the data extraction tool is a general tool in the tool library and can be bound to multiple LLM models. The data extraction tool includes but is not limited to: connecting to a relational database and obtaining library table data; accessing the file system and obtaining file data.

[0062] Obtaining the data to be classified from the data source of the pending task, including: when the sample sampling quantity in the task parameters of the pending task is not empty, obtaining the data to be classified according to the sample sampling quantity, and when the execution result is normal, selecting all the data to be classified and re-executing.

[0063] ② When the tool model and tool hyperparameters in the to-do task are not empty, the corresponding tool model is extracted from the data classification tool bound to the second LLM model, and the hyperparameters are obtained from the tool hyperparameters and set to the tool model. The second LLM model calls the tool model to perform the data classification task on the classified data according to the task objectives in the to-do task to obtain the execution result; otherwise, a classification prompt word is constructed for the to-do task, and the second LLM model performs the data classification task according to the classification prompt word to obtain the execution result.

[0064] It should be noted that the role of the second LLM model is defined as a data classification expert in the classification prompt words, and the data description in the to-be-classified tasks is used as the context of the data to be classified; the task objectives in the to-be-classified tasks are used as the goals of the second LLM model; the task requirements in the to-be-classified tasks are used as the classification constraints of the second LLM model, and the second LLM model is required to automatically select the appropriate tool model from the bound data classification tools for classification by analyzing the data to be classified and the task objectives, and organize the execution results according to the set format.

[0065] Preferably, considering that when aligning data attributes, in some cases the similarities of multiple fields are greater than the similarity threshold and the difference in similarity is relatively small, at this time, the number of candidate sets is greater than 1, and the attribute with the largest similarity is first selected to replace the attribute to be aligned. The data extraction tool samples data from the data source according to the sample sampling number in the task parameters for data classification. If the execution result is abnormal, the classification agent selects the field with the second largest similarity from the candidate set to replace the attribute to be aligned, repeats the above process, and calls the data classification tool again to execute the pending task to obtain the final execution result.

[0066] The classification agent feeds back the execution result to the classification large model; the classification large model selects a new to-do task from the task tree according to the execution result, including:

[0067] When the execution result is normal, the next task is taken out from the task tree as a new to-do task in a breadth-first manner;

[0068] When the execution result is timeout and the execution number of the current to-do task does not exceed the threshold, the current to-do task is taken as a new to-do task again;

[0069] When the execution result is abnormal or the execution number of the current to-do task exceeds the threshold, the current to-do task and all subtasks of the current to-do task are no longer executed, and the next task is taken out from the task tree as a new to-do task in a breadth-first manner.

[0070] Preferably, the execution result fed back by the classification agent to the classification large model further includes: detailed information of the to-do task and a classification result, for the normal execution result, the classification large model constructs a prompt word according to the detailed information of the to-do task, sets a series of rules and constraints in the prompt word for secondary reasoning, and verifies whether the secondary reasoning result matches the classification result, if not, the new to-do task is selected as abnormal according to the execution result.

[0071] The classification large model distributes the selected new to-do task to the classification agent, and repeats step S3 until there is no new to-do task, and the data classification is completed.

[0072] Finally, the classification large model saves the new and modified to-do tasks with normal execution result to the rule library, including: generating a task identifier for the new to-do task, automatically filling in the parent task identifier and subtask identifier according to the hierarchical relationship in the task tree; covering the information of the modified to-do task with the existing information in the rule library, such as the lowered similarity threshold.

[0073] It should be noted that the classification result is part of the data asset, like the original collected data, which is stored in a vectorized manner, facilitating the retrieval and management of the data asset, forming an important data list and a sensitive data asset map.

[0074] Preferably, in order to improve the accuracy of the data asset, the tasks in the rule library are periodically executed by the classification large model to update the data in each category and continuously optimize the rule library.

[0075] Compared with existing technologies, the dynamic task-based data classification method provided in this embodiment utilizes a classification model to distribute pending tasks submitted through the interface. During the iterative execution process, the classification agent calls an appropriate data classification tool to complete data classification, and the execution results are fed back to the classification model. The classification model dynamically determines the distribution strategy based on the execution results. The two parties collaborate to achieve this goal, while ensuring the quality of data classification, saving time and labor costs and improving data processing efficiency. While performing classification, it supports the addition of new tasks and the modification of existing tasks, realizing the automatic expansion and optimization of task rules in the rule library, and quickly adapting to data changes and new application scenarios.

[0076] Those skilled in the art will appreciate that all or part of the process steps of the above-described embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, such as a magnetic disk, an optical disk, a read-only memory, or a random access memory.

[0077] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with this technical field within the technical scope disclosed by the present invention should be covered by the scope of protection of the present invention.

Claims

1. A data classification method based on dynamic tasks, characterized in that: The following steps are involved: S1. Acquire multiple tasks from the rule base according to the received task identifier, construct a task tree, and display it on the interface; S2. After adjusting the tasks in the task tree on the interface, transfer them to the classification model and select the root task as the to-do task to be distributed to the classification agent. The classification model and the classification agent are based on the large model application built on LangChain. S3. The classification agent identifies whether the pending tasks have aligned data attributes, invokes a data classification tool to execute the pending tasks for which the data attributes have aligned, and feeds back the execution results to the classification model. The classification model selects new pending tasks from the task tree based on the execution results and distributes them to the classification agent. Step S3 is repeated until there are no new pending tasks, and the data classification is completed. The classification agent identifies whether the pending task has aligned data attributes, including: identifying the type of the pending task according to the data source, task identifier and alignment identifier of the pending task, and if it is a newly added or modified structured data classification task, it is regarded as the task to be aligned; otherwise, the pending task has aligned data attributes; obtaining metadata information of each field in the data table according to the data source of the task to be aligned; comparing each attribute in the data description of the task to be aligned with the metadata information of each field, screening out attributes that do not exist in the metadata information, and treating them as attributes to be aligned; obtaining a similarity threshold from the task parameters of the task to be aligned, and identifying whether a field can be selected from each field based on the similarity threshold; if so, replacing the attribute to be aligned with the selected field, and then the pending task has aligned data attributes; otherwise, the pending task has not aligned data attributes; The method of identifying whether a field can be selected from each field based on the similarity threshold includes: obtaining semantic vectors of the attribute to be aligned and each field respectively; if there is a field whose similarity with the semantic vector of the attribute to be aligned exceeds the similarity threshold, the field is placed in a candidate set, and the field corresponding to the maximum similarity is selected from the candidate set to replace the attribute to be aligned; otherwise, the similarity threshold is gradually reduced in proportion but not less than the minimum similarity threshold, and the fields greater than the reduced similarity threshold are placed in the candidate set, and the field corresponding to the maximum similarity is selected from the candidate set to replace the attribute to be aligned; when the maximum similarity is less than the minimum similarity threshold, the field cannot be selected; If the execution result is abnormal and the number of fields in the candidate set is greater than 1, the field with the second largest similarity is selected from the candidate set to replace the attribute to be aligned, and the data classification tool is called again to execute the pending task to obtain the final execution result.

2. The data classification method based on dynamic tasks according to claim 1, characterized in that: The method further includes: for to-do tasks with misaligned data attributes, feeding back abnormal execution results to the classification model; and the classification model saving newly added and modified to-do tasks with normal execution results to a rule library.

3. The data classification method based on dynamic tasks according to claim 1, characterized in that: The method of obtaining multiple tasks from the rule base according to the received task identifier is to take the task corresponding to the task identifier as the root task, and to traverse the rule base layer by layer according to the subtask identifier of the root task and obtain all subtasks until the subtask identifier is empty.

4. The data classification method based on dynamic tasks according to claim 1, characterized in that: The to-do tasks include: task identifier, parent task identifier, child task identifier, alignment identifier, task name, data source, data description, task goal, task description, task requirements, task parameters, tool model and tool hyperparameters; the task parameters include similarity threshold and sample sampling quantity.

5. The data classification method based on dynamic tasks according to claim 4, characterized in that: The calling of a data classification tool to execute the pending tasks for the aligned data attributes includes: Using the data extraction tool bound to the second LLM model corresponding to the classification agent, obtain the data to be classified from the data source of the to-be-classified tasks; When the tool model and tool hyperparameters in the to-do task are not empty, the corresponding tool model is extracted from the data classification tool bound to the second LLM model, and the hyperparameters are obtained from the tool hyperparameters and set to the tool model. The second LLM model calls the tool model to perform the data classification task on the to-be-classified data according to the task objectives in the to-do task to obtain the execution result; otherwise, a classification prompt word is constructed for the to-be-classified task, and the second LLM model performs the data classification task according to the classification prompt word to obtain the execution result.

6. The data classification method based on dynamic tasks according to claim 5, characterized in that: The obtaining of the data to be classified from the data source of the pending task includes: when the sample sampling quantity in the task parameters of the pending task is not empty, obtaining the data to be classified according to the sample sampling quantity, and when the execution result is normal, selecting all the data to be classified and re-executing.

7. The data classification method based on dynamic tasks according to claim 1 or 5, characterized in that: The classification model selects new to-do tasks from the task tree based on the execution results, including: When the execution result is normal, the next task is taken from the task tree as the new to-do task in a breadth-first manner; When the execution result is timeout and the number of executions of the current to-do task does not exceed the threshold, the current to-do task will be used as a new to-do task again; When the execution result is abnormal, or the execution count of the current to-do task exceeds the threshold, the current to-do task and all subtasks with the current to-do task as the parent task will no longer be executed, and the next task will be taken from the task tree as the new to-do task in a breadth-first manner.

Images