Label processing method, data classification method and related device
By constructing a tag knowledge graph and correcting abnormal tags, the problem of insufficient data classification accuracy caused by manual tag setting was solved, thereby improving the accuracy of data classification and user experience.
Patent Information
- Application Number
- CN202511052268.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-29
- Publication Date
- 2025-11-11
AI Technical Summary
In existing technologies, label setting is mostly done manually, which lacks verification, resulting in insufficient classification accuracy, data clutter, and a poor user experience.
By constructing a tag knowledge graph, using tag data and hierarchical relationship data, semantic understanding is performed to identify and correct abnormal tags, thereby improving the accuracy of tag definitions.
It improves the accuracy of data classification, ensures that the classification system accurately identifies and classifies the data to be classified, reduces data clutter, and enhances the user experience.
Smart Images
Figure CN120929609A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of financial technology or other related fields, and in particular to a label processing method, a data classification method and related apparatus. Background Technology
[0002] Data classification allows data with common attributes or characteristics to be grouped together for easier user access. In related technologies, data classification typically involves identifying the data to be classified and matching the identification results with built-in labels in the classification system to determine the data category, thus achieving data classification.
[0003] However, currently, most labels are set manually, and the accuracy of the labels lacks verification. Inaccurate labels can lead to messy classification data and a poor user experience. Summary of the Invention
[0004] This application provides a tag processing method, a data classification method, and a related apparatus to improve the accuracy of data classification.
[0005] In a first aspect, embodiments of this application provide a label processing method, including:
[0006] Obtain the label data and hierarchical relationship data between labels in the classification system; the label data includes label names and corresponding label descriptions;
[0007] Based on the tag data and the hierarchical relationship data, a tag knowledge graph is constructed;
[0008] Semantic understanding is performed on the tag data of each node in the tag knowledge graph to obtain the tag semantic data corresponding to the tag name and the descriptive semantic data corresponding to the tag description;
[0009] Based on the tag semantic data and the description semantic data, determine whether there are abnormal tags in the knowledge graph;
[0010] If an anomaly exists, the anomaly label is corrected to obtain the target label knowledge graph.
[0011] Using the methods described above, a knowledge graph constructed based on label data and hierarchical relationships can correct labels that are abnormal, thereby improving the accuracy of label definitions and thus improving the accuracy of classification when using a classification system to classify data.
[0012] In some embodiments, determining whether there are anomalous tags in the knowledge graph based on the tag semantic data and the description semantic data includes:
[0013] Based on the semantic data, determine whether there is ambiguity in the tag descriptions of each tag, and / or determine whether there are duplicate tag descriptions in the tag descriptions of each tag. If so, determine that there are abnormal tags; or,
[0014] For the same tag, determine whether the corresponding descriptive semantic data and tag semantic data are consistent. If they are inconsistent, then an abnormal tag is identified.
[0015] In some embodiments, correcting the abnormal label includes:
[0016] If it is determined that there is ambiguity in the tag description of each tag, then obtain the remaining tag descriptions associated with the tag based on the tag knowledge graph;
[0017] The tag descriptions are redefined based on the remaining tag descriptions.
[0018] In some embodiments, correcting the abnormal label includes:
[0019] If it is determined that at least two tags have duplicate tag descriptions, then the tags with duplicate tag descriptions are merged.
[0020] or,
[0021] If it is determined that the tag descriptions of at least two tags are duplicated, then the remaining tag descriptions associated with each tag description are obtained according to the tag knowledge graph.
[0022] Based on the remaining tag descriptions, different descriptive information is added to each of the tag descriptions to distinguish duplicate tag descriptions.
[0023] In some embodiments, correcting the abnormal label includes:
[0024] If the descriptive semantic data and the tag semantic data corresponding to the same tag are inconsistent, then the remaining tag data associated with the tag are obtained according to the tag knowledge graph.
[0025] Based on the remaining tag data, determine whether the descriptive semantic data is abnormal, or whether the tag semantic data is abnormal;
[0026] If the descriptive semantic data is abnormal, the tag description of the tag is regenerated based on the remaining tag data and the tag semantic data;
[0027] If the tag semantic data is abnormal, the tag name is regenerated based on the tag data and the descriptive semantic data.
[0028] In some embodiments, constructing a tag knowledge graph based on the tag data and the hierarchical relationship data includes:
[0029] Using the tag data as points and the hierarchical relationship data as edges, an initial knowledge graph is constructed;
[0030] Based on the initial knowledge graph, determine the interrelationship data between the tags;
[0031] The initial knowledge graph is processed based on the mutual relationship data and the hierarchical relationship data to obtain the tag knowledge graph;
[0032] In the tag knowledge graph, the mutual relationship data and the hierarchical relationship data are the same.
[0033] In some embodiments, the method further includes:
[0034] Obtain sample data;
[0035] The sample data is classified using the target label knowledge graph;
[0036] If there are target data in the sample data that cannot be classified, then the target label knowledge graph is improved based on the target data.
[0037] In some embodiments, improving the target tag knowledge graph based on the target data includes:
[0038] Output first information to the outside; the first information includes the target data;
[0039] The system receives second information from external input and improves the target tag knowledge graph based on the second information; the second information includes the tag name and / or tag description of the target data.
[0040] In some embodiments, improving the target tag knowledge graph based on the second information includes:
[0041] Identify the label name and / or label description, and extract the corresponding feature data;
[0042] Based on the feature data, the hierarchical relationship data corresponding to the feature data is determined by matching the target label knowledge graph.
[0043] Based on the hierarchical relationship data, the tag name and / or tag description, create a new tag node in the target tag knowledge graph.
[0044] Secondly, embodiments of this application provide a data classification method, including:
[0045] Obtain the data to be classified;
[0046] The data is classified using the target label knowledge graph described in any of the first aspects.
[0047] Thirdly, this application provides an electronic device, including: a processor, a transceiver, and a memory; the processor is communicatively connected to the transceiver and the memory respectively;
[0048] The memory is used to store computer programs; the transceiver is used to communicate and interact with external devices; the processor is used to execute the computer instructions stored in the memory to implement the method of either the first aspect or the second aspect.
[0049] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the method of either the first or second aspect.
[0050] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the method of either the first or second aspect.
[0051] The tag processing method provided in this application involves acquiring tag data and hierarchical relationship data between tags in a classification system. The tag data includes tag names and corresponding tag descriptions. A tag knowledge graph is constructed based on the tag data and the hierarchical relationship data. Semantic understanding is performed on the tag data of each node in the tag knowledge graph to obtain tag semantic data corresponding to the tag name and descriptive semantic data corresponding to the tag description. Based on the tag semantic data and the descriptive semantic data, it is determined whether there are abnormal tags in the knowledge graph. If so, the abnormal tags are corrected to obtain the target tag knowledge graph. This method, by constructing a knowledge graph based on tag data and hierarchical relationships, can correct abnormal tags, thereby improving the accuracy of tag definitions and ultimately enhancing the accuracy of classification when using a classification system to classify data. Attached Figure Description
[0052] Figure 1 A flowchart illustrating a label processing method provided in this application embodiment. Figure 1 ;
[0053] Figure 2 A flowchart illustrating a label processing method provided in this application embodiment. Figure 2 ;
[0054] Figure 3 A flowchart illustrating a data classification method provided in an embodiment of this application;
[0055] Figure 4 This is a schematic diagram of the structure of a tag processing device provided in an embodiment of this application;
[0056] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation
[0057] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0058] In the embodiments of this application, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect, without limiting their order. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that the terms "first" and "second" do not necessarily imply that they are different.
[0059] It should be noted that, in the embodiments of this application, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.
[0060] It should be noted that the tag processing method, data classification method and related apparatus provided in this application can be used in the field of fintech, or in any field other than fintech. This application does not limit the application field of the tag processing method, data classification method and related apparatus.
[0061] As mentioned earlier, most labeling systems currently rely on manual label setting. However, the accuracy of these manual labels lacks verification, and inaccurate labels can lead to messy categorized data, hindering users from accessing and querying the data.
[0062] In view of this, embodiments of this application provide a tag processing method, a data classification method, and a related apparatus. By constructing a knowledge graph from the tag data and hierarchical relationship data in the classification system, and verifying the data of each node in the knowledge graph, and reconstructing tags with anomalies, the accuracy of tags can be effectively improved, thereby improving the accuracy of data classification.
[0063] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.
[0064] Figure 1 This is a flowchart illustrating a label processing method provided in an embodiment of this application, as shown below. Figure 1 As shown, it includes:
[0065] S101. Obtain the label data and hierarchical relationship data between the labels in the classification system.
[0066] In some embodiments, the execution entity of this application may be a processing unit or module (e.g., a central processing unit) with processing capabilities in the classification system.
[0067] In some embodiments, the tag data includes a tag name and a corresponding tag description. The tag description is information used to explain or describe the tag; that is, additional information about the tag.
[0068] The hierarchical relationship data between tags represents the mutual inclusion relationships between the tags.
[0069] For example, if the tags (names) include Personal Finance, Corporate Banking, Personal Deposits, Personal Loans, Corporate Banking, Corporate Digital Operations Management, and Corporate Customer Marketing, then the hierarchical relationship data would be: Personal Finance includes Personal Deposits, Personal Loans, Payroll, etc.; Corporate Banking includes Corporate Digital Operations Management and Corporate Customer Marketing, etc.; thus, the first level is: Personal Finance, Corporate Banking; the second level is: Personal Deposits, Personal Loans, Corporate Digital Operations Management, and Corporate Customer Marketing.
[0070] The processing unit can retrieve the stored label data and the hierarchical relationship data between the labels from the storage unit in the classification system.
[0071] S102. Construct a tag knowledge graph based on the tag data and the hierarchical relationship data.
[0072] In some embodiments, when the tag data and the hierarchical relationship data are obtained, a tag knowledge graph including each tag can be constructed using the tag data as points and the corresponding hierarchical relationship data as edges.
[0073] For example, each tag can be treated as a node (node attributes include tag name and corresponding tag description). Edges can be created based on the hierarchical relationships between tags (e.g., if tag A is the parent tag of tag B, an edge is created from A to B; the relationship type can be "belongs to" or "parent-child," etc.), resulting in a corresponding tag knowledge graph. Optionally, a graph database can be used to store the created knowledge graph, and graph visualization tools can be used to display it.
[0074] In some embodiments, to further improve the accuracy of the constructed knowledge graph, the processing unit may also construct the knowledge graph based on the following method:
[0075] For example, an initial knowledge graph is constructed using the tag data as points and the hierarchical relationship data as edges; based on the initial knowledge graph, the mutual relationship data between each tag is determined; the initial knowledge graph is processed based on the mutual relationship data and the hierarchical relationship data to obtain the tag knowledge graph; wherein, in the tag knowledge graph, the mutual relationship data and the hierarchical relationship data are the same.
[0076] The data on the relationships between labels can include both direct and indirect relationships. For example, a direct relationship can be a direct edge connecting two nodes; an indirect relationship can be the strength of association between nodes connected by a path with a path length greater than 1, which can be measured by calculating the shortest path between nodes, the probability of a random walk, or Personalized PageRank.
[0077] For example, the hierarchical relationship data is represented as a graph G_h (a directed graph where edges point from parent nodes to child nodes), and the mutual relationship data is represented as a graph G_r. The intersection of graphs G_h and G_r is obtained to get the signature knowledge graph. Here, the intersection is the intersection of edges. That is, an edge e exists in both G_r and G_h.
[0078] S103. Perform semantic understanding on the tag data of each node in the tag knowledge graph to obtain the tag semantic data corresponding to the tag name and the description semantic data corresponding to the tag description.
[0079] In some embodiments, a large semantic recognition model can be invoked to perform semantic understanding on the tag names of each tag to obtain tag semantic data, and semantic understanding on the tag description data of each tag to obtain description semantic data.
[0080] S104. Based on the tag semantic data and the description semantic data, determine whether there are any abnormal tags in the knowledge graph. If they exist, proceed to step S105; otherwise, end the process.
[0081] In some embodiments, semantic ambiguity judgment can be performed on the label semantic data and the description semantic data to determine whether there are abnormal labels.
[0082] In some embodiments, semantic duplication judgment can be performed on the semantic data of each tag to determine whether there are any abnormal tags, and / or semantic duplication judgment can be performed on the semantic data of the description of each tag to determine whether there are any abnormal tags.
[0083] In some embodiments, consistency judgment can be performed on the tag semantic data and the description semantic data of the same tag to determine whether there are abnormal tags in both.
[0084] S105. If it exists, correct the abnormal label to obtain the target label knowledge graph.
[0085] In some embodiments, if semantic ambiguity is determined in the label semantic data and / or the description semantic data, then an abnormal label is determined to exist.
[0086] In some embodiments, if it is determined that there are tags with duplicate semantic data, and / or tags describing duplicate semantic data, then it is determined that there are anomalous tags.
[0087] In some embodiments, if it is determined that there are tags whose semantic data and descriptive semantic data are inconsistent, then it is determined that there are abnormal tags.
[0088] In some embodiments, if abnormal tags exist, they can be corrected. For example, ambiguous tag semantic data and / or descriptive semantic data can be corrected, tags with duplicate tag semantic data can be merged, and inconsistencies between descriptive semantic data and tag semantic data can be corrected by adjusting the descriptive semantic data or tag semantic data.
[0089] The tag processing method provided in this application involves acquiring tag data and hierarchical relationship data between tags in a classification system. The tag data includes tag names and corresponding tag descriptions. A tag knowledge graph is constructed based on the tag data and the hierarchical relationship data. Semantic understanding is performed on the tag data of each node in the tag knowledge graph to obtain tag semantic data corresponding to the tag name and descriptive semantic data corresponding to the tag description. Based on the tag semantic data and the descriptive semantic data, it is determined whether there are abnormal tags in the knowledge graph. If so, the abnormal tags are corrected to obtain the target tag knowledge graph. This method, by constructing a knowledge graph based on tag data and hierarchical relationships, can correct abnormal tags, thereby improving the accuracy of tag definitions and ultimately enhancing the accuracy of classification when using a classification system to classify data.
[0090] Based on the above embodiments, the process of determining the existence of abnormal tags and correcting abnormalities will be further explained below.
[0091] In some embodiments, based on the semantic description data, it is determined whether there is ambiguity in the label description of each label; if so, it is determined that there is an abnormal label.
[0092] For example, a large model can be used to perform semantic analysis on the semantic descriptions of each tag to determine whether there is ambiguity in the semantic data. If there is ambiguity, the tag description is determined to be an ambiguous tag description, and the tag is then marked as an abnormal tag.
[0093] In some embodiments, it is determined whether there are duplicate tag descriptions in the tag descriptions of each tag based on the semantic data. If so, it is determined that there are abnormal tags.
[0094] For example, the semantic description data of each tag can be compared to determine whether there is any identical semantic description data. If so, it can be determined that there are tags with the same tag description, and then the tags with the same tag description can be marked as abnormal tags.
[0095] In some embodiments, for the same tag, it is determined whether the corresponding descriptive semantic data and tag semantic data are consistent. If they are inconsistent, it is determined that there is an abnormal tag.
[0096] For example, the similarity between the descriptive semantic data and the tag semantic data of the same tag can be judged. If the similarity is greater than or equal to a preset value, it is determined that the descriptive semantic data and the tag semantic data of the tag are consistent. If the similarity is less than the preset value, it is determined that the descriptive semantic data and the tag semantic data of the tag are inconsistent, and the tag is marked as an abnormal tag.
[0097] In some embodiments, if it is determined that the tag description of a certain tag is ambiguous, then other tag descriptions associated with the tag are obtained according to the tag knowledge graph; and the tag description is redefined according to the other tag descriptions.
[0098] For example, if it is determined that the label description of label A is ambiguous, the label descriptions of the parent and child labels of label A can be obtained. The label descriptions of label A, its parent and child labels, and the relationship between label A and its parent and child labels are then input into a large model. The large model is used to redefine the label description of label A to eliminate the ambiguity in the label description of label A.
[0099] In some embodiments, if it is determined that at least two tags have duplicate tag descriptions, the tags with duplicate tag descriptions are merged.
[0100] For example, if the label descriptions of labels A and B are duplicated, either label A or label B can be deleted to merge them. Furthermore, the label descriptions and names of the undelete labels, as well as the label names of the deleted labels, can be input into the larger model to redefine the label names of the undelete labels using the larger model.
[0101] In some embodiments, if it is determined that the tag descriptions of at least two tags are duplicated, then the remaining tag descriptions associated with each tag description are obtained according to the tag knowledge graph; based on the remaining tag descriptions, different descriptive information is added to each tag description to distinguish the duplicate tag descriptions.
[0102] For example, if it is determined that the label descriptions of label A and label B are duplicated, the label descriptions of the parent and child labels of labels A and B can be obtained separately. These label descriptions, along with their parent and child label descriptions and the relationship between label A and its parent and child labels, are then input into a larger model. This larger model adds different descriptive information to the label descriptions of labels A and B to distinguish the duplicated descriptions. For instance, if both label A and label B are described as "loans," after modification by the larger model, the label description of label A is changed to "personal-loans," and the label description of label B is changed to "corporate-loans."
[0103] In some embodiments, if the descriptive semantic data and the tag semantic data corresponding to the same tag are inconsistent, then other tag data associated with the tag are obtained according to the tag knowledge graph; based on the other tag data, it is determined whether the descriptive semantic data or the tag semantic data is abnormal; if the descriptive semantic data is abnormal, then the tag description of the tag is regenerated according to the tag data and the tag semantic data; if the tag semantic data is abnormal, then the tag name of the tag is regenerated according to the tag data and the descriptive semantic data.
[0104] For example, if it is determined that the descriptive semantic data and tag semantic data of tag A are inconsistent, the tag data of the parent and child tags of tag A can be obtained. The descriptive semantic data of tag A, as well as the descriptive semantic data of the parent and child tags and their hierarchical relationship with tag A, are input into the large model. The large model is used to determine whether the descriptive semantic data of tag A is abnormal. If it is abnormal, the tag description of tag A is redefined using the large model. If it is normal, the tag semantic data of tag A, as well as the tag semantic data of the parent and child tags and their hierarchical relationship with tag A, are input into the large model. The large model is used to determine whether the tag descriptive semantic data of tag A is abnormal. If it is abnormal, the tag name of tag A is regenerated using the large model.
[0105] The process of verifying the accuracy of labels in a classification system has been explained above. The following section will combine... Figure 2 The process of further refining the labels in the classification system is explained.
[0106] Figure 2 A flowchart illustrating a label processing method provided in this application embodiment. Figure 2 ,like Figure 2 As shown, it includes:
[0107] S201. Obtain sample data.
[0108] In some embodiments, externally input sample data may be received via a data interface. The sample data may include multiple data points to be classified.
[0109] S203. Classify the sample data using the target label knowledge graph.
[0110] In some embodiments, semantic understanding is performed on each data to be classified in the sample data to determine the data features of each data to be classified, and the data features are matched with each node of the target label knowledge graph to determine whether there is a node in the target label knowledge graph that matches each data to be classified. If there is a node, the label corresponding to the node is marked as the label of the data; if there is no node, the step shown in S203 is executed.
[0111] S203. If there are target data in the sample data that cannot be classified, then the target label knowledge graph is improved based on the target data.
[0112] In some embodiments, if the target data exists, it is determined that the target tag knowledge graph lacks a corresponding tag, and then first information is output to the outside; the first information includes the target data; second information is received from the outside, and the target tag knowledge graph is improved according to the second information; the second information includes the tag name and / or tag description of the target data.
[0113] For example, the external source could be the client of the relevant user, which outputs initial information to instruct the relevant user to manually set labels for the data.
[0114] Upon receiving the second information input by the relevant user, the tag name and / or tag description included in the second information can be identified, and the corresponding feature data can be extracted; the feature data can be matched in the target tag knowledge graph to determine the hierarchical relationship data corresponding to the feature data; and a new tag node can be created in the target tag knowledge graph based on the hierarchical relationship data, the tag name and / or tag description.
[0115] For example, a large model can be used to analyze the tag names and / or tag descriptions provided by the user, extract corresponding feature data, and match this feature data with each node in the target tag knowledge graph to obtain the parent and / or child tags that are related to the feature data, thereby obtaining the hierarchical relationship corresponding to the target data. After determining the hierarchical relationship, tag nodes can be created in the target tag knowledge graph using the hierarchical relationship data and the tag names and / or tag descriptions.
[0116] The tag processing method provided in this application involves acquiring sample data; classifying the sample data using a target tag knowledge graph; and, if the sample data contains target data that cannot be classified, improving the target tag knowledge graph based on the target data. This method can utilize externally input information to create new tags that are missing in the classification system, thereby improving the completeness of the tags and enhancing the accuracy of subsequent data classification, thus avoiding data classification chaos caused by missing tags.
[0117] Based on the above embodiments, this application also provides a data classification method.
[0118] Figure 3 This is a flowchart illustrating a data classification method provided in an embodiment of this application, as shown below. Figure 3 As shown, it includes:
[0119] S301. Obtain the data to be classified.
[0120] In some embodiments, externally input data to be classified may be received via a data interface.
[0121] S302. Classify the data using the target label knowledge graph.
[0122] In some embodiments, the target label knowledge graph can be a knowledge graph obtained based on any of the above embodiments.
[0123] After obtaining the data to be classified, features can be extracted from the data to be classified, and the extracted features can be matched with the features of the data and each node of the target label knowledge graph. The labels of the matched nodes are then marked as the labels of the data, thereby completing the classification of the data.
[0124] In summary, the data classification method for label processing provided in this application analyzes and judges the built-in label category data and the hierarchical relationship data between corresponding label categories in the classification system. This ensures that the data to be classified conforms to the classification standards of the classification system, avoiding the impact of erroneously entered messy data on the classification system. Simultaneously, by performing semantic judgment on the descriptive data of the label category data, the system can accurately identify the labels and ensure the system's understanding of each label is singular. Therefore, by judging the accuracy of the labels, the accuracy of the classification system in classifying the data to be classified is improved. When missing labels are detected in the classification system, relevant labels can be created by obtaining feedback from relevant personnel, improving the accuracy of subsequent data classification results.
[0125] Based on the above embodiments, this application also provides a label processing device.
[0126] Figure 4 This is a schematic diagram of the structure of a label processing device 40 provided in an embodiment of this application, as shown below. Figure 4 As shown, it includes:
[0127] The acquisition module 401 is used to acquire label data and hierarchical relationship data between labels in the classification system; the label data includes label names and corresponding label descriptions.
[0128] The construction module 402 is used to construct a tag knowledge graph based on the tag data and the hierarchical relationship data.
[0129] The identification module 403 is used to perform semantic understanding on the tag data of each node in the tag knowledge graph to obtain the tag semantic data corresponding to the tag name and the descriptive semantic data corresponding to the tag description.
[0130] The processing module 404 is used to determine whether there are abnormal tags in the knowledge graph based on the tag semantic data and the description semantic data; if there are, the abnormal tags are corrected to obtain the target tag knowledge graph.
[0131] In some embodiments, the processing module 404 is configured to determine whether there is ambiguity in the tag description of each tag based on the description semantic data, and / or determine whether there is duplicate tag description in the tag description of each tag; if so, determine that there is an abnormal tag; or, for the same tag, determine whether the corresponding description semantic data and tag semantic data are consistent; if they are inconsistent, determine that there is an abnormal tag.
[0132] In some embodiments, the processing module 404 is configured to, if it is determined that there is ambiguity in the tag description of each tag, obtain the remaining tag descriptions associated with the tag according to the tag knowledge graph; and redefine the tag descriptions according to the remaining tag descriptions.
[0133] In some embodiments, the processing module 404 is configured to merge the tags with duplicate tag descriptions if it is determined that at least two tags have duplicate tag descriptions; or, if it is determined that at least two tags have duplicate tag descriptions, obtain the remaining tag descriptions associated with each tag description according to the tag knowledge graph; and add different descriptive information to each tag description according to the remaining tag descriptions to distinguish the duplicate tag descriptions.
[0134] In some embodiments, the processing module 404 is configured to: if the descriptive semantic data and the tag semantic data corresponding to the same tag are inconsistent, obtain the remaining tag data associated with the tag based on the tag knowledge graph; determine whether the descriptive semantic data or the tag semantic data is abnormal based on the remaining tag data; if the descriptive semantic data is abnormal, regenerate the tag description of the tag based on the remaining tag data and the tag semantic data; if the tag semantic data is abnormal, regenerate the tag name of the tag based on the tag data and the descriptive semantic data.
[0135] In some embodiments, the construction module 402 is configured to construct an initial knowledge graph using the tag data as points and the hierarchical relationship data as edges; determine the mutual relationship data between each tag based on the initial knowledge graph; process the initial knowledge graph based on the mutual relationship data and the hierarchical relationship data to obtain the tag knowledge graph; wherein, in the tag knowledge graph, the mutual relationship data and the hierarchical relationship data are the same.
[0136] In some embodiments, the processing module 404 is used to acquire sample data; classify the sample data using the target label knowledge graph; and if there is target data in the sample data that cannot be classified, improve the target label knowledge graph based on the target data.
[0137] In some embodiments, the processing module 404 is configured to output first information to the outside; the first information includes the target data; receive second information input from the outside, and improve the target tag knowledge graph according to the second information; the second information includes the tag name and / or tag description of the target data.
[0138] In some embodiments, the processing module 404 is configured to identify the tag name and / or tag description, extract corresponding feature data; match the feature data in the target tag knowledge graph to determine the hierarchical relationship data corresponding to the feature data; and create a new tag node in the target tag knowledge graph based on the hierarchical relationship data, the tag name and / or tag description.
[0139] The label processing apparatus provided in this application embodiment can execute the label processing method shown in any of the above embodiments, and its principle and technical effect are similar, so it will not be described again here.
[0140] This application also provides an electronic device.
[0141] Figure 5 This is a schematic diagram of the structure of the electronic device 50 provided in the embodiments of this application, such as... Figure 5 As shown, the electronic device may include: a transceiver 501, a processor 502, and a memory 503. The electronic device may be a processing unit as described in any of the above embodiments.
[0142] Processor 502 executes computer execution instructions stored in memory, causing processor 502 to perform the scheme in the above embodiments. Processor 502 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.
[0143] The memory 503 is connected to the processor 502 via the system bus and completes communication between them. The memory 503 is used to store computer program instructions.
[0144] Transceiver 501 can perform the functions of receiving and sending data and instructions.
[0145] Optionally, the electronic device 50 may also include a communication interface 504, which allows communication and interaction with external or internal devices via the communication interface 503. External devices may be, for example, client devices (e.g., mobile phones, tablets). In specific implementations, if the communication interface 504, memory 503, and processor 502 are implemented independently, they can be interconnected via a bus to complete communication with each other.
[0146] The system bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The system bus can be divided into address bus, data bus, control bus, etc. For ease of representation, only one thick line is used in the diagram, but this does not indicate that there is only one bus or one type of bus. Transceivers are used to enable communication between database access devices and other computers (e.g., clients, read-write libraries, and read-only libraries). Memory may include random access memory (RAM) and may also include non-volatile memory.
[0147] Optionally, in a specific implementation, if the communication interface 504, memory 503, and processor 502 are integrated on a single chip, then the communication interface 504, memory 503, and processor 502 can communicate through an internal interface.
[0148] This application also provides a chip for executing instructions, which is used to execute the technical solutions in the above embodiments.
[0149] This application also provides a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the technical solutions of the above embodiments. The implementation principle and technical effects are similar, and will not be repeated here.
[0150] In one possible implementation, a computer-readable medium may include random access memory (RAM), read-only memory (ROM), compact discread-only memory (CD-ROM) or other optical disc storage, disk storage or other magnetic storage devices, or any other medium targeted to carry or to store the required program code in the form of instructions or data structures, and accessible by a computer. Furthermore, any connection is appropriately referred to as a computer-readable medium. For example, if software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL), or wireless technologies such as infrared, radio, and microwave, then coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of medium. As used herein, disks and optical discs include optical discs, laser discs, optical discs, Digital Versatile Discs (DVDs), floppy disks, and Blu-ray discs, where disks typically reproduce data magnetically, while optical discs optically reproduce data using lasers. Combinations of the above should also be included within the scope of computer-readable media.
[0151] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the technical solutions of the above embodiments. Its implementation principle and technical effects are similar, and will not be repeated here.
[0152] In the specific implementation of the aforementioned terminal device or server, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of this application can be directly manifested as execution by a hardware processor, or execution by a combination of hardware and software modules within the processor.
[0153] Those skilled in the art will understand that all or part of the steps in any of the above method embodiments can be implemented by hardware associated with program instructions. The aforementioned program can be stored in a computer-readable storage medium, and when the program is executed, all or part of the steps in the above method embodiments are performed.
[0154] If the technical solution of this application is implemented in software form and sold or used as a product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the technical solution of this application can be embodied in the form of a software product, which is stored in a storage medium and includes a computer program or several instructions. This computer software product causes a computer device (which may be a personal computer, server, network device, or similar electronic device) to execute all or part of the steps of the method described in the embodiments of this application.
[0155] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this application.
[0156] It should be further noted that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.
[0157] It should be understood that the above-described device embodiments are merely illustrative, and the device of this application can also be implemented in other ways. For example, the division of units / modules in the above embodiments is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units, modules, or components may be combined, or integrated into another system, or some features may be ignored or not executed.
[0158] Furthermore, unless otherwise specified, the functional units / modules in the various embodiments of this application can be integrated into one unit / module, or each unit / module can exist physically separately, or two or more units / modules can be integrated together. The integrated units / modules described above can be implemented in hardware or in the form of software program modules.
[0159] When integrated units / modules are implemented in hardware, the hardware can be digital circuits, analog circuits, etc. The physical implementation of the hardware structure includes, but is not limited to, transistors, memristors, etc. Unless otherwise specified, the processor can be any suitable hardware processor, such as a CPU, GPU, FPGA, DSP, and ASIC, etc. Unless otherwise specified, the storage unit can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc.
[0160] If the integrated unit / module is implemented as a software program module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.
[0161] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.
[0162] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.
Claims
1. A label processing method, characterized in that, include: Obtain the label data and hierarchical relationship data between the labels in the classification system; The tag data includes the tag name and the corresponding tag description; Based on the tag data and the hierarchical relationship data, a tag knowledge graph is constructed; Semantic understanding is performed on the tag data of each node in the tag knowledge graph to obtain the tag semantic data corresponding to the tag name and the descriptive semantic data corresponding to the tag description; Based on the tag semantic data and the description semantic data, determine whether there are abnormal tags in the knowledge graph; If an anomaly exists, the anomaly label is corrected to obtain the target label knowledge graph.
2. The method according to claim 1, characterized in that, The step of determining whether there are anomalous tags in the knowledge graph based on the tag semantic data and the description semantic data includes: Based on the semantic data, determine whether there is ambiguity in the tag descriptions of each tag, and / or determine whether there are duplicate tag descriptions in the tag descriptions of each tag. If so, determine that there are abnormal tags; or, For the same tag, determine whether the corresponding descriptive semantic data and tag semantic data are consistent. If they are inconsistent, then an abnormal tag is identified.
3. The method according to claim 2, characterized in that, The correction of the abnormal labels includes: If it is determined that there is ambiguity in the tag description of each tag, then obtain the remaining tag descriptions associated with the tag based on the tag knowledge graph; The tag descriptions are redefined based on the remaining tag descriptions.
4. The method according to claim 2, characterized in that, The correction of the abnormal labels includes: If it is determined that at least two tags have duplicate tag descriptions, then the tags with duplicate tag descriptions are merged. or, If it is determined that the tag descriptions of at least two tags are duplicated, then the remaining tag descriptions associated with each tag description are obtained according to the tag knowledge graph. Based on the remaining tag descriptions, different descriptive information is added to each of the tag descriptions to distinguish duplicate tag descriptions.
5. The method according to claim 2, characterized in that, The correction of the abnormal labels includes: If the descriptive semantic data and the tag semantic data corresponding to the same tag are inconsistent, then the remaining tag data associated with the tag are obtained according to the tag knowledge graph. Based on the remaining tag data, determine whether the descriptive semantic data is abnormal, or whether the tag semantic data is abnormal; If the descriptive semantic data is abnormal, the tag description of the tag is regenerated based on the remaining tag data and the tag semantic data; If the tag semantic data is abnormal, the tag name is regenerated based on the tag data and the descriptive semantic data.
6. The method according to any one of claims 1-5, characterized in that, The step of constructing a tag knowledge graph based on the tag data and the hierarchical relationship data includes: Using the tag data as points and the hierarchical relationship data as edges, an initial knowledge graph is constructed; Based on the initial knowledge graph, determine the interrelationship data between the tags; The initial knowledge graph is processed based on the mutual relationship data and the hierarchical relationship data to obtain the tag knowledge graph; In the tag knowledge graph, the mutual relationship data and the hierarchical relationship data are the same.
7. The method according to claim 6, characterized in that, The method further includes: Obtain sample data; The sample data is classified using the target label knowledge graph; If there are target data in the sample data that cannot be classified, then the target label knowledge graph is improved based on the target data.
8. The method according to claim 7, characterized in that, The step of improving the target tag knowledge graph based on the target data includes: Output first information to the outside; the first information includes the target data; The system receives second information from external input and improves the target tag knowledge graph based on the second information; the second information includes the tag name and / or tag description of the target data.
9. The method according to claim 8, characterized in that, The step of improving the target tag knowledge graph based on the second information includes: Identify the label name and / or label description, and extract the corresponding feature data; Based on the feature data, the hierarchical relationship data corresponding to the feature data is determined by matching the target label knowledge graph. Based on the hierarchical relationship data, the tag name and / or tag description, create a new tag node in the target tag knowledge graph.
10. A data classification method, characterized in that, include: Obtain the data to be classified; The data is classified using the target label knowledge graph described in any one of claims 1-9.
11. An electronic device, characterized in that, include: The processor, transceiver, and memory are provided; the processor is communicatively connected to both the transceiver and the memory. The memory is used to store computer programs; The transceiver is used for communication and interaction with external devices; The processor is configured to execute the computer program to implement the method of any one of claims 1-9 or 10.
12. A computer-readable storage medium, characterized in that, It contains a computer program that is executed by a processor to implement the method of any one of claims 1-9 or 10.
13. A computer program product, characterized in that, Includes a computer program that, when executed by a controller, implements the method of any one of claims 1-9 or 10.