Label system management method and device, medium and program product

Abstract

The embodiment of the invention provides a label system management method and device, a medium and a program product, and relates to the technical field of label management. The method comprises the following steps: acquiring an initial label system; determining to-be-classified labels needing to be classified, and determining a pre-classification result corresponding to the to-be-classified labels based on a pre-trained classification model; and obtaining performance index change information of the to-be-determined label system relative to the initial label system, and updating the initial label system based on the pre-classification result under the condition that the pre-classification result is determined to meet a preset classification condition based on the performance index change information to obtain an updated label system. According to the embodiment of the invention, the to-be-classified labels are automatically pre-classified by training the classification model, and the classification operation is executed according to the performance change of the label system before and after pre-classification, so that rapid and accurate classification of new labels is realized, and the efficiency and stability of label system management are effectively improved.

Description

Technical Field

[0001] The present application relates to the technical field of label management, and in particular to a label system management method, device, medium and program product. Background Art

[0002] In the field of telephone sales, building and maintaining an effective labeling system can identify and classify user characteristics, and formulate and implement different business processes and sales strategies for different user types, which helps to improve customer service and sales efficiency.

[0003] At present, big model technology makes it very easy to generate specific labels based on text conversations, but the construction, management and optimization of the label system still mostly rely on manual operations, resulting in inefficient management of the label system. In addition, the stability of label system management is limited because it is greatly affected by the professional level and subjective experience of personnel. Summary of the invention

[0004] The purpose of the embodiments of the present application is to provide a label system management method, device, medium and program product to improve the efficiency and stability of label system management.

[0005] In a first aspect, an embodiment of the present application provides a method for managing a label system, including:

[0006] Acquire an initial label system; wherein the initial label system includes at least one label sub-dimension, and each label sub-dimension is associated with at least one label;

[0007] Determine a label to be classified that needs to be classified, and determine a pre-classification result corresponding to the label to be classified based on a pre-trained classification model; wherein the pre-classification result is represented by classifying the label to be classified into a target sub-dimension, and the target sub-dimension is one of the at least one label sub-dimension;

[0008] The performance indicator change information of the pending label system relative to the initial label system is obtained. When it is determined based on the performance indicator change information that the pre-classification result meets the preset classification condition, the initial label system is updated based on the pre-classification result to obtain an updated label system; wherein the pending label system is a label system formed after associating the labels to be classified to the target sub-dimension.

[0009] In an embodiment of the present application, a classification model is trained to automatically pre-classify labels to be classified, and classification operations are performed based on changes in label system performance before and after pre-classification, thereby achieving fast and accurate classification of new labels and effectively improving the efficiency and stability of label system management.

[0010] In some possible embodiments, determining the to-be-classified tags that need to be classified includes:

[0011] Based on several candidate tags acquired in advance, tag matching is performed on the user sales scenario data in real time, and the number of times each candidate tag is matched is recorded respectively;

[0012] When the number of times that a target candidate tag among the plurality of candidate tags is matched and hit exceeds a preset number threshold, the target candidate tag is determined as a tag to be classified that needs to be classified.

[0013] In the embodiment of the present application, by determining whether classification is required based on the number of hits of the new tag, the accuracy of tag management is further improved.

[0014] In some possible embodiments, the training method of the classification model includes:

[0015] Based on the initial label system, sample labels and their associated sample label sub-dimensions are obtained as training data, and the preset open source large model is trained to obtain a trained classification model.

[0016] In an embodiment of the present application, training data is constructed according to the constructed label system to train the classification model, thereby further improving the accuracy of label classification.

[0017] In some possible embodiments, obtaining performance indicator change information of the pending labeling system relative to the initial labeling system includes:

[0018] Based on the preset historical user telephone sales scenario data, the first hit conversion influence direction of the target sub-dimension relative to the initial label system is obtained, and the second hit conversion influence direction of the target sub-dimension relative to the pending label system is obtained; wherein, if the hit conversion rate corresponding to the target sub-dimension is greater than the average hit conversion rate of all label sub-dimensions under the current label system, then the hit conversion influence direction of the target sub-dimension relative to the current label system is positive; if the hit conversion rate corresponding to the target sub-dimension is less than the average hit conversion rate of all label sub-dimensions under the current label system, then the hit conversion influence direction of the target sub-dimension relative to the current label system is negative;

[0019] The determining, based on the performance indicator change information, that the pre-classification result meets a preset classification condition includes:

[0020] If it is determined that the first hit-conversion influence direction is consistent with the second hit-conversion influence direction, it is determined that the pre-classification result meets a preset classification condition.

[0021] In an embodiment of the present application, a consistency comparison is performed on the hit conversion impact direction of the pre-classified target sub-dimension before and after classification to determine whether the pre-classification result meets the classification conditions, thereby further improving the stability of label classification.

[0022] In some possible embodiments, obtaining performance indicator change information of the pending labeling system relative to the initial labeling system further includes:

[0023] Based on the historical user telephone sales scenario data, obtaining the data fluctuation index corresponding to the to-be-classified label in the to-be-determined label system; wherein the data fluctuation index includes a data standard score;

[0024] The determining, based on the performance indicator change information, that the pre-classification result meets the preset classification condition is specifically:

[0025] If it is determined that the first hit conversion impact direction is consistent with the second hit conversion impact direction, and the data fluctuation index is within a preset tolerance threshold, it is determined that the pre-classification result meets the preset classification condition.

[0026] In the embodiment of the present application, by further acquiring the data fluctuation index corresponding to the to-be-classified label after pre-classification to determine whether the pre-classification result meets the conditions, the stability of label classification is further improved.

[0027] In some possible embodiments, after obtaining the performance indicator change information of the pending labeling system relative to the initial labeling system, the method further includes:

[0028] In the case where it is determined based on the performance indicator change information that the pre-classification result does not meet the preset classification condition, the labels to be classified are respectively associated with different label sub-dimensions to form corresponding candidate label systems, and the comparison performance indicator change information corresponding to the labels to be classified in each of the candidate label systems relative to the initial label system is obtained;

[0029] Based on the comparison performance indicator change information, a number of candidate label systems that meet the pre-selection conditions are screened out from the candidate label systems, and a number of corresponding candidate classification results are determined based on the candidate label systems.

[0030] In an embodiment of the present application, when the pre-classification results do not meet the conditions, the labels to be classified are associated with different sub-dimensions to form different candidate label systems, and a number of qualified alternative classification results are selected according to the changes in indicators of these candidate label systems, thereby further improving the efficiency of label management.

[0031] In some possible embodiments, after the initial label system is updated based on the pre-classification result to obtain an updated label system, the method further includes:

[0032] Based on the updated label system, sample labels and their associated sample label sub-dimensions are obtained as training data to retrain the classification model to obtain an updated classification model.

[0033] In an embodiment of the present application, after classifying new tags and updating the tag system, the updated tag system is used to construct training data to retrain the classification model, thereby further improving the accuracy and timeliness of tag classification management.

[0034] In a second aspect, an embodiment of the present application provides a management device for a label system, including:

[0035] A label system acquisition module, used to acquire an initial label system; wherein the initial label system includes at least one label sub-dimension, and each label sub-dimension is associated with at least one label;

[0036] A label pre-classification module, used to determine a label to be classified that needs to be classified, and determine a pre-classification result corresponding to the label to be classified based on a pre-trained classification model; wherein the pre-classification result is represented by classifying the label to be classified into a target sub-dimension, and the target sub-dimension is one of the at least one label sub-dimension;

[0037] A label system update module is used to obtain performance indicator change information of the pending label system relative to the initial label system, and when it is determined based on the performance indicator change information that the pre-classification result meets the preset classification conditions, the initial label system is updated based on the pre-classification result to obtain an updated label system; wherein the pending label system is a label system formed after associating the to-be-classified labels to the target sub-dimension.

[0038] In a third aspect, an embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor can implement the method described in any embodiment of the first aspect when executing the program.

[0039] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method described in any embodiment of the first aspect can be implemented.

[0040] In a fifth aspect, an embodiment of the present application provides a computer program product, wherein the computer program product includes a computer program, wherein when the computer program is executed by a processor, the method described in any embodiment of the first aspect can be implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 A flowchart of a tag system management method provided in an embodiment of the present application;

[0043] Figure 2 A schematic diagram of the structure of a label system management device provided in an embodiment of the present application;

[0044] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0046] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0047] It should be noted that existing technologies have achieved the ability to extract features and generate labels from customer conversations, behavioral data, etc. based on pre-trained large models. These labels can help companies understand the characteristics of different customers, thereby helping to formulate or optimize sales strategies for different customers.

[0048] However, although labels can be automatically generated, effectively organizing these labels into a useful label system usually requires manual intervention by data analysts and business teams, including deciding which labels should be retained, how they should be classified, and how to connect with existing business processes and policies. In addition, the maintenance and update of the label system also mainly rely on manual operations, which usually requires regular checks on the relevance and effectiveness of labels, and adjustments based on new business needs or market conditions.

[0049] It is understandable that manually building and maintaining a labeling system is a time-consuming and inefficient process. The inefficiency will lead to a slow response to market changes and an inability to capture the latest trends in customer behavior in real time. In addition, since the manual management of the labeling system requires reliance on rich professional knowledge and experience, it limits the flexibility and scalability of the labeling system. There is also the problem of low stability in labeling system management.

[0050] In response to the problems existing in the above-mentioned prior art, the present application provides a label system management method, which pre-classifies new labels by training a large model, and then performs the final label classification operation based on the changes in relevant indicators of the label system before and after the pre-classification, thereby avoiding the problem of relying on human expertise or experience, and being able to quickly and accurately classify new labels, effectively improving the efficiency and stability of label system management.

[0051] like Figure 1 As shown, the embodiment of the present application provides a method for managing a label system, which may include the following steps:

[0052] S1. Acquire an initial label system; wherein the initial label system includes at least one label sub-dimension, and each label sub-dimension is associated with at least one label.

[0053] Specifically, the initial label system is the label system that needs to be updated at present, in which the label system can be updated regularly according to the set period, or the update request for the label system can be initiated manually, or the label system update can be automatically triggered according to the generation of new labels, or the label system update can be automatically triggered according to the number of times the new label is matched and hit to reach the set value. It should be noted that the update of the label system is mainly the process of automatically classifying the currently unclassified labels (such as newly generated labels) into the corresponding label sub-dimension.

[0054] The initial label system can be constructed artificially based on business experience. Exemplarily, the label system may include several main label dimensions, such as basic context, insurance awareness, communication willingness, purchase willingness, etc.; each main label dimension corresponds to several label sub-dimensions, for example, the main dimension of basic context includes label sub-dimensions such as "cannot buy insurance" and "never bought free insurance"; each label sub-dimension may be associated with at least one specific label, for example, the label sub-dimension of "cannot buy insurance" may include labels such as "cannot buy insurance" and "I am an elderly person", and the label sub-dimension of "never bought free insurance" may include labels such as "it was not handled by me" and "never received accident insurance".

[0055] S2. Determine the labels to be classified that need to be classified, and determine the pre-classification results corresponding to the labels to be classified based on the pre-trained classification model; wherein the pre-classification results are represented by classifying the labels to be classified into a target sub-dimension, and the target sub-dimension is one of the at least one label sub-dimension.

[0056] Furthermore, the training method of the classification model includes:

[0057] Based on the initial label system, sample labels and their associated sample label sub-dimensions are obtained as training data, and the preset open source large model is trained to obtain a trained classification model.

[0058] Specifically, after determining the label to be classified that needs to be classified according to the set rules, the label to be classified is input into the pre-trained classification model to obtain the pre-classification result output by the classification model. Exemplarily, the pre-classification result can be one of the label sub-dimensions (target sub-dimension) of the initial label system, indicating that the label to be classified currently input is recommended to be classified (associated) into the target sub-dimension.

[0059] Among them, the classification model can be trained using the open source large model LLAMA3, using multiple sets of pre-acquired labels and their corresponding label sub-dimensions as sample data to train the open source large model until the model reaches a preset convergence state or reaches a preset upper limit on the number of training times, so that the large model learns the correspondence between labels and label sub-dimensions, so that unclassified labels can be reasonably classified into corresponding label sub-dimensions.

[0060] S3. Obtain performance indicator change information of the pending label system relative to the initial label system. When it is determined based on the performance indicator change information that the pre-classification result meets the preset classification conditions, the initial label system is updated based on the pre-classification result to obtain an updated label system; wherein the pending label system is a label system formed after associating the labels to be classified to the target sub-dimension.

[0061] Specifically, the labels to be classified can be classified into corresponding target sub-dimensions according to the pre-classification results, forming another pending label system. Then, the performance indicators of the initial label system before pre-classification and the performance indicators corresponding to the pending label system formed by pre-classification are obtained respectively, and then the performance indicator change information of the two is obtained.

[0062] Exemplarily, the performance indicators of the tag system may include: the hit count, hit conversion count, and hit conversion rate calculated based on the user sales scenario data of the preset historical period for matching the tag system. The user sales scenario data may include the user's sales conversation data and the user's behavioral feature data, etc., and one user corresponds to a set of user sales scenario data; the hit count refers to the number of users whose certain tag (or a certain tag sub-dimension) is hit by the user sales scenario data; the hit conversion count refers to the number of users who have been converted (have reached or concluded a transaction) among all users whose certain tag (or a certain tag sub-dimension) is hit by the user sales scenario data; the hit conversion rate refers to the ratio of the hit conversion count to the hit count.

[0063] Then, according to the performance indicator change information, it can be judged whether the preset classification conditions are met, such as whether the difference in the hit conversion rate of the two is less than the set value. If the conditions are met, the initial label system is directly updated using the pre-classification results, that is, the labels to be classified are classified (associated) to the target sub-dimension of the initial label system; if the conditions are not met, the initial label system is not updated, and the labels to be classified continue to be determined. Exemplarily, when the conditions are not met, a reminder message can be issued to remind the management personnel to manually intervene to classify the labels to be classified.

[0064] Based on this, the classification model is trained to automatically pre-classify the labels to be classified, and the classification operation is performed according to the performance changes of the label system before and after the pre-classification, so as to achieve fast and accurate classification of new labels, effectively improving the efficiency and stability of label system management.

[0065] In some possible embodiments, in step S2, determining the tags to be classified that need to be classified may include:

[0066] S201, performing tag matching on user sales scenario data in real time based on a plurality of candidate tags acquired in advance, and recording the number of times each candidate tag is matched;

[0067] S202: When the number of times that a target candidate tag among a plurality of candidate tags is matched exceeds a preset number threshold, the target candidate tag is determined as a tag to be classified that needs to be classified.

[0068] It should be noted that after new tags are generated, these new tags can be temporarily stored in a specific location, and then matched based on the user sales scenario data obtained in real time, and the number of times these new tags are hit is recorded. Among them, when a user's sales scenario data (such as conversation data) matches a tag once, it is considered that the tag is hit once.

[0069] Then, when the number of times a tag is matched exceeds a preset threshold, it is determined that the tag needs to be classified into the tag system. In addition, based on a preset period, a tag with the most matching hits can be regularly selected from all new tags to be determined as the tag to be classified.

[0070] In some possible embodiments, in step S3, obtaining performance indicator change information of the pending labeling system relative to the initial labeling system may include:

[0071] S3011. Based on the preset historical user telephone sales scenario data, obtain the first hit conversion influence direction of the target sub-dimension relative to the initial label system, and obtain the second hit conversion influence direction of the target sub-dimension relative to the pending label system; wherein, if the hit conversion rate corresponding to the target sub-dimension is greater than the average hit conversion rate of all label sub-dimensions under the current label system, then the hit conversion influence direction of the target sub-dimension relative to the current label system is positive; if the hit conversion rate corresponding to the target sub-dimension is less than the average hit conversion rate of all label sub-dimensions under the current label system, then the hit conversion influence direction of the target sub-dimension relative to the current label system is negative;

[0072] In step S3, determining whether the pre-classification result meets the preset classification condition based on the performance indicator change information includes:

[0073] S3012: If it is determined that the first hit conversion influence direction is consistent with the second hit conversion influence direction, it is determined that the pre-classification result meets the preset classification condition.

[0074] Specifically, in step S3011, firstly, the user telephone sales scenario data of the preset historical period is matched with the initial label system and the pending label system respectively, wherein, in the process of matching the user telephone sales scenario data with the initial label system, firstly, the overall average hit conversion rate R1 of all labels in the initial label system is calculated, and then the average hit conversion rate r1 of all labels under the target sub-dimension is calculated, and then R1 and r1 are compared: if r1 is greater than R1, it is determined that the target sub-dimension has a positive impact on the hit conversion index relative to the index system (the hit conversion impact direction is positive); if r1 is less than R1, it is determined that the target sub-dimension has a negative impact on the hit conversion index relative to the index system (the hit conversion impact direction is negative). Based on this, the first hit conversion impact direction of the target sub-dimension relative to the initial label system can be obtained, and then after adding the to-be-classified labels to form the pending label system, the second hit conversion impact direction of the target sub-dimension relative to the pending label system can be obtained in the same way.

[0075] Then, determine whether the first hit conversion impact direction is consistent with the second hit conversion impact direction. If they are consistent, it means that the target sub-dimensions before and after adding the label to be classified have the same impact on the label system and meet the classification conditions; if they are inconsistent, it means that the target sub-dimensions before and after adding the label to be classified have different impacts on the label system and do not meet the classification conditions.

[0076] Based on this, by comparing the consistency of the hit conversion impact direction of the pre-classified target sub-dimension before and after classification, it is determined whether the pre-classification result meets the classification conditions, thereby further improving the stability of label classification.

[0077] In some possible embodiments, in step S3, obtaining performance indicator change information of the pending labeling system relative to the initial labeling system may also include:

[0078] S3021. Based on historical user telephone sales scenario data, obtain data fluctuation indicators corresponding to the to-be-classified labels in the to-be-determined label system; wherein the data fluctuation indicators include data standard scores;

[0079] In step S3, it is determined based on the performance indicator change information whether the pre-classification result meets the preset classification condition, specifically:

[0080] S3022: If it is determined that the first hit conversion impact direction is consistent with the second hit conversion impact direction, and the data fluctuation index is within a preset tolerance threshold, it is determined that the pre-classification result meets the preset classification condition.

[0081] It should be noted that, on the basis of step 3011, the process of step S3021 can also be added. Specifically, after the tags to be classified are added to form a tag system to be determined, the relevant data fluctuation indicators corresponding to the tags to be classified relative to the tag system to be determined can be calculated based on the user telephone sales scenario data of a preset historical period, such as the Z-score absolute value (data standard score) of the hit conversion rate of the tags to be classified, where the Z-score calculation formula is: z = (r-μ) / σ, where r represents the hit conversion rate corresponding to the tag to be classified, μ represents the average value of the hit conversion rate, and σ represents the standard deviation of the hit conversion rate.

[0082] Then, if it is determined that the first hit conversion impact direction is consistent with the second hit conversion impact direction, and the corresponding | z | If the value of the first hit conversion influence direction is less than the preset tolerance threshold, it is determined that the pre-classification result meets the preset classification condition; otherwise, if the conversion influence direction of the first hit is inconsistent with the conversion influence direction of the second hit, or the corresponding label to be classified is | z | If the value is not less than the preset tolerance threshold, it is determined that the pre-classification result does not meet the preset classification condition.

[0083] In some possible embodiments, after obtaining the performance indicator change information of the pending labeling system relative to the initial labeling system, the following may also be included:

[0084] S3031. When it is determined based on the performance indicator change information that the pre-classification result does not meet the preset classification condition, the labels to be classified are respectively associated with different label sub-dimensions to form corresponding candidate label systems, and the comparison performance indicator change information of the labels to be classified in each candidate label system relative to the initial label system is obtained;

[0085] S3032: Based on the information on changes in the performance indicators of each comparison, a number of candidate label systems that meet the pre-selection conditions are selected from each candidate label system, and a number of corresponding candidate classification results are determined based on each candidate label system.

[0086] It should be noted that when it is determined that the results of automatic pre-classification based on the classification model do not meet the conditions, other indicator data can also be calculated and corresponding alternative classification results can be screened out to facilitate manual intervention to quickly confirm the classification from these alternative classification results.

[0087] Specifically, first, all label sub-dimensions of the initial label system can be traversed, and the labels to be classified can be classified (associated) to these label sub-dimensions respectively, to form multiple corresponding candidate label systems; for each formed candidate label system, the corresponding hit conversion influence direction of the corresponding label sub-dimension before and after the addition of the label to be classified is obtained (refer to step S3011 to obtain); then, the corresponding candidate label systems with consistent hit conversion influence directions before and after the addition of the new label are preliminarily screened out; then, based on each of the preliminarily screened candidate label systems, the statistical information corresponding to these candidate label systems, such as the variance and standard deviation of the hit conversion rate, etc., are calculated respectively, and then a number of candidate label systems (the number can be set according to needs, for example, 5) with the smallest variance (or standard deviation, etc.) are selected as alternative label systems, and then the classification results corresponding to these alternative label systems are used as alternative classification results.

[0088] After obtaining the alternative classification results, a manual intervention prompt can be initiated, or these alternative classification results can be sent to a manual management terminal so that the management personnel can quickly determine which label sub-dimension of the initial label system to add the current label to be classified to, or add a label sub-dimension based on these alternative classification results.

[0089] In some possible embodiments, after step S3, the following steps may also be included:

[0090] S4. Based on the updated label system, sample labels and their associated sample label sub-dimensions are obtained as training data to retrain the classification model to obtain an updated classification model.

[0091] It should be noted that after each classification of the labels to be classified to form a new label system, in order to ensure the classification accuracy and timeliness of the classification model, it is necessary to reconstruct the sample data based on the updated label system to train the classification model so as to adjust and update the classification model.

[0092] Based on this, after classifying the new labels and updating the label system, the updated label system is used to construct training data to retrain the classification model, thereby further improving the accuracy and timeliness of label classification management.

[0093] It should be noted that, compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0094] 1. Improved efficiency: By automatically building and maintaining the label system, the proportion of manual operations is effectively reduced, thereby improving the overall operational efficiency and response speed of label management.

[0095] 2. Enhanced real-time performance: Dynamic adjustment and optimization mechanisms enable the labeling system to quickly adapt to changes in the market and customer behavior, enhancing business agility and competitiveness.

[0096] 3. Reduce professional dependence: The method of automated management of the label system reduces the dependence on the professional knowledge of data analysts, making the management and optimization of the label system more popular and easier for non-professionals to participate and use.

[0097] 4. Improve decision-making quality: By building and maintaining a structured and logically clear labeling system, it provides companies with more accurate and in-depth customer insights, helping to make more effective decisions.

[0098] Please refer to Figure 2 , Figure 2 FIG. 1 shows a block diagram of a management device for a label system provided in some embodiments of the present application. It should be understood that the management device for the label system is similar to the above-mentioned Figure 1 Corresponding to the method embodiment, the various steps involved in the above method embodiment can be executed. The specific functions of the management device of the label system can be found in the description above. To avoid repetition, the detailed description is appropriately omitted here.

[0099] Figure 2 The label system management device includes at least one software function module that can be stored in a memory in the form of software or firmware or fixed in the label system management device, and the label system management device includes:

[0100] The label system acquisition module 210 is used to acquire an initial label system; wherein the initial label system includes at least one label sub-dimension, and each label sub-dimension is associated with at least one label;

[0101] The tag pre-classification module 220 is used to determine the tags to be classified that need to be classified, and determine the pre-classification results corresponding to the tags to be classified based on the pre-trained classification model; wherein the pre-classification results are represented by classifying the tags to be classified into a target sub-dimension, and the target sub-dimension is one of the at least one tag sub-dimension;

[0102] The label system update module 230 is used to obtain the performance indicator change information of the pending label system relative to the initial label system. When it is determined based on the performance indicator change information that the pre-classification result meets the preset classification conditions, the initial label system is updated based on the pre-classification result to obtain an updated label system; wherein the pending label system is a label system formed after associating the labels to be classified to the target sub-dimension.

[0103] It can be understood that the above-mentioned device item embodiment corresponds to the method item embodiment of the present invention. A label system management device provided by the embodiment of the present invention can implement the label system management method provided by any method item embodiment of the present invention.

[0104] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the device described above can refer to the corresponding process in the aforementioned method, and will not be described in detail here.

[0105] like Figure 3 As shown, some embodiments of the present application provide an electronic device 300, which includes: a memory 310, a processor 320, and a computer program stored in the memory 310 and executable on the processor 320, wherein the processor 320 reads the program from the memory 310 through the bus 330 and executes the program to implement the method of any embodiment included in the above-mentioned label system management method.

[0106] Processor 320 can process digital signals and can include various computing structures, such as complex instruction set computer structure, reduced instruction set computer structure, or a structure that implements a combination of multiple instruction sets. In some examples, processor 320 can be a microprocessor.

[0107] The memory 310 may be used to store instructions executed by the processor 320 or data related to the execution of instructions. These instructions and / or data may include codes for implementing some or all functions of one or more modules described in the embodiments of the present application. The processor 320 of the disclosed embodiment may be used to execute instructions in the memory 310 to implement the method shown above. The memory 310 includes a dynamic random access memory, a static random access memory, a flash memory, an optical memory, or other memory known to those skilled in the art.

[0108] Some embodiments of the present application further provide a computer-readable storage medium having a computer program stored thereon. The computer program is executed by a processor to execute the method described in the method embodiment.

[0109] Some embodiments of the present application further provide a computer program product, which, when executed on a computer, enables the computer to execute the method described in the method embodiment.

[0110] It should be noted that each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the embodiments can be referred to each other. For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0111] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0112] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0113] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0114] The above description is only an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0115] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0116] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

Claims

1. A method for managing a label system, characterized in that: include: Acquire an initial label system; wherein the initial label system includes at least one label sub-dimension, and each label sub-dimension is associated with at least one label; Determine a label to be classified that needs to be classified, and determine a pre-classification result corresponding to the label to be classified based on a pre-trained classification model; wherein the pre-classification result is represented by classifying the label to be classified into a target sub-dimension, and the target sub-dimension is one of the at least one label sub-dimension; The performance indicator change information of the pending label system relative to the initial label system is obtained. When it is determined based on the performance indicator change information that the pre-classification result meets the preset classification condition, the initial label system is updated based on the pre-classification result to obtain an updated label system; wherein the pending label system is a label system formed after associating the labels to be classified to the target sub-dimension.

2. The label system management method according to claim 1, characterized in that: The step of determining the tags to be classified that need to be classified includes: Based on several candidate tags acquired in advance, tag matching is performed on the user sales scenario data in real time, and the number of times each candidate tag is matched is recorded respectively; When the number of times that a target candidate tag among the plurality of candidate tags is matched and hit exceeds a preset number threshold, the target candidate tag is determined as a tag to be classified that needs to be classified.

3. The label system management method according to claim 1, characterized in that: The training method of the classification model includes: Based on the initial label system, sample labels and their associated sample label sub-dimensions are obtained as training data, and the preset open source large model is trained to obtain a trained classification model.

4. The label system management method according to claim 1, characterized in that: The obtaining of performance indicator change information of the pending labeling system relative to the initial labeling system includes: Based on the preset historical user telephone sales scenario data, the first hit conversion influence direction of the target sub-dimension relative to the initial label system is obtained, and the second hit conversion influence direction of the target sub-dimension relative to the pending label system is obtained; wherein, if the hit conversion rate corresponding to the target sub-dimension is greater than the average hit conversion rate of all label sub-dimensions under the current label system, then the hit conversion influence direction of the target sub-dimension relative to the current label system is positive; if the hit conversion rate corresponding to the target sub-dimension is less than the average hit conversion rate of all label sub-dimensions under the current label system, then the hit conversion influence direction of the target sub-dimension relative to the current label system is negative; The determining, based on the performance indicator change information, that the pre-classification result meets a preset classification condition includes: If it is determined that the first hit-conversion influence direction is consistent with the second hit-conversion influence direction, it is determined that the pre-classification result meets a preset classification condition.

5. The label system management method according to claim 4, characterized in that: The obtaining of performance indicator change information of the pending labeling system relative to the initial labeling system also includes: Based on the historical user telephone sales scenario data, obtaining the data fluctuation index corresponding to the to-be-classified label in the to-be-determined label system; wherein the data fluctuation index includes a data standard score; The determining, based on the performance indicator change information, that the pre-classification result meets the preset classification condition is specifically: If it is determined that the first hit conversion impact direction is consistent with the second hit conversion impact direction, and the data fluctuation index is within a preset tolerance threshold, it is determined that the pre-classification result meets the preset classification condition.

6. The label system management method according to claim 1, characterized in that: After obtaining the performance indicator change information of the pending label system relative to the initial label system, the method further includes: In the case where it is determined based on the performance indicator change information that the pre-classification result does not meet the preset classification condition, the labels to be classified are respectively associated with different label sub-dimensions to form corresponding candidate label systems, and the comparison performance indicator change information corresponding to the labels to be classified in each of the candidate label systems relative to the initial label system is obtained; Based on the comparison performance indicator change information, a number of candidate label systems that meet the pre-selection conditions are screened out from the candidate label systems, and a number of corresponding candidate classification results are determined based on the candidate label systems.

7. The label system management method according to claim 1, characterized in that: After the initial label system is updated based on the pre-classification result to obtain an updated label system, the method further includes: Based on the updated label system, sample labels and their associated sample label sub-dimensions are obtained as training data to retrain the classification model to obtain an updated classification model.

8. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor can implement the label system management method described in any one of claims 1 to 7 when executing the program.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for managing a label system according to any one of claims 1 to 7 is executed.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method for managing the label system according to any one of claims 1 to 7 is implemented.

Images