A method, apparatus, and device for updating a classification model based on user behavior.

Abstract

This invention discloses a method, apparatus, and device for updating a classification model based on user behavior. The method includes: acquiring the original user table dataset and newly added data corresponding to the original classification model, wherein the original table dataset contains the original data; acquiring a new table dataset based on the original table dataset and the newly added data; determining the validity of the new table dataset; if the new table dataset is determined to be valid, acquiring the valid fields from the new table dataset; updating the original table dataset based on the valid fields; and updating the original classification model using the updated original table dataset. By acquiring the original table dataset and newly added data from the original classification model to establish a new table dataset, and by increasing the dimensionality of the classification model through the selected valid fields, and then updating the original classification model using the updated original table dataset, the accuracy of the classification model calculation is improved. Simultaneously, iterative updates of the classification model are achieved without manual intervention, thus saving labor costs.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method, apparatus, and device for updating a classification model based on user behavior. Background Technology

[0002] In the internet, user behavior generates a large amount of data, including historical data and new data. In order to recommend targeted information to users or to carry out necessary monitoring and management during user operations, it is usually necessary to classify and predict based on the user's operation behavior in network applications and for pre-set goals.

[0003] The traditional approach is that when new data is acquired, business personnel first analyze and select potentially effective data for algorithm engineers. Then, the algorithm engineers use data mining techniques to further filter data that contributes highly to classification value and input it into the model for retraining and iterating.

[0004] In existing technologies, many models, after being carefully selected and feature-engineered by algorithm engineers to complete the initial modeling, consume a lot of manual labor costs. Moreover, the input parameters are relatively fixed, and the dimensions and directions covered by the input parameters are also relatively fixed. This inevitably leads to the loss of data in certain dimensions, which makes the model's recognition and judgment capabilities insufficient in certain directions. Summary of the Invention

[0005] This invention provides a method, apparatus, and device for updating a classification model based on user behavior, so as to update the classification model for effective new data in user behavior.

[0006] According to one aspect of the present invention, a method for updating a classification model based on user behavior is provided, comprising:

[0007] Obtain the original user table dataset and newly added data corresponding to the original classification model. The original table dataset contains the original data.

[0008] Obtain the new table dataset based on the original table dataset and the newly added data;

[0009] Determine the validity of the new table dataset. If the new table dataset is found to be valid, retrieve the valid fields from the new table dataset and update the original table dataset based on the valid fields.

[0010] The original classification model is updated using the updated original dataset.

[0011] Preferably, obtaining a new table dataset based on the original table dataset and the newly added data includes: determining the new fields in the newly added data; performing missing value processing on the new fields to obtain a first processing field; performing extreme value processing or normalization processing on the first processing field to obtain a second processing field; and adding the newly added data corresponding to the second processing field to the original table dataset to obtain the new table dataset.

[0012] Preferably, determining the new fields in the new data includes: obtaining all fields in the new data; filtering all fields to obtain structured fields, and using the structured fields as new fields.

[0013] Preferably, the missing value processing of the newly added fields to obtain the first processing field includes: taking each newly added field as a target field in turn; determining whether the ratio of the missing data amount of the newly added data corresponding to the target field to the total data amount is greater than a first preset threshold; if so, taking the target field as the first processing field; otherwise, deleting the target field.

[0014] Preferably, the process of performing extreme value processing or normalization processing on the first processing field to obtain the second processing field includes: arranging the newly added data corresponding to the first processing field from smallest to largest to obtain a first processing set; obtaining the difference or ratio between two specified positions in the first processing set; determining whether the difference or ratio is less than a second preset threshold; if so, performing extreme value processing on the first processing field to obtain the second processing field; otherwise, performing normalization processing on the first processing field to obtain the second processing field.

[0015] Preferably, determining the validity of the new table dataset includes: obtaining relevant data metrics for the original table dataset and the new table dataset, wherein the relevant data metrics include precision, recall, and area under the receiver operating feature curve; calculating the difference between each identical relevant data metric in the original table dataset and the new table dataset; determining whether the difference meets a preset condition, and if it does, determining that the new table dataset is valid; otherwise, determining that the new table dataset is invalid.

[0016] Preferably, obtaining relevant data metrics for the original table dataset and the new table dataset includes: dividing the original table dataset and the new table dataset into a first dataset and a second dataset respectively, and obtaining the actual labels of the first dataset and the second dataset input by the user; training the original classification model using the first dataset and the actual labels of the first dataset to obtain a trained classification model; calculating the calculated labels for the second dataset using the trained classification model, and calculating the relevant data metrics for the actual labels and the calculated labels.

[0017] Preferably, when the new table dataset is determined to be valid, the valid fields in the new table dataset are obtained, and the original table dataset is updated based on the valid fields. This includes: obtaining the contribution of each field in the new table dataset and the original table dataset; determining the minimum contribution of the fields in the original table dataset, and taking the fields in the new table dataset whose contribution is greater than the minimum contribution as valid fields; and adding the valid fields and the corresponding new data to the original table dataset to update the original table dataset.

[0018] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0019] At least one processor; and

[0020] A memory communicatively connected to the at least one processor; wherein,

[0021] The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform a classification model update method for user behavior as described in any embodiment of the present invention.

[0022] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement a classification model update method for user behavior as described in any embodiment of the present invention.

[0023] The technical solution of this invention establishes a new data table by acquiring the original data table dataset and newly added data from the original classification model. Furthermore, it increases the dimension of the classification model by selecting effective fields, and then updates the original classification model using the updated original data table dataset. This improves the accuracy of the classification model calculation and enables iterative updates of the classification model without manual intervention, thereby saving labor costs.

[0024] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1This is a flowchart of a classification model update method for user behavior provided in Embodiment 1 of the present invention;

[0027] Figure 2 This is a flowchart of another method for updating a classification model for user behavior according to Embodiment 1 of the present invention;

[0028] Figure 3 This is a flowchart of another method for updating a classification model for user behavior according to Embodiment 2 of the present invention;

[0029] Figure 4 This is a schematic diagram of a classification model update device for user behavior according to Embodiment 3 of the present invention;

[0030] Figure 5 This is a schematic diagram of the structure of an electronic device that implements a classification model update method for user behavior according to an embodiment of the present invention. Detailed Implementation

[0031] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0032] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0033] Example 1

[0034] Figure 1This invention provides a flowchart of a method for updating a classification model based on user behavior, as described in Embodiment 1. This embodiment is applicable to updating the classification model for valid new data in user behavior. The method can be executed by a device for updating a classification model based on user behavior, which can be implemented in hardware and / or software and can be configured in a computer. Figure 1 As shown, the method includes:

[0035] S110. Obtain the original user table dataset and newly added data corresponding to the original classification model.

[0036] Specifically, the original classification model is a classification model trained on the original dataset. The classification model is used to classify the data input by technicians. The original dataset contains the original data, which refers to the historical data of users in multiple dimensions. Users refer to the objects processed by the classification model. Historical data includes, but is not limited to, the user's gender, age, and other behavioral data related to the classification model. New data refers to new data collected by the controller other than the original dataset. That is, when the controller detects new data, it will obtain the original dataset of users and the new data corresponding to the original classification model so as to update the original classification model in a timely manner.

[0037] S120. Obtain the new table dataset based on the original table dataset and the newly added data.

[0038] Figure 2 This invention provides a flowchart of a method for determining risks in industrial control network data according to Embodiment 1. Step S120 mainly includes the following steps S121 to S124:

[0039] S121. Determine the new fields in the newly added data.

[0040] Preferably, determining the new fields in the new data includes: obtaining all fields in the new data; filtering all fields to obtain structured fields, and using the structured fields as new fields.

[0041] Specifically, when the controller detects new data, it retrieves all fields from the new data. A field refers to a data item, and each field corresponds to all new data information under that field. For example, a field could be "gender," and all new data related to "gender" would be the new data corresponding to that "gender" field. The controller filters all fields contained in the new data, separating numerical data from factor data. Numerical data refers to data composed of numbers, while factor data refers to grouped data. For example, the field "number of meetings attended" contains various different numerical information, making it numerical data. The field "gender" is divided into two groups, "male" and "female." The controller can represent this type of factor data numerically, using 1 to represent "male" and 2 to represent "female." In other words, the controller can treat the fields corresponding to both numerical data and numerically represented factor data as structured fields and use these structured fields as new fields.

[0042] S122. The missing value processing of the newly added field is used to obtain the first processed field.

[0043] Preferably, the missing value processing of the newly added fields to obtain the first processing field includes: taking each newly added field as a target field in turn; determining whether the ratio of the missing data amount of the newly added data corresponding to the target field to the total data amount is greater than a first preset threshold; if so, taking the target field as the first processing field; otherwise, deleting the target field.

[0044] Specifically, after the controller obtains a new field, it will handle missing values ​​for the new field. That is, each new field will be treated as a target field in turn. Then, the ratio of the missing data of the new data corresponding to the target field to the total data volume will be compared with a first preset threshold. The first preset threshold is set in advance by the developers within the controller according to the data volume requirements of the classification model update. If the data volume of the new data corresponding to the target field is less than the first preset threshold, the controller will delete the target field. Otherwise, the controller will treat the target field as the first processing field and fill the empty value positions in the first processing field with 0.

[0045] For example, when the total number of new data entries corresponding to each target field is 10, the developers set the first preset threshold to 0.5. There are 4 missing data entries corresponding to the target field "Number of Views". The ratio of the number of missing data entries to the total number of data entries is 0.4, which is less than the first preset threshold. The controller will delete "Number of Views" and will not use this field as new data for subsequent model updates. However, there are 2 missing data entries corresponding to the target field "Number of Clicks". The ratio of the number of missing data entries to the total number of data entries is 0.2, which is greater than the first preset threshold. The controller will use the new data entries corresponding to "Number of Clicks" as the first processing field and fill the data positions of the 2 missing data entries in "Number of Clicks" with 0.

[0046] S123. Perform extreme value processing or normalization on the first processing field to obtain the second processing field.

[0047] Preferably, the process of performing extreme value processing or normalization processing on the first processing field to obtain the second processing field includes: arranging the newly added data corresponding to the first processing field from smallest to largest to obtain a first processing set; obtaining the difference or ratio between two specified positions in the first processing set; determining whether the difference or ratio is less than a second preset threshold; if so, performing extreme value processing on the first processing field to obtain the second processing field; otherwise, performing normalization processing on the first processing field to obtain the second processing field.

[0048] Specifically, after obtaining the first processing field, the controller performs either extreme value processing or normalization processing on the first processing field to obtain the second processing field. The controller selects different processing methods based on the degree of dispersion of the newly added data. When the dispersion of the newly added data is small, extreme value processing is used; when the dispersion of the newly added data is large, normalization processing is used. When determining the degree of dispersion of the newly added data, the controller arranges the corresponding newly added data in the first processing field from small to large to obtain the first processing set. Then, it calculates the difference or ratio between two specified positions in the first processing set. The degree of dispersion of the newly added data can be determined by the calculated difference or ratio. The controller compares the difference or ratio with a second preset threshold, which is predetermined by the R&D personnel. The settings within the controller indicate that if the difference or ratio is less than the second preset threshold, it means the dispersion of the new data is small. In this case, the controller will perform extreme value processing on the first processing field. If the difference or ratio is greater than the second preset threshold, it means the dispersion of the new data is large. In this case, the controller will perform normalization processing on the first processing field. For example, if the new data value at the 80% position and the new data value at the 90% position in the first processing set generated by the first processing field K are 1.5 and 1.6 respectively, and the controller calculates that the difference between the two positions is 0.1, which is less than the second preset threshold of 1, it means that the dispersion of the new data corresponding to the first processing field K is small. In this case, extreme value processing should be performed on the first processing field K to generate the second processing field.

[0049] Furthermore, extreme value processing is necessary because some extreme values ​​in the data can affect the stability of the model. Therefore, after obtaining the first processing field, the controller will delete the maximum and minimum values ​​in the first processing set. For example, the controller takes the new data in the first set that are located between 10% and 90% as the new data corresponding to the second processing field. For data with a high degree of dispersion, the controller can map all the natural numbers in the new data to a small numerical range, and then use the data in that range as the new data corresponding to the second processing field. After normalization, the controller's computing power can be improved, memory pressure can be reduced, and computation time can be saved.

[0050] S124. Add the newly added data corresponding to the second processing field to the original table dataset to create a new table dataset.

[0051] Specifically, the controller can add the newly added data corresponding to the second processing field to the original table dataset to obtain a new table dataset. The time span of the new table dataset is consistent with that of the original table dataset. For example, if the original data in the original table dataset was collected from August 1 to August 5, 2022, the collection time of the new table dataset should also be from August 1 to August 5, 2022.

[0052] S130. Determine the validity of the new table dataset. When the new table dataset is determined to be valid, obtain the valid fields in the new table dataset and update the original table dataset based on the valid fields.

[0053] Specifically, after obtaining the new table dataset, the controller will first determine the validity of the new table dataset. Only when the new table dataset is valid will subsequent update operations be performed. The purpose of determining the validity of the new table dataset is to avoid invalid updates to the original classification model and to ensure the accuracy of the original classification model calculation.

[0054] Preferably, when the new table dataset is determined to be valid, the valid fields in the new table dataset are obtained, and the original table dataset is updated based on the valid fields. This includes: obtaining the contribution of each field in the new table dataset and the original table dataset; determining the minimum contribution of the fields in the original table dataset, and taking the fields in the new table dataset whose contribution is greater than the minimum contribution as valid fields; and adding the valid fields and the corresponding new data to the original table dataset to update the original table dataset.

[0055] Specifically, after confirming the validity of the new table dataset, the controller obtains the contribution of all fields in both the new and original table datasets and sorts them. Contribution refers to the degree to which a field contributes to the classification model. After the data is input into the classification model for classification calculation and results are generated, the contribution of each field in the dataset within the classification model can be obtained. The controller determines the minimum contribution value of a field in the original table dataset and then selects fields in the new table dataset with a contribution value greater than that value as valid fields. For example, if the original table dataset has three fields with contributions of 15, 12, and 8, and the new table dataset has three fields with contributions of 10, 7, and 3, then since the field with a contribution of 10 in the new table dataset is greater than the minimum contribution of 8 in the original table dataset, the controller will select the field with a contribution of 10 in the new table dataset as a valid field. Finally, the valid fields and the corresponding newly added datasets are added to the original table dataset to update the original table dataset, meaning the updated original table dataset includes four fields.

[0056] S140. Update the original classification model using the updated original dataset.

[0057] Specifically, after the controller obtains the updated original table dataset, it can update the original classification model. Furthermore, whenever the controller detects new data, it will create a new table dataset. When the new table dataset is valid, it will determine the valid fields of the new table and then update the original table dataset based on the valid fields, so as to realize the iterative process of the original table dataset in the original classification model.

[0058] The technical solution of this invention establishes a new data table by acquiring the original data table dataset and newly added data from the original classification model. Furthermore, it increases the dimension of the classification model by selecting effective fields, and then updates the original classification model using the updated original data table dataset. This improves the accuracy of the classification model calculation and enables iterative updates of the classification model without manual intervention, thereby saving labor costs.

[0059] Example 2

[0060] Figure 3 This is a flowchart of a classification model update method for user behavior provided in Embodiment 2 of the present invention. This embodiment, based on Embodiment 1 above, specifically explains the determination of the validity of the new table dataset. For example... Figure 3 As shown, the method includes:

[0061] S210. Obtain relevant data metrics for the original table dataset and the new table dataset.

[0062] Specifically, the relevant data metrics include precision, recall, and the area under the receiver operating characteristic (ROC) curve. Precision refers to the percentage of correctly calculated results out of the total number of results. Recall is the probability that a result that is actually positive is predicted as positive. AUC is the area under the ROC curve calculated from the ROC curve. Since the ROC curve is generally above the line y=x, the value of AUC ranges between 0.5 and 1. The closer the AUC is to 1.0, the more accurate the calculation result of the classification model. Since precision, recall, and the area under the ROC curve are existing technologies for judging the calculation results of classification models, the specific calculation methods will not be described in detail in this embodiment.

[0063] Preferably, obtaining relevant data metrics for the original table dataset and the new table dataset includes: dividing the original table dataset and the new table dataset into a first dataset and a second dataset respectively, and obtaining the actual labels of the first dataset and the second dataset input by the user; training the original classification model using the first dataset and the actual labels of the first dataset to obtain a trained classification model; calculating the calculated labels for the second dataset using the trained classification model, and calculating the relevant data metrics for the actual labels and the calculated labels.

[0064] Specifically, calculating relevant data metrics requires the controller to divide the original table dataset and the new table dataset into a first dataset and a second dataset, respectively. The first dataset serves as the training set for training the classification model, and the second dataset serves as the test set for testing the classification model. The controller will perform the division according to the criteria set by the developers. For example, if the developers set 75% for the first dataset and 25% for the second dataset, the controller will divide both the original table dataset (75% training set and 25% test set) and the new table dataset (75% training set and 25% test set). After the division, the controller will obtain the first and second datasets input by the user. The actual labels of the test set are used to train the original classification model using the actual labels of the first dataset from the original dataset and the first dataset from the new dataset. Then, the trained classification model is used to calculate the calculated labels on the second dataset. Finally, the controller calculates relevant data metrics based on the actual and calculated labels. For example, if the actual label of a user M in the second dataset is "attended" while the calculated label is "not attended", the controller can determine that the actual label and the calculated label do not match, and then calculate relevant data metrics such as precision, recall, and area under the receiver operating feature curve.

[0065] S220. Calculate the difference between the same related data indicators in the original table dataset and the new table dataset.

[0066] Specifically, the controller obtains the precision, recall, and area under the receiver operating feature (AUC) calculated from the original dataset and the new dataset, and calculates the difference between the same related data metrics. For example, if the precision of the original dataset is 0.5, the recall is 0.8, and the AUC is 0.8, while the precision of the new dataset is 0.6, the recall is 0.82, and the AUC is 0.78, the controller can calculate the difference in precision as -0.1 (0.5-0.6), the difference in recall as -0.02 (0.8-0.82), and the difference in AUC as 0.2 (0.78-0.8).

[0067] S230. Determine whether the difference meets the preset conditions. If yes, execute S240; otherwise, execute S250.

[0068] Specifically, after calculating the difference between the various related data indicators, the controller will determine whether the difference meets the preset conditions. The preset conditions are set in advance by the R&D personnel in the controller based on the performance of the classification model, and can be adjusted according to the calculation requirements. When the difference meets the preset conditions, S240 can be executed; otherwise, S250 can be executed.

[0069] S240. Determine if the new table dataset is valid.

[0070] For example, the preset condition set by the R&D personnel is that any two of the differences among the three relevant data indicators are less than 0. Since the difference in accuracy is -0.1 and the difference in recall is -0.02, both of which are less than 0, the preset condition is met, and the new table dataset is determined to be valid.

[0071] S250, Determined that the new table dataset is invalid.

[0072] For example, the preset condition set by the R&D personnel is that the difference between the three relevant data indicators is less than 0. Since the difference of AUC is 0.2, which is greater than 0, the preset condition is not met. At this time, the new table dataset is determined to be invalid. When the new table dataset is invalid, it means that the performance of the training classification model trained on the new table dataset is not as good as the training classification model trained on the original table dataset. At this time, no further update operation is performed on the invalid new table dataset.

[0073] The technical solution of this invention establishes a new data table by acquiring the original data table from the original classification model and adding new data. When the validity of the new data table is determined, the valid fields are then filtered. This avoids updating the original classification model in a way that would have a negative effect. Furthermore, the dimensions of the classification model can be increased by the filtered valid fields. The updated original data table is then used to update the original classification model, which improves the accuracy of the classification model calculation. It also enables iterative updates of the classification model without the need for manual intervention, thereby saving labor costs.

[0074] Example 3

[0075] Figure 4 This is a schematic diagram of a classification model update device for user behavior provided in Embodiment 3 of the present invention. Figure 4 As shown, the device includes: a new data acquisition module 310, used to acquire the original user table dataset and new data corresponding to the original classification model, wherein the original table dataset contains the original data; a new table dataset acquisition module 320, used to acquire a new table dataset based on the original table dataset and the new data; an original table dataset update module 330, used to determine the validity of the new table dataset, and when the new table dataset is determined to be valid, acquire the valid fields in the new table dataset and update the original table dataset based on the valid fields; and a classification model update module 340, used to update the original classification model using the updated original table dataset.

[0076] Preferably, the new table dataset acquisition module 320 specifically includes: a new field determination unit, used to determine the new fields in the new data; a first processing field acquisition unit, used to process the missing values ​​of the new fields to obtain the first processing field; a second field acquisition unit, used to process the first processing field by performing extreme value processing or normalization processing to obtain the second processing field; and a new table dataset generation unit, used to add the new data corresponding to the second processing field to the original table dataset to obtain the new table dataset.

[0077] Preferably, the newly added field determination unit is specifically used for: obtaining all fields in the newly added data; filtering all fields to obtain structured fields, and using the structured fields as newly added fields.

[0078] Preferably, the first processing field acquisition unit is specifically used for: sequentially taking each newly added field as a target field; determining whether the ratio of the missing data amount of the newly added data corresponding to the target field to the total data amount is greater than a first preset threshold; if so, taking the target field as the first processing field; otherwise, deleting the target field.

[0079] Preferably, the second processing field acquisition unit is specifically used for: arranging the newly added data corresponding to the first processing field from small to large to obtain a first processing set; obtaining the difference or ratio between two specified positions in the first processing set; determining whether the difference or ratio is less than a second preset threshold; if so, performing extreme value processing on the first processing field to obtain the second processing field; otherwise, performing normalization processing on the first processing field to obtain the second processing field.

[0080] Preferably, the original table dataset update module 330 specifically includes: a relevant data indicator acquisition unit, used to acquire relevant data indicators of the original table dataset and the new table dataset, wherein the relevant data indicators include precision, recall, and area under the receiver operation feature curve; a difference calculation unit, used to calculate the difference between each identical relevant data indicator in the original table dataset and the new table dataset; and a new table dataset validity determination unit, used to determine whether the difference meets the preset conditions. If it does, the new table dataset is determined to be valid; otherwise, the new table dataset is determined to be invalid.

[0081] Preferably, the relevant data indicator acquisition unit is specifically used for: dividing the original table dataset and the new table dataset into a first dataset and a second dataset respectively, and obtaining the actual labels of the first dataset and the second dataset input by the user; training the original classification model using the first dataset and the actual labels of the first dataset to obtain a trained classification model; calculating the calculated labels of the second dataset using the trained classification model, and calculating the relevant data indicators of the actual labels and the calculated labels.

[0082] Preferably, the original table dataset update module 330 further includes: an original table dataset update unit, used to obtain the contribution of each field in the new table dataset and the original table dataset; determine the minimum contribution of the fields in the original table dataset, and take the fields in the new table dataset whose contribution is greater than the minimum contribution as valid fields; and add the valid fields and the new data corresponding to the valid fields to the original table dataset to update the original table dataset.

[0083] The technical solution of this invention establishes a new data table by acquiring the original data table dataset and newly added data from the original classification model. Furthermore, it increases the dimension of the classification model by selecting effective fields, and then updates the original classification model using the updated original data table dataset. This improves the accuracy of the classification model calculation and enables iterative updates of the classification model without manual intervention, thereby saving labor costs.

[0084] The user behavior classification model update device provided in this embodiment of the invention can execute the user behavior classification model update method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0085] Example 4

[0086] Figure 5 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0087] like Figure 5 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0088] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0089] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as a classification model update method for user behavior.

[0090] In some embodiments, a classification model update method for user behavior can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the classification model update method for user behavior described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform a classification model update method for user behavior by any other suitable means (e.g., by means of firmware).

[0091] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0092] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0093] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0094] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0095] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0096] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0097] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0098] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for updating a classification model based on user behavior, characterized in that, include: Obtain the original user table dataset and newly added data corresponding to the original classification model, wherein the original table dataset contains the original data; Obtain a new table dataset based on the original table dataset and the newly added data; Determine the validity of the new table dataset. If the new table dataset is determined to be valid, obtain the valid fields in the new table dataset and update the original table dataset based on the valid fields. The original classification model is updated using the updated original dataset. The step of obtaining the new table dataset based on the original table dataset and the newly added data includes: Identify the new fields in the newly added data; The missing value is processed in the newly added field to obtain the first processed field; The first processing field is subjected to extreme value processing or normalization processing to obtain the second processing field; The new table dataset is obtained by adding the newly added data corresponding to the second processing field to the original table dataset, wherein the time span of the new table dataset is consistent with that of the original table dataset.

2. The method according to claim 1, characterized in that, The step of determining the new field in the new data includes: Retrieve all fields from the newly added data; Filter all fields to obtain structured fields, and use these structured fields as the new fields.

3. The method according to claim 1, characterized in that, The process of handling missing values ​​in the newly added field to obtain the first processed field includes: Each of the newly added fields will be used as the target field in turn; Determine whether the ratio of the missing data amount to the total data amount of the new data corresponding to the target field is greater than a first preset threshold. If so, the target field is used as the first processing field; otherwise, the target field is deleted.

4. The method according to claim 1, characterized in that, The step of performing extreme value processing or normalization processing on the first processing field to obtain the second processing field includes: Arrange the newly added data corresponding to the first processing field from smallest to largest to obtain the first processing set, and obtain the difference or ratio between two specified positions in the first processing set; Determine whether the difference or ratio is less than a second preset threshold. If so, perform extreme value processing on the first processing field to obtain a second processing field; otherwise, perform normalization processing on the first processing field to obtain a second processing field.

5. The method according to claim 1, characterized in that, The determination of the validity of the new table dataset includes: Obtain relevant data metrics for the original table dataset and the new table dataset, wherein the relevant data metrics include precision, recall, and area under the receiver operation feature curve; Calculate the difference between each of the same related data indicators in the original table dataset and the new table dataset; Determine whether the difference meets the preset conditions. If it does, determine that the new table dataset is valid; otherwise, determine that the new table dataset is invalid.

6. The method according to claim 5, characterized in that, The process of obtaining relevant data metrics for the original table dataset and the new table dataset includes: The original table dataset and the new table dataset are divided into a first dataset and a second dataset, respectively, and the actual labels of the first dataset and the second dataset input by the user are obtained. The original classification model is trained using the first dataset and the actual labels of the first dataset to obtain a trained classification model; The trained classification model is used to calculate the calculated labels on the second dataset, and the relevant data indicators of the actual labels and the calculated labels are calculated.

7. The method according to claim 6, characterized in that, The step of obtaining valid fields from the new table dataset when it is determined that the new table dataset is valid, and updating the original table dataset based on the valid fields, includes: Obtain the contribution of each field in the new table dataset and the original table dataset; Determine the minimum contribution of a field in the original table dataset, and then select the fields in the new table dataset whose contribution is greater than the minimum contribution as the valid fields. The original table dataset is updated by adding the valid fields and the corresponding new data to the original table dataset.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor to enable the at least one processor to perform the method of claims 1-7.

9. A computer storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the method described in claims 1-7.

Images