Method and device for quickly authenticating new channel user and electronic equipment

Abstract

The application provides a method and device for quickly authenticating users of a newly added channel and a computing device. The method comprises the following steps: selecting a base channel from existing channels, obtaining user sample data of the base channel, and performing clustering calculation to obtain a plurality of class clusters of a user group of the base channel; obtaining user sample data of a newly added channel, wherein the amount of the user sample data of the newly added channel is less than the amount of the user sample data of the base channel; filtering the user sample data of the base channel according to the similarity between the user sample data of the newly added channel and the class clusters, so as to expand a training data set of a user authentication model of the newly added channel; training the user authentication model of the newly added channel by using the expanded training data, and authenticating new users of the newly added channel by using the model. The application effectively solves the problem of insufficient user sample data of the newly added channel by filtering user sample data from the user sample data of the base channel to expand the user sample data of the newly added channel.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of computer information processing, and in particular, to a method and device for quickly authenticating users of a newly added channel, and a computer device. BACKGROUND

[0002] Information push is a new technology for reducing information overload on the Internet by regularly transmitting information that users need. Push technology reduces the time spent searching on the network by automatically transmitting information to users. It searches and filters information according to the interests of users and regularly pushes it to users to help them efficiently discover valuable information. Network information push is the use of push technology to actively send information that users are interested in to users on a regular basis according to their needs. For example, relevant information is pushed to users through channels such as short messages.

[0003] However, sending corresponding push information to different users is actually a multi-classification problem, and accordingly, a multi-classification model is needed to perform corresponding processing. In fact, classification problems are basic problems of machine learning, and regression, clustering, and the like correspond thereto. However, with the continuous development of Internet service businesses, new channels will be continuously added, and relevant information will be pushed to users through the new channels. Because the amount of user sample data of the new channel is small, far less than the amount of user sample data of existing channels, it is impossible to use machine learning methods to analyze the user sample data of the new channel, and it is impossible to achieve customized information push for different users of the new channel, thereby leading to inaccurate push information, low push efficiency, and the like. In addition, there is still much room for improvement in model calculation accuracy and authentication accuracy.

[0004] Therefore, it is necessary to provide an improved authentication method for quickly authenticating users of a new channel. SUMMARY

[0005] To solve the technical problems in the prior art that the amount of sample data of a new channel is insufficient, the amount of sample data of the new channel is insufficient, which leads to the inability to use machine learning methods to analyze the user sample data of the new channel, the inability to achieve customized information push for different users of the new channel, and the problems of inaccurate push information and low push efficiency.

[0006] The first aspect of the present application provides a method for quickly authenticating users of a newly added channel, wherein the newly added channel is a channel added on the basis of an existing channel, comprising: selecting a base channel from the existing channels, obtaining user sample data of the base channel, and performing clustering calculation to obtain a plurality of class clusters of a user group of the base channel; obtaining user sample data of the newly added channel, wherein the amount of user sample data of the newly added channel is less than that of the base channel; filtering the user sample data of the base channel according to the similarity between the user sample data of the newly added channel and the class clusters, so as to expand the training data set of a user authentication model of the newly added channel; training the user authentication model of the newly added channel using the expanded training data set, and using the model to authenticate new users of the newly added channel.

[0007] According to an optional embodiment of the present application, at least part of the features of the user sample data of the newly added channel and the user sample data of the base channel are the same, and the same features are used for the clustering calculation and the similarity calculation.

[0008] According to an optional embodiment of the present application, further comprising: performing clustering calculation on the user sample data of the newly added channel to obtain at least one class cluster of a user group of the newly added channel; and the similarity calculation is a similarity calculation between the class cluster of the user group of the newly added channel and the class cluster of the user group of the base channel.

[0009] According to an optional embodiment of the present application, further comprising: taking the clustering center of the class cluster of the user group of the newly added channel as a new clustering center, taking the clustering center of the class cluster of the user group of the base channel as a base clustering center, and sequentially calculating the Euclidean distance between all new clustering centers and each base clustering center; when the calculated Euclidean distance is greater than a certain threshold, judging that the user group corresponding to the corresponding new clustering center is similar to the user group corresponding to the base clustering center, and adding the user sample data of the user group corresponding to the base clustering center to the user sample data of the newly added channel as extended sample data.

[0010] According to an optional embodiment of the present application, further comprising: determining common features in the user sample data of the base channel and the user sample data of the newly added channel, and filtering user sample data from the user sample data of the base channel according to the determined common features, wherein the common features include at least one of the following features: user basic features, multi-lending risk performance features, and transaction behavior data.

[0011] According to an optional embodiment of the present application, comprising: using a K-means algorithm to perform clustering calculation on the user sample data of the base channel; and / or using a K-means algorithm to perform clustering calculation on the user sample data of the newly added channel.

[0012] According to an optional embodiment of the present application, the obtaining of the user sample data of the new channel comprises: obtaining the user sample data from one or more applet interfaces.

[0013] According to an optional embodiment of the present application, the training of the new channel user authentication model using the augmented training data set and the authentication of the new user of the new channel using the model comprises: training the new channel user authentication model using the augmented training data set, wherein the training data comprises a new channel type, transaction behavior data corresponding to the new channel type, user basic features, multi-investment risk performance features, and user authentication evaluation values; obtaining user feature data of the new user of the new channel, and calculating a user authentication evaluation value of the new user using the trained new channel user authentication model.

[0014] The second aspect of the present application provides a device for quickly authenticating a user of a new channel, wherein the new channel is a channel newly added on the basis of an existing channel, and the device comprises: a clustering calculation module configured to select a base channel from the existing channel, obtain user sample data of the base channel, and perform clustering calculation to obtain a plurality of class clusters of a user group of the base channel; a data acquisition module configured to acquire user sample data of the new channel, wherein the amount of the user sample data of the new channel is less than the amount of the user sample data of the base channel; a screening processing module configured to screen the user sample data of the base channel according to the similarity between the user sample data of the new channel and the class clusters, so as to augment a training data set of a new channel user authentication model; and an authentication processing module configured to train the new channel user authentication model using the augmented training data set, and to authenticate a new user of the new channel using the model.

[0015] According to an optional embodiment of the present application, at least part of the user sample data of the new channel and the user sample data of the base channel are the same, and the same features are used for the clustering calculation and the similarity calculation.

[0016] According to an optional embodiment of the present application, the device further comprises: a clustering calculation module configured to perform clustering calculation on the user sample data of the new channel to obtain at least one class cluster of a user group of the new channel; and the similarity calculation is a similarity calculation between the class cluster of the user group of the new channel and the class cluster of the user group of the base channel.

[0017] According to an optional embodiment of the present application, the method further comprises a calculation and judgment module, which takes the cluster centers of the user groups of the new channel as new cluster centers, takes the cluster centers of the user groups of the base channel as base cluster centers, and sequentially calculates the Euclidean distances between each new cluster center and each base cluster center; when the calculated Euclidean distance is greater than a specific threshold, it is judged that the user group corresponding to the corresponding new cluster center is similar to the user group corresponding to the base cluster center, and the user sample data of the user group corresponding to the base cluster center is added to the user sample data of the new channel as expansion sample data.

[0018] According to an optional embodiment of the present application, the method further comprises a determination module, which is configured to determine common features in the user sample data of the base channel and the user sample data of the new channel, and filter user sample data from the user sample data of the base channel according to the determined common features, wherein the common features include at least one of the following features: user basic features, multi-investment risk performance features, and transaction behavior data.

[0019] According to an optional embodiment of the present application, the method further comprises: using a K-means algorithm to perform cluster calculation on the user sample data of the base channel; and / or using a K-means algorithm to perform cluster calculation on the user sample data of the new channel.

[0020] According to an optional embodiment of the present application, the method further comprises: using a K-means algorithm to perform cluster calculation on the user sample data of the base channel; and / or using a K-means algorithm to perform cluster calculation on the user sample data of the new channel.

[0021] According to an optional embodiment of the present application, the method further comprises: using the expanded training data set to train the user authentication model of the new channel, wherein the training data includes the new channel type, the transaction behavior data corresponding to the new channel type, the user basic features, the multi-investment risk performance features, and the user authentication evaluation value; obtaining the user feature data of a new user of the new channel, and using the trained user authentication model of the new channel to calculate the user authentication evaluation value of the new user.

[0022] In addition, the third aspect of the present application also provides a computer device comprising a processor and a memory, wherein the memory is configured to store a computer executable program, and when the computer executable program is executed by the processor, the processor executes the method for quickly authenticating users of a new channel according to the first aspect of the present application.

[0023] In addition, the fourth aspect of the present application also provides a computer program product, which stores a computer executable program, and when the computer executable program is executed, the method for quickly authenticating users of a new channel according to the first aspect of the present application is implemented.

[0024] Advantages

[0025] Compared with the prior art, the application can make the user sample data of the new channel reach a specified amount by screening the user sample data from the user sample data of the base channel to expand the user sample data of the new channel, can effectively solve the problem of insufficient user sample data of the new channel, can effectively perform new user authentication processing by using the expanded user sample data to establish a training data set, training the new channel user authentication model using the training data set, and using the model to authenticate new users of the new channel, can accurately predict target users who can pass user authentication, and can effectively implement customized information pushing for different target users.

[0026] Further, the K-means algorithm is used to perform clustering calculation on the user sample data of the base channel, so that multiple class clusters corresponding to the base channel and class center values of each class cluster can be accurately obtained, the K-means algorithm is used to perform clustering calculation on the user sample data of the new channel, so that multiple class clusters corresponding to the new channel and class center values of each class cluster can be accurately obtained, the Euclidean distance between the user sample data of the new channel and the class center of the class cluster of the base channel is used to accurately screen the user sample data of the base channel to expand the user sample data of the new channel, so that the problem of insufficient user sample data of the new channel can be effectively solved, the model parameters in the new channel user authentication model can be adjusted to optimize the model parameters and improve the model authentication accuracy, the new channel user authentication model is trained using the expanded training data set, and the model is used to authenticate new users of the new channel, so that new user authentication processing can be more effectively performed, target users who can pass user authentication can be more accurately predicted, and customized information pushing for different target users can be more effectively implemented. BRIEF DESCRIPTION OF DRAWINGS

[0027] In order to make the technical problems solved by the application, the technical means adopted and the technical effects achieved more clear, the specific embodiments of the application will be described in detail below with reference to the accompanying drawings. However, it should be declared that the drawings described below are only the drawings of the exemplary embodiments of the application, and for those skilled in the art, other drawings of embodiments can be obtained from these drawings without creative labor.

[0028] Figure 1 is a flowchart of an example of the method for quickly authenticating users of a new channel of embodiment 1 of the application.

[0029] Figure 2 is a flowchart of another example of the method for quickly authenticating users of a new channel of embodiment 1 of the application.

[0030] Figure 3 is a flowchart of still another example of the method for quickly authenticating a user of a new channel according to Embodiment 1 of the present application.

[0031] Figure 4 is a schematic diagram of an example of the new channel authentication apparatus according to Embodiment 2 of the present application.

[0032] Figure 5 is a schematic diagram of another example of the new channel authentication apparatus according to Embodiment 2 of the present application.

[0033] Figure 6 is a schematic diagram of still another example of the new channel authentication apparatus according to Embodiment 2 of the present application.

[0034] Figure 7 is a structural block diagram of an exemplary embodiment of a computer device according to the present application.

[0035] Figure 8 is a structural block diagram of an exemplary embodiment of a computer program product according to the present application. DETAILED DESCRIPTION

[0036] Exemplary embodiments of the present application will now be described more fully hereinafter with reference to the accompanying drawings. The exemplary embodiments of the present application may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these exemplary embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the application to those skilled in the art. Like reference numerals refer to like elements throughout the specification.

[0037] Features, structures, characteristics or other details described in relation to one specific embodiment are not excluded from other embodiments unless explicitly excluded.

[0038] In the description of the specific embodiments, features, structures, characteristics or other details described in relation to the present application are to make those skilled in the art fully understand the embodiments. However, it is not excluded that one or more of the specific features, structures, characteristics or other details can not be practiced by those skilled in the art without the technical solution of the present application.

[0039] The flowcharts shown in the drawings are only exemplary illustrations, and do not necessarily include all contents and operations / steps, nor necessarily executed in the order described. For example, some operations / steps can be further decomposed, and some operations / steps can be combined or partially combined, so the actual execution order can be changed according to the actual situation.

[0040] The block diagrams shown in the drawings are merely functional entities and do not necessarily have to correspond to physically independent entities. That is, the functional entities can be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0041] It should be understood that although first, second, third, etc. designations can be used herein to describe various devices, elements, components or parts, this should not be limited by these designations. These designations are used to distinguish one from another. For example, a first device can also be referred to as a second device without departing from the essential technical solutions of the present application.

[0042] The term "and / or" or "and / or" includes all combinations of any one and one or more of the associated listed items.

[0043] In view of the above problems, the present application proposes a method for quickly authenticating users of a new channel. The method screens user sample data from the user sample data of the base channel to expand the user sample data of the new channel, so that the amount of user sample data of the new channel can reach a specified amount, effectively solving the problem of insufficient user sample data of the new channel. By using the expanded user sample data to establish a training data set and using the training data set to train a new channel user authentication model, and then using the model to authenticate new users of the new channel, the new user authentication process can be effectively processed, the target users who can pass the user authentication can be accurately predicted, and the customized information push for different target users can be effectively implemented. The specific process will be described in detail below.

[0044] It should be noted that in the present application, resources refer to any available material, information, time, information resources include computing resources and various types of data resources. Data resources include various special-purpose data in various fields. In essence, the present application can be applied to the authentication process of various Internet services, not limited to Internet resources, including physical goods, water, electricity, and meaningful materials, etc. However, for the sake of convenience, the authentication process for resource guarantee services is used as an example to illustrate the implementation of the method of the present application, but those skilled in the art should understand that the present application can also be used for other resource service authentication. The specific process of the method of the present application will be described in detail below in conjunction with examples.

[0045] Example 1

[0046] Below, the embodiments of the method of the present application will be described with reference to Figures 1 to 3 the flowchart of the method of the present application.

[0047] Figure 1 The flowchart of the method of the present application for quickly authenticating users of a new channel. As shown inFigure 1 The method comprises the following steps.

[0048] In step S101, a base channel is selected from existing channels, user sample data of the base channel is obtained, and clustering calculation is performed to obtain multiple class clusters of a user group of the base channel.

[0049] In step S102, user sample data of the new channel is obtained, and the amount of the user sample data of the new channel is less than the amount of the user sample data of the base channel.

[0050] In step S103, the user sample data of the base channel is filtered according to the similarity between the user sample data of the new channel and the class clusters, so as to expand a training data set of a new channel user authentication model.

[0051] In step S104, the new channel user authentication model is trained using the expanded training data set, and the model is used to authenticate new users of the new channel.

[0052] First, in step S101, a base channel is selected from existing channels, user sample data of the base channel is obtained, and clustering calculation is performed to obtain multiple class clusters of a user group of the base channel.

[0053] In this example, the authentication method is used to quickly authenticate users of a new channel, and the new channel is a channel added on the basis of an existing channel.

[0054] It should be noted that, in the present application, the existing channel refers to an existing channel for pushing various Internet services, wherein the Internet services include resource use services, resource guarantee services or mutual assistance services, group purchase, and ride services, etc. For example, the existing channel includes a webpage, a social tool, a short message, an email, a resource-related product APP, and a resource management platform, etc.

[0055] Specifically, a base channel is selected from existing channels, user sample data of the base channel is obtained, and clustering calculation is performed to obtain multiple class clusters of a user group of the base channel, and each class cluster represents a user group.

[0056] It should be noted that the base channel refers to an existing channel whose use time for pushing various Internet services reaches a specified use time, and / or the number of sample data reaches a specific number. For example, the base channel is a resource service product APP, a guarantee product APP, and a resource allocation product APP, etc.

[0057] As a specific embodiment, the base channels are selected from the existing channels according to the channel usage time and the proportion of the user sample data generated by the channel in the total sample data of a certain Internet service, and the number of the base channels is one or more, and in this example, the number of the base channels is one.

[0058] For example, the usage time of the base channel is greater than a certain time (for example, 6 months to 24 months), and the user sample data is greater than a certain amount. In this case, the base channel is determined according to the set proportion of the user sample data of the new channel to the user sample data of the base channel, wherein the set proportion corresponding to different Internet service or resource service products is configured, for example, the set proportion is 1:10-1:20, and the user sample data of the new channel is, for example, about 10,000, and the user sample data of the base channel is determined to be 100,000-200,000 according to the set proportion and the user sample data of the new channel, and the base channel is selected according to the determined user sample data.

[0059] It should be noted that the above is only described as an example and cannot be understood as a limitation of the present application. In other examples, the base channel can also be determined according to the channel usage time or the proportion of the user sample data generated by the channel in the total sample data of a certain Internet service, and the base channel is selected from the existing channels.

[0060] Further, the user sample data of the selected base channel is obtained and clustering calculation is performed.

[0061] For clustering calculation, for example, the K-means algorithm is used, and the user basic features and transaction behavior data are used to perform clustering calculation on the user sample data of the base channel to obtain a plurality of class clusters (also referred to as class set in this specification) corresponding to the base channel.

[0062] It should be noted that in other examples, at least one of the following features can also be used for clustering calculation: user basic features, multi-investment risk performance features, transaction behavior data, promotion information browsing or clicking times, and purchase times of similar service products. The above is only described as an example and cannot be understood as a limitation of the present application.

[0063] Specifically, the center vector of each class cluster is calculated using a centroid-based algorithm. Specifically, the sample data set is divided into k class clusters, and the value of k is selected by a technician according to the type of Internet service and the sample quantity. The specific process of the above algorithm is as follows.

[0064] First, the center vectors μ1, μ2, …, μk of the k classes are initialized. k .

[0065] Allocation phase. Determine the class cluster each sample belongs to according to the current class center value:

[0066] Loop, for each sample x i Calculate the distance of the sample from each class center μ j : D ij =||x i -μ j ||, assign the sample to the nearest class cluster, end the loop.

[0067] Update phase. Update the class center of each class cluster:

[0068] Loop, for each class cluster, update the class center of each class cluster according to the allocation scheme in the last step, which is the mean of all samples in the class cluster: End the loop.

[0069] Thus, by the above method, the clustering process is performed to obtain a plurality of class clusters corresponding to the base channel and the class center value of each class cluster.

[0070] Therefore, by using the K-means algorithm, the clustering calculation is performed on the user sample data of the base channel, and the plurality of class clusters corresponding to the base channel and the class center value of each class cluster can be accurately obtained.

[0071] In order to continuously expand the Internet service business, new channels will be continuously added as the business develops, which will be used for information push of related Internet service business. However, due to the small amount of user sample data of the newly added channel, which is much less than the amount of user sample data of the existing channel, for example, less than 10% of the amount of user sample data of the existing channel, or even less, so that the machine learning method cannot be used to analyze the user sample data of the newly added channel, and the customized information push for different users of the newly added channel cannot be realized. Therefore, it is necessary to expand the user sample data of the newly added channel, so as to be able to use the machine learning method to analyze the user sample data, and then realize the customized information push for different users.

[0072] In step S102, the user sample data of the newly added channel is obtained, and the amount of user sample data of the newly added channel is less than the amount of user sample data of the base channel.

[0073] As a specific embodiment, the user sample data of the newly added channel corresponding to the resource guarantee service is obtained, and the amount of user sample data of the newly added channel is determined, wherein the newly added channel is, for example, an application programming interface (API).

[0074] Specifically, the user sample data is obtained from one or more application programming interfaces (APIs), and the categories of the application programs include diet, transportation, shopping, etc.

[0075] Further, the user sample data amount of the new channel is determined, for example, the user sample data amount is 10,000.

[0076] It should be noted that the user sample data amount of the new channel is less than the user sample data amount of the base channel, specifically, less than 10% of the user sample data amount (for example, 150,000) of the selected base channel. Not limited to this, the above is only described as an example, and cannot be understood as a limitation of the present application.

[0077] In an embodiment, the user sample data of the new channel and the user sample data of the base channel have at least part of the same features, and the user sample data of the new channel and the user sample data of the base channel are determined, and the same features are used for respective clustering calculation.

[0078] In this embodiment, the authentication method further includes the following step S201 (that is, step S102 is divided into steps S102 and S201, see Figure 2 ): using K-means algorithm and using user basic features and transaction behavior data, the user sample data of the new channel is clustered and calculated. By executing step S201, at least one class cluster of the user group corresponding to the new channel is obtained, and in this example, the new channel includes multiple class clusters.

[0079] Specifically, the center vector of each class cluster of the new channel is calculated using the algorithm based on the centroid.

[0080] It should be noted that in this example, since the clustering calculation method in step S102 is the same as the clustering calculation method in step S101, the description of this part is omitted. In addition, at least one of the following features can be used for clustering calculation: user basic features, multi-lending risk performance features, transaction behavior data, promotion information browsing or clicking times, and purchase times of similar service products. The above is only described as an example, and cannot be understood as a limitation of the present application.

[0081] Therefore, by using K-means algorithm, the user sample data of the new channel is clustered and calculated, and multiple class clusters corresponding to the new channel and the class center value of each class cluster can be accurately obtained.

[0082] In order to solve the problem of insufficient user sample data of the new channel, the present application selects the user sample data similar or similar to the user sample data of the new channel (for example, API channel) from the user sample data of the existing mature channel (that is, existing channel) of the APP, and adds it to the training sample of the user risk model based on the API channel (that is, to the training data set), to solve the problem of insufficient user sample of the API channel.

[0083] Next, the expansion process of the user sample data of the new channel will be described in detail.

[0084] In step S103, the user sample data of the base channel is filtered according to the similarity between the user sample data of the new channel and the clusters, so as to expand the training data set of the new channel user authentication model.

[0085] In an embodiment, the user sample data of the base channel is filtered according to the similarity between the user sample data of the new channel and the respective cluster centers (also referred to as base cluster centers in the present application) of the plurality of clusters (each cluster representing a user group) of the base channel.

[0086] Specifically, the similarity between the user sample data of the new channel and each base cluster center is calculated, and the base cluster center with a similarity to the user sample data within a certain range is selected from the base cluster centers. For example, the Euclidean distance is used for similarity calculation. However, it is not limited thereto, and in other examples, Mahalanobis distance, Manhattan distance, etc. can also be used for similarity calculation.

[0087] More specifically, the Euclidean distance between the user sample data of the new channel and the base cluster center is calculated, and the base cluster center with a calculated Euclidean distance within a certain range is selected, which is one or more base cluster centers.

[0088] Specifically, when the selected base cluster center is a plurality of base cluster centers (or one base cluster center), the user sample data of the cluster (i.e. user group) corresponding to the plurality of base cluster centers (or the one base cluster center) is selected from the user sample data of the base channel as user sample data similar to the user sample data of the new channel, so as to expand the user sample data of the new channel.

[0089] In a preferred embodiment, the user sample data of the base channel is filtered according to the similarity between the respective new cluster centers of the plurality of clusters corresponding to the new channel determined in step S102 and the respective cluster centers (i.e. base cluster centers) of the plurality of clusters of the base channel.

[0090] It should be noted that the similarity calculation is the similarity calculation between the cluster of the user group of the new channel and the cluster of the user group of the base channel. The cluster center of the cluster of the user group of the new channel is the new cluster center, and the cluster center of the cluster of the user group of the base channel is the base cluster center.

[0091] Specifically, similarity calculation is performed between each new cluster center and the base cluster center. For example, the Euclidean distance is used for similarity calculation. However, this is not limited thereto, and in other examples, the Mahalanobis distance, Manhattan distance, or the like can be used for similarity calculation.

[0092] More specifically, the Euclidean distance between all new cluster centers and each base cluster center is calculated in turn, and when the calculated Euclidean distance (for example, the Euclidean distance d1 between the new cluster center X1 and the base cluster center J1) is greater than a certain threshold value d 阈 , it is determined that the user group corresponding to the corresponding new cluster center (for example, the new cluster center X1) is similar to the user group corresponding to the base cluster center (for example, the base cluster center J1), and the user sample data of the user group corresponding to the base cluster center (for example, the base cluster center J1) is added to the user sample data of the new channel as extended sample data.

[0093] In yet another embodiment, common features are determined in the user sample data of the base channel and the user sample data of the new channel, and user sample data is filtered from the user sample data of the base channel according to the determined common features, wherein the common features include at least one of the following features: user basic features, multi-lending risk performance features, and transaction behavior data.

[0094] Further, by expanding the user sample data of the new channel from the user sample data of the base channel by the above method, the amount of user sample data of the new channel can reach a specified amount, and can be used to establish a training data set of a user authentication model.

[0095] It should be noted that the above is only described as an example and should not be construed as limiting the present application. In other examples, user sample data is filtered from the user sample data of the base channel by calculating similarity and by common features.

[0096] Therefore, by similarity between the user sample data of the new channel and the clusters of the base channel, the user sample data of the base channel can be accurately filtered to expand the training data set of the user authentication model of the new channel.

[0097] Next, in step S104, the new channel user authentication model is trained using the expanded training data set, and the model is used to authenticate new users of the new channel.

[0098] As a specific embodiment, the following algorithm is used to establish a new channel user authentication model.

[0099] The user re-authentication model is constructed using the XGBoost algorithm, the TextCNN algorithm, the random forest algorithm, the logistic regression algorithm, the RNN or LSTM network algorithm.

[0100] In this step, as an embodiment, an RNN or a logistic regression algorithm is used to establish a new channel user authentication model, and the user re-authentication model is trained using a training data set, which includes a new channel type, historical user basic features annotated with user authentication evaluation values in the new channel, transaction behavior data and multi-investment risk performance features corresponding to the new channel type, and extended sample data selected from the base channel, which includes user feature data of historical users annotated with user authentication evaluation values, and uses the user authentication pass probability to represent the user authentication evaluation value.

[0101] Specifically, the user feature data is the input feature, and the output feature is the user authentication evaluation value.

[0102] For the training data samples of the training data set in model construction, the following steps are also included: a step of judging whether the true probability and the predicted probability of the historical user are consistent.

[0103] Specifically, the relative deviation of the true probability (i.e., the true pass probability) and the output predicted probability (i.e., the predicted pass probability) is calculated, wherein when the calculated relative deviation is less than a set value, it is judged that the true probability and the predicted probability are consistent; and when the calculated relative deviation value is greater than or equal to the set value, it is judged that the true probability and the predicted probability are consistent. In this way, it can be more accurately judged whether the true probability and the predicted probability are consistent.

[0104] It should be noted that the above is only described as an example and cannot be understood as a limitation of the present application. In other embodiments, the difference between the true probability and the output predicted probability can also be calculated to judge whether the true probability and the predicted probability are consistent. In the present example, the set value is determined by business personnel according to the statistical value of the historical true probability. However, it is not limited thereto, and in other examples, it can also be determined by other means.

[0105] As an embodiment, after the step of judging whether the true probability and the predicted probability are consistent, it is judged whether the proportion of the number of the predicted probabilities consistent with the true probability in the number of all the training data samples is greater than a preset proportion value.

[0106] If yes, the new channel user authentication model converges, and a trained new channel user authentication model is obtained.

[0107] If no, the model parameters in the new channel user authentication model are adjusted, and the predicted probability of the historical user is predicted again through the new channel user authentication model until the proportion of the number of the predicted probabilities consistent with the true probability in the number of all the training data samples is greater than the preset proportion value.

[0108] It should be noted that, for the step of determining whether the proportion of the number of the predicted probabilities consistent with the true probabilities in the number of all the training data samples is greater than a preset proportion value, etc., the machine learning method used to build the model can be replaced, and a new channel user authentication model can be re-established, etc. The above is only described as an example and cannot be understood as a limitation of the present application.

[0109] Thus, by adjusting the model parameters in the new channel user authentication model, the model parameters can be optimized, and the model authentication accuracy can be improved.

[0110] In an embodiment, user feature data of a new user of a new channel is obtained, and a trained new channel user authentication model is used to calculate a user authentication evaluation value of the new user.

[0111] For example, in the user authentication process of an internet service such as a resource allocation service or a resource guarantee service, when the calculated user authentication evaluation value is greater than a certain value, it is determined that the new user has passed the user re-authentication, and it is determined that the new user can obtain, for example, a resource allocation service or a resource guarantee service.

[0112] In another specific embodiment, when the calculated user authentication evaluation value is less than or equal to a certain value, it is determined that the new user has not passed the user authentication, and it is determined that the new user cannot obtain, for example, a resource allocation service or a resource guarantee service.

[0113] Therefore, by training the new channel user authentication model using the augmented training data set and using the model to authenticate new users of the new channel, new user authentication processing can be more effectively performed, target users who can pass user authentication can be more accurately predicted, and customized information pushing for different target users can be more effectively implemented.

[0114] It should be noted that the above is only described as an example and cannot be understood as a limitation of the present application.

[0115] The process of the above authentication method is only used to illustrate the present application, and the order and number of steps are not particularly limited. In addition, the steps in the above method can also be split into two (see steps S103 and S301 in detail), three, or some steps can be combined into one step, which is adjusted according to the actual example. Figure 3 Step S103 is split into step S103 and step S301), three, or some steps can be combined into one step, which is adjusted according to the actual example.

[0116] Those skilled in the art can understand that all or part of the steps of the above embodiments are implemented as a program (computer program) executed by a computer data processing device. When the computer program is executed, the above method provided by the present application can be implemented. Moreover, the computer program can be stored in a computer readable storage medium, which can be a readable storage medium such as a disk, an optical disk, a ROM, a RAM, etc., or a storage array composed of multiple storage media, such as a disk or tape storage array. The storage medium is not limited to centralized storage, and it can also be distributed storage, such as cloud storage based on cloud computing.

[0117] Compared with the prior art, the present application can make the user sample data of the new added channel reach a specified amount by screening the user sample data from the user sample data of the base channel to expand the user sample data of the new added channel, which can effectively solve the problem of insufficient user sample data of the new added channel; by using the expanded user sample data to establish a training data set and using the training data set to train the new added channel user authentication model, and then using the model to authenticate new users of the new added channel, new user authentication processing can be effectively performed, target users who can pass user authentication can be accurately predicted, and customized information pushing for different target users can be effectively implemented.

[0118] Further, by using the K-means algorithm to perform clustering calculation on the user sample data of the base channel, multiple class clusters corresponding to the base channel and class center values of each class cluster can be accurately obtained; by using the K-means algorithm to perform clustering calculation on the user sample data of the new added channel, multiple class clusters corresponding to the new added channel and class center values of each class cluster can be accurately obtained; by using the Euclidean distance between the user sample data of the new added channel and the class center of the class cluster of the base channel, the user sample data of the base channel can be accurately screened to expand the user sample data of the new added channel, which can effectively solve the problem of insufficient user sample data of the new added channel; by adjusting the model parameters in the new added channel user authentication model, the model parameters can be optimized, and the model authentication accuracy can be improved; by using the expanded training data set to train the new added channel user authentication model, and using the model to authenticate new users of the new added channel, new user authentication processing can be more effectively performed, target users who can pass user authentication can be more accurately predicted, and customized information pushing for different target users can be more effectively implemented.

[0119] Embodiment 2

[0120] The device embodiment of the present application is described below, which can be used to execute the method embodiment of the present application. For the details described in the device embodiment of the present application, it should be considered as a supplement to the above-mentioned method embodiment; for the details not disclosed in the device embodiment of the present application, it can be implemented by referring to the above-mentioned method embodiment.

[0121] With reference to Figure 4 , Figure 5 and Figure 6 , the second aspect of the present application provides a new channel authentication device 400 for quickly authenticating users of a new channel, wherein the new channel is a channel newly added on the basis of an existing channel, and the new channel authentication device 400 comprises: a clustering calculation module 401 for selecting a base channel in the existing channel, obtaining user sample data of the base channel, and performing clustering calculation to obtain a plurality of class clusters of a user group of the base channel; a data acquisition module 402 for acquiring user sample data of the new channel, wherein the amount of user sample data of the new channel is less than the amount of user sample data of the base channel; a screening processing module 403 for screening the user sample data of the base channel according to the similarity between the user sample data of the new channel and the class clusters, so as to expand a training data set of a new channel user authentication model; and an authentication processing module 404 for training the new channel user authentication model using the expanded training data set, and authenticating new users of the new channel using the model.

[0122] Specifically, at least part of the features of the user sample data of the new channel and the user sample data of the base channel are the same, and the same features are used for the clustering calculation and the similarity calculation.

[0123] Further, the user sample data of the new channel is subjected to clustering calculation to obtain at least one class cluster of a user group of the new channel; and the similarity calculation is the similarity calculation between the class cluster of the user group of the new channel and the class cluster of the user group of the base channel.

[0124] As shown in Figure 5 , the new channel authentication device 400 further comprises a calculation and judgment module 501 for taking the clustering center of the class cluster of the user group of the new channel as a new clustering center, taking the clustering center of the class cluster of the user group of the base channel as a base clustering center, and sequentially calculating the Euclidean distance between all new clustering centers and each base clustering center; when the calculated Euclidean distance is greater than a specific threshold value, it is judged that the user group corresponding to the corresponding new clustering center is similar to the user group corresponding to the base clustering center, and the user sample data of the user group corresponding to the base clustering center is added to the user sample data of the new channel as extended sample data.

[0125] As shown in Figure 6As shown, the newly-added channel authentication device 400 further comprises a determination module 601, configured to determine common features in the user sample data of the base channel and the user sample data of the newly-added channel, and filter user sample data from the user sample data of the base channel according to the determined common features, wherein the common features comprise at least one of the following features: user basic features, multi-lending risk performance features, and transaction behavior data.

[0126] For the clustering calculation of the base channel, a K-means algorithm is used to perform clustering calculation on the user sample data of the base channel.

[0127] For the clustering calculation of the newly-added channel, a K-means algorithm is used to perform clustering calculation on the user sample data of the newly-added channel.

[0128] In an embodiment, the user sample data of the base channel is filtered according to the similarity between the user sample data of the newly-added channel and each class center (also referred to as a base clustering center in the present application) of the plurality of class clusters (each class cluster representing a user group) of the base channel.

[0129] Specifically, the similarity between the user sample data of the newly-added channel and each base clustering center is calculated, and the base clustering center having a similarity to the user sample data within a certain range is selected from the base clustering centers. For example, the Euclidean distance is used for similarity calculation. However, it is not limited thereto, and in other examples, the Mahalanobis distance, Manhattan distance, etc. can also be used for similarity calculation.

[0130] More specifically, the Euclidean distance between the user sample data of the newly-added channel and the base clustering center is calculated, and the base clustering center having a calculated Euclidean distance within a certain range is selected, which is one or more base clustering centers.

[0131] Specifically, when the selected base clustering center is a plurality of base clustering centers (or one base clustering center), the user sample data of the class cluster (i.e., user group) corresponding to the plurality of base clustering centers (or the one base clustering center) is selected as user sample data similar to the user sample data of the newly-added channel from the user sample data of the base channel, for expanding the user sample data of the newly-added channel.

[0132] In a preferred embodiment, the user sample data of the base channel is filtered according to the similarity between each new clustering center of the plurality of class clusters of the newly-added channel and each class center (i.e., base clustering center) of the plurality of class clusters of the base channel.

[0133] It should be noted that the similarity calculation is the similarity calculation of the clusters of the user group of the new channel and the clusters of the user group of the base channel. The cluster center of the clusters of the user group of the new channel is a new cluster center, and the cluster center of the clusters of the user group of the base channel is a base cluster center.

[0134] Specifically, the similarity calculation is performed between each new cluster center and the base cluster center. For example, the Euclidean distance is used for the similarity calculation. However, the method is not limited to this, and in other examples, the Mahalanobis distance, Manhattan distance, or the like can also be used for the similarity calculation.

[0135] It should be noted that the above is described only as an example and cannot be understood as a limitation of the present application.

[0136] As a specific embodiment, obtaining sample data from the new channel includes: obtaining user sample data from one or more application program interfaces (APIs).

[0137] Specifically, the new channel user authentication model is trained using the augmented training data set, and the training data includes a new channel type, transaction behavior data corresponding to the new channel type, user basic features, multi-investment risk performance features, and a user authentication evaluation value.

[0138] Further, the user feature data of a new user of the new channel is obtained, and the trained new channel user authentication model is used to calculate a user authentication evaluation value of the new user.

[0139] Further, it is determined whether the new user passes the user authentication according to the calculated user authentication evaluation value, so as to determine whether the new user can obtain, for example, a resource allocation service or a resource guarantee service.

[0140] In a specific embodiment, when the calculated user authentication evaluation value is greater than a specific value, it is determined that the new user passes the user re-authentication, and it is determined that the new user can obtain, for example, a resource allocation service or a resource guarantee service.

[0141] In another specific embodiment, when the calculated user authentication evaluation value is less than or equal to a specific value, it is determined that the new user does not pass the user authentication, and it is determined that the new user cannot obtain, for example, a resource allocation service or a resource guarantee service.

[0142] It should be noted that in Embodiment 2, the description of the same part as Embodiment 1 is omitted.

[0143] Those skilled in the art can understand that each module in the above device embodiment can be distributed in the device as described, or can be changed accordingly and distributed in one or more devices different from the above embodiment. The modules of the above embodiment can be combined into one module, or can be further split into multiple sub-modules.

[0144] Compared with the prior art, the present application can make the user sample data of the new added channel reach a specified amount by screening the user sample data from the user sample data of the base channel to expand the user sample data of the new added channel, can effectively solve the problem of insufficient user sample data of the new added channel; can effectively perform new user authentication processing, can accurately predict target users who can pass user authentication, and can effectively implement customized information pushing for different target users by using the expanded user sample data to establish a training data set, and using the training data set to train the new added channel user authentication model, and using the model to authenticate new users of the new added channel.

[0145] Further, the K-means algorithm is used to perform clustering calculation on the user sample data of the base channel, so that multiple class clusters corresponding to the base channel and class center values of each class cluster can be accurately obtained; the K-means algorithm is used to perform clustering calculation on the user sample data of the new added channel, so that multiple class clusters corresponding to the new added channel and class center values of each class cluster can be accurately obtained; the Euclidean distance between the user sample data of the new added channel and the class center of the class cluster of the base channel is used to accurately screen the user sample data of the base channel to expand the user sample data of the new added channel, so that the problem of insufficient user sample data of the new added channel can be effectively solved; by adjusting the model parameters in the new added channel user authentication model, the model parameters can be optimized, and the model authentication accuracy can be improved; by using the expanded training data set to train the new added channel user authentication model, and using the model to authenticate new users of the new added channel, new user authentication processing can be more effectively performed, target users who can pass user authentication can be more accurately predicted, and customized information pushing for different target users can be more effectively implemented.

[0146] Embodiment 3

[0147] The computer device embodiment of the present application is described below, which can be regarded as a specific entity implementation of the above-mentioned method and system embodiments of the present application. The details described in the computer device embodiment of the present application should be regarded as a supplement to the above-mentioned method or system embodiments; for details not disclosed in the computer device embodiment of the present application, reference can be made to the above-mentioned method or system embodiments.

[0148] Figure 7 is a structural block diagram of an exemplary embodiment of a computer device according to the present application. The following refers toFigure 7 The computer device 200 according to this embodiment of the present application is described. Figure 7 The computer device 200 shown is only one example of a possible device and should not be taken as limiting the present application to just that particular device.

[0149] As shown, the computer device 200 is in the form of a general-purpose computing device. The components of the computer device 200 can include, but are not limited to, at least one processing unit 210, at least one storage unit 220, a bus 230 that connects the various device components (including the storage unit 220 and the processing unit 210), a display unit 240, etc. Figure 7 The storage unit 220 can include a readable medium in the form of volatile storage such as random access memory (RAM) 2201 and / or cache memory 2202, and also can include non-volatile storage such as read-only memory (ROM) 2203.

[0150] Figure 1 The storage unit 220 can include a readable medium in the form of volatile storage such as random access memory (RAM) 2201 and / or cache memory 2202, and also can include non-volatile storage such as read-only memory (ROM) 2203.

[0151] The storage unit 220 can include a readable medium in the form of volatile storage such as random access memory (RAM) 2201 and / or cache memory 2202, and also can include non-volatile storage such as read-only memory (ROM) 2203.

[0152] The storage unit 220 can include a readable medium in the form of volatile storage such as random access memory (RAM) 2201 and / or cache memory 2202, and also can include non-volatile storage such as read-only memory (ROM) 2203.

[0153] The bus 230 can represent one or more of several types of bus structures, including a storage bus or bus controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of a variety of bus architectures.

[0154] ​Computer device 200 can also communicate with one or more external devices 300 such as a keyboard or pointing device, a Bluetooth device, etc.; one or more devices that enable a user to interact with computer device 200; and / or one or more devices (e.g., a router, a modem, a server, etc.) that enable computer device 200 to communicate with one or more other computing devices. Such communication can occur via Input / Output (I / O) interface 250. Still yet, computer device 200 can communicate with one or more networks, such as a local area network (LAN), a general wide area network (WAN), and / or a public network (e.g., the Internet) via network adapter 260. As depicted, network adapter 260 can communicate with the other components of computer device 200 via bus 230. It should be appreciated that although not shown, other hardware and / or software modules could be used in conjunction with computer device 200. Examples, include, but are not limited to, microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data archival storage systems, etc.

[0155] Those skilled in the art will readily understand that the exemplary embodiments described with regard to the present application can be implemented by software, or by software in combination with necessary hardware. Thus, the technical solutions according to the embodiments of the present application can be embodied in the form of a software product, which can be stored in a computer readable storage medium (which can be a CD-ROM, a U disk, a mobile hard disk, etc.) or a network, and includes a number of instructions to enable a computing device (which can be a personal computer, a server, or a network device, etc.) to perform the above-described methods according to the present application. When the computer program is executed by a data processing device, the computer program product enables the above-described methods of the present application to be implemented.

[0156] As shown in Figure 7 the computer program can be stored on one or more computer program products. The computer program product can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or apparatus, or any suitable combination of the above. More specific examples (a non-exhaustive list) of the computer program product include an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0157] The computer program product can include a data signal transporting one or more instructions that, when executed by a processor, can cause the processor to implement a method described herein. The computer program product can include a data signal transporting a computer program, which can cause a processor to implement a method described herein. Additionally, the computer program product can include a computer readable medium having stored thereon a computer program, which can cause a processor to implement a method described herein.

[0158] Program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++, or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computing device, partly on the user's computing device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device or entirely on the remote computing device or server. In the latter scenario, the remote computing device can be connected to the user's computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computing device, such as through the Internet using an Internet Service Provider (ISP).

[0159] In light of the above, the present application can be implemented in hardware, or as software modules running on one or more processors, or as a combination of both. Those skilled in the art will appreciate that the various components in the embodiments of the present application can be implemented using a general purpose data processing apparatus such as a microprocessor or a digital signal processor (DSP) to execute software instructions that implement the methods described herein. The present application can also be implemented as a computer program product, which can be executed by a computer program and computer program product (e.g., a computer program and computer program product) for performing some or all of the methods described herein. Such computer program product can be stored on a computer program product, or can have one or more signals in the form of a computer readable medium. Such signals can be downloaded from an Internet website, or provided on a carrier signal, or in any other form.

[0160] The above-described specific embodiments further illustrate the purpose, technical solutions and beneficial effects of the present application, and it should be understood that the present application is not inherently related to any specific computer, virtual device or electronic equipment, and various general-purpose devices can also implement the present application. The above-described is only a specific embodiment of the present application and is not used to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for quickly authenticating users of a newly added channel, characterized in that: The new channel is a channel newly added on the basis of an existing channel, and the method comprises: selecting one or more base channels from the existing channels according to the channel use time and / or the proportion of the user sample data generated by the channel in the total sample data of an internet service, obtaining the user sample data of the base channels, and performing clustering calculation to obtain multiple clusters of user groups of the base channels; each cluster represents a user group; obtaining user sample data of the new channel, the number of users of the new channel being less than 10% of the user sample data amount of the selected base channels, performing clustering calculation on the user sample data of the new channel to obtain at least one cluster of user groups of the new channel; the clustering calculation uses a centroid-based algorithm to calculate the center vector of each cluster; the user sample data of the new channel and the user sample data of the base channels have at least some same features, and the same features are used for the clustering calculation and the similarity calculation; taking the clustering center of the cluster of the user group of the new channel as a new clustering center, and taking the clustering center of the cluster of the user group of the base channel as a base clustering center, and sequentially calculating the Euclidean distance between all new clustering centers and each base clustering center; when the calculated Euclidean distance is greater than a specific threshold, judging that the user group corresponding to the corresponding new clustering center is similar to the user group corresponding to the base clustering center, and adding the user sample data of the user group corresponding to the base clustering center to the user sample data of the new channel as expanded sample data to expand the training data set of the new channel user authentication model; training the new channel user authentication model using the expanded training data set, and using the model to authenticate new users of the new channel.

2. The method for quickly authenticating users of a new channel according to claim 1, characterized in that: the obtaining of the user sample data of the new channel comprises: obtaining the user sample data from one or more applet interfaces. 3.The method of claim 1, wherein, Further comprising: determining common features in the user sample data of the base channels and the user sample data of the new channel, and screening user sample data from the user sample data of the base channels according to the determined common features, wherein the common features include at least one of the following features: user basic features, multi-investment risk performance features, and transaction behavior data.

4. The method for quickly authenticating users of a new channel according to claim 1, comprising: using a K-means algorithm to perform clustering calculation on the user sample data of the base channels; and / or using a K-means algorithm to perform clustering calculation on the user sample data of the new channel.

5. The method for quickly authenticating a user of a new channel according to claim 1, wherein, The training of the new channel user authentication model using the expanded training data set, and the authentication of new users of the new channel using the model comprise: training the new channel user authentication model using the expanded training data set, wherein the training data comprises a new channel type, transaction behavior data corresponding to the new channel type, user basic features, multi-investment risk performance features, and a user authentication evaluation value; Obtain user feature data of a new user of a new channel, and use a trained user authentication model of the new channel to calculate a user authentication evaluation value of the new user.

6. A device for quick authentication of a user of a new channel, comprising: a first authentication module for authenticating the user of the new channel; and a second authentication module for authenticating the user of the new channel. The new channel is a channel added on the basis of an existing channel, and the device comprises: A clustering calculation module selects one or more base channels from the existing channels according to channel use time and / or a proportion of user sample data generated by the channel in total sample data of an internet service, obtains user sample data of the base channels, and performs clustering calculation to obtain a plurality of class clusters of user groups of the base channels; each class cluster represents a user group; A data acquisition module is configured to obtain user sample data of the new channel, the number of users of the new channel is less than 10% of the amount of user sample data of the selected base channels, perform clustering calculation on the user sample data of the new channel to obtain at least one class cluster of user groups of the new channel, use a centroid-based algorithm to calculate a center vector of each class cluster, and use at least some same features of the user sample data of the new channel and the user sample data of the base channels for the clustering calculation and the similarity calculation; A calculation and judgment module is configured to take a clustering center of a class cluster of user groups of the new channel as a new clustering center, take a clustering center of a class cluster of user groups of the base channels as a base clustering center, sequentially calculate Euclidean distances between all new clustering centers and each base clustering center, judge that a user group corresponding to a corresponding new clustering center is similar to a user group corresponding to the base clustering center when the calculated Euclidean distance is greater than a specific threshold, and add user sample data of the user group corresponding to the base clustering center to user sample data of the new channel as expanded sample data to expand a training data set of a user authentication model of the new channel. An authentication processing module is configured to train the user authentication model of the new channel using the expanded training data set, and use the model to authenticate new users of the new channel. 7.A computer device comprising a processor and a memory, the memory being configured to store a computer executable program, when the computer program is executed by the processor, the processor performs the method for quickly authenticating users of a new channel according to any one of claims 1-5.

Images