WO2020257991A1

WO2020257991A1 - User identification method and related product

Info

Publication number: WO2020257991A1
Application number: PCT/CN2019/092592
Authority: WO
Inventors: 石露
Original assignee: 深圳市欢太科技有限公司; Oppo广东移动通信有限公司
Priority date: 2019-06-24
Filing date: 2019-06-24
Publication date: 2020-12-30
Also published as: CN113383362B; CN113383362A

Abstract

Disclosed in embodiments of the present application are a user identification method and a related product. The method comprises: when user identification on a target user ID is required, determining whether there are N fake traffic creating populations that have been identified, the N fake traffic creating populations being classified according to population user rules, and fake traffic creating user IDs comprised in any one of the N fake traffic creating populations being greater than a preset number threshold; if yes, obtaining input features of the target user ID, the input features comprising a user location feature, a user APP usage feature, a user equipment usage feature, and a user click through rate (CTR) feature; identifying a similarity between the target user ID and each of the N fake traffic creating populations on the basis of the input features of the target user ID; and if there is a fake traffic creating population in the N fake traffic creating populations whose similarity to the target user ID is greater than a preset similarity threshold, determining that the target user ID is a fake traffic creating user ID. The embodiments of the present application can improve the identification accuracy of fake traffic creating users.

Description

User identification methods and related products

Technical field

This application relates to the field of communication technology, and specifically to a user identification method and related products.

Background technique

On the resource display platform, the more useful resources are pushed to users in important positions, the greater the user's perception of value and the better the effect of resources. Currently, as the display resources become more and more limited, the indicator of the resource display in a better position is getting better and better as its user clicks or user behavior. In order to get more click-through rates, some content producers will use the method of brushing to obtain them. On the one hand, they can obtain better location resources for themselves, and on the other hand, they can get more exposure from real users. But from the perspective of the resource display platform, it will cause the platform to be unfair to resources, and users will also have mistrust of the platform. Therefore, how to identify the users who brushed the amount has become an urgent problem to be solved.

The current user identification of swiping users is mainly to identify users one by one. For example, one account is logged in on multiple mobile phones, and there are multiple accounts on one mobile phone for registration and login, and one mobile phone has continuous and uninterrupted access to the same URL or the number of visits exceeds normal. Identification of users and other means. Currently, the accuracy of user identification for swiping is low.

Summary of the invention

The embodiments of the present application provide a user identification method and related products, which can improve the identification accuracy of a user who has swiped.

In the first aspect, an embodiment of the present application provides a user identification method, including:

When user identification of the target user ID is required, determine whether there are identified N brush groups, the N brush groups are classified according to group user rules, any one of the N brush groups The ID of the users who brush the amount included in the group is greater than the preset number threshold, and N is a positive integer;

If it exists, obtain the input characteristics of the target user ID, the input characteristics including user location characteristics, user APP usage characteristics, user equipment usage characteristics, and user click-through rate CTR characteristics;

Identifying the similarity between the target user ID and each of the N brushing groups based on the input characteristics of the target user ID;

If there is a brushing group whose similarity with the target user ID is greater than a preset similarity threshold among the N brushing groups, it is determined that the target user ID is a brushing user ID.

In a second aspect, an embodiment of the present application provides a user identification device, the user identification device including a first determining unit, an acquiring unit, an identifying unit, and a second determining unit, wherein:

The first determining unit is configured to determine whether there are identified N brushing groups when the target user ID needs to be identified. The N brushing groups are classified according to group user rules. The user ID contained in any one of the two groups is greater than the preset number threshold, and N is a positive integer;

The acquiring unit is configured to acquire the input characteristics of the target user ID when the first determining unit determines that there are N swiping groups that have been identified, and the input characteristics include user location characteristics and user APP usage Features, user equipment usage features and user click-through rate CTR features;

The identification unit is configured to identify the similarity between the target user ID and each of the N brush amount groups based on the input characteristics of the target user ID;

The second determining unit is configured to determine the brush group that has a similarity with the target user ID greater than a preset similarity threshold among the N brush groups The target user ID is the ID of the user who swipes.

In a third aspect, an embodiment of the present application provides a server, including a processor and a memory, the memory is used to store one or more programs, and the one or more programs are configured to be executed by the processor. The program includes instructions for executing the steps in the first aspect of the embodiments of the present application.

In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the foregoing computer-readable storage medium stores a computer program for electronic data exchange, wherein the foregoing computer program enables a computer to execute Some or all of the steps described in one aspect.

In a fifth aspect, embodiments of the present application provide a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute Example part or all of the steps described in the first aspect. The computer program product may be a software installation package.

It can be seen that the user identification method described in the embodiment of the present application specifically includes the following steps: when the target user ID needs to be identified, it is determined whether there are identified N brushing groups, and the N brushing groups are According to the group user rule classification, any one of the N brushing groups contains a brushing user ID greater than a preset number threshold, and N is a positive integer; if it exists, the input characteristics of the target user ID are obtained, and the input characteristics include users Location characteristics, user APP usage characteristics, user equipment usage characteristics, and user click-through rate CTR characteristics; based on the input characteristics of the target user ID to identify the similarity between the target user ID and each of the N brush groups; if N There is a scraping group whose similarity with the target user ID is greater than a preset similarity threshold in the scraping group, and the target user ID is determined as the scraping user ID. In the implementation of the embodiments of this application, when the target user ID is used for user identification, the target user can be recognized for similarity with the identified swiping group. If the similarity is greater than the preset similarity threshold, the target user ID can be directly identified In order to swipe the user ID, since the swiping users often have the characteristics of the swiping group, the similarity recognition with the swiping group can quickly and accurately determine whether the target user is the swiping user ID, thereby improving the recognition of the swiping user Accuracy.

Description of the drawings

In order to more clearly describe the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only These are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work.

FIG. 1 is a schematic flowchart of a user identification method disclosed in an embodiment of the present application;

2 is a schematic flowchart of another user identification method disclosed in an embodiment of the present application;

FIG. 3 is a schematic flow chart of an algorithm for identifying users who swipe credits disclosed in an embodiment of the present application;

4 is a schematic flowchart of another user identification method disclosed in an embodiment of the present application;

Fig. 5 is a schematic structural diagram of a user identification device disclosed in an embodiment of the present application;

Fig. 6 is a schematic structural diagram of a server disclosed in an embodiment of the present application.

Detailed ways

In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only It is a part of the embodiments of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific sequence. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes unlisted steps or units, or optionally also includes Other steps or units inherent to these processes, methods, products or equipment.

Reference to "embodiments" herein means that a specific feature, structure or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art clearly and implicitly understand that the embodiments described herein can be combined with other embodiments.

The mobile terminals involved in the embodiments of this application may include various handheld devices with wireless communication functions, vehicle-mounted devices, wearable devices, computing devices or other processing devices connected to wireless modems, as well as various forms of user equipment (User Equipment, UE), mobile station (Mobile Station, MS), terminal device (terminal device), etc. For ease of description, the devices mentioned above are collectively referred to as mobile terminals.

The following describes the embodiments of the present application in detail.

Please refer to FIG. 1. FIG. 1 is a schematic flowchart of a user identification method disclosed in an embodiment of the present application. As shown in FIG. 1, the user identification method includes the following steps.

101. When the target user ID needs to be identified, the server determines whether there are identified N scraping groups. The N scraping groups are classified according to the group user rules, and any one of the N scraping groups is obtained. The number of brush user IDs included in the group is greater than the preset number threshold, and N is a positive integer.

In the embodiments of the present application, the server serves the client, and the content of the service includes providing resources to the client and storing client data. The server is a targeted service program, and the device running the server can be called a server. The server can establish connections with multiple clients at the same time, and can provide services to multiple clients at the same time. In the embodiment of the application, the server can be used to identify the user ID of the swiping amount.

Among them, the client, content provider, and server can form a content distribution system. The client is a content distribution client. The client can provide a display interface for displaying various content resources. Different content resources occupy different positions of the display interface. The content distribution system will count the clicks or clicks of each content resource of each client. The amount of downloads is determined to be displayed in different positions of the display interface of the client according to the amount of clicks or downloads of each content resource. For content resources that can be provided by the content provider, the server displays the content of the content provider on the display interface of the client. There can be multiple content providers, multiple clients, and multiple servers. The content resources can be application resources, audio and video resources, etc. The following is an example of APP resources.

The content distribution system usually counts the number of clicks or downloads of various apps, and displays the apps in different locations of the content distribution platform (ie, the client) based on the statistical data, recommends apps with high downloads to users, and gives Dedicated resources form a list to operate. Based on this, the APP producer (ie, the content provider) hopes that the APP can obtain a higher amount of clicks or downloads, so that the APP can get a better display position or can be recommended to users by the content distribution platform. In order to obtain a higher number of clicks or downloads, APP publishers may use a swiping application to swipe the clicks or downloads of the APP. The APP publisher sends a swipe task request through the swipe application, and the terminal installed with the swipe application obtains the swipe task request. Then, the terminal uses the installed swipe application to generate users who do not really exist, that is, the swipe The user clicks or watches the APP that needs to be swiped, thereby increasing the number of clicks or downloads of the APP. When making certain decisions with such unreal APP clicks or downloads, it often brings a lot of adverse effects to the content distribution platform. The apps recommended based on the unreal APP clicks or download traffic data may not be high-quality apps. Thereby affecting the user's trust in the content distribution platform. In order to reduce the negative impact of unreal APP clicks or downloads, the content distribution platform needs to identify which of the users who click or watch a certain APP are users who are scam users.

In order to identify whether the target user ID is the user ID of the scalping user, the server first determines whether there are N scalping groups that have been identified. The N scalping groups are classified according to the group user rules. Group user rules can be based on the location of the device corresponding to the user ID, the time series used by the application corresponding to the user ID, the cumulative use time of the application corresponding to the user ID, the frequency of use of the application corresponding to the user ID, and the application corresponding to the user ID. The usage time ratio of all applications with the client is determined. For example, the device location can be the same, the time series are similar, the cumulative usage time is greater than a certain duration threshold (for example, 2 hours), the usage frequency is greater than a certain frequency threshold (for example, 100 times), the application corresponding to the user ID and all of the client The user IDs whose usage time ratio of the application is greater than a certain ratio threshold (for example, 80%) are classified into the same swiping group.

Wherein, the brush amount user ID included in the brush amount group is greater than the preset number threshold, and the preset number threshold can be set in advance and stored in the memory (for example, non-volatile memory) of the server. The preset number threshold may be an integer greater than or equal to 2, for example, the preset number threshold may be set to 5.

Optionally, the group user rule is determined based on the location of the device corresponding to the user ID and the time sequence used by the application corresponding to the user ID. Before performing step 101, you can also perform the following steps:

The server determines that the distance between the corresponding device positions in the plurality of user IDs that have been identified is less than the preset distance threshold, and the time sequence used by the application in the plurality of user IDs is in the first preset time period. The user ID of the amount of brushing within is classified into the first type of brushing group.

In this embodiment of the application, before the server determines whether there are identified N swiping groups, the server can use group user rules to classify the multiple swiping user IDs that have been identified, and the multiple swiping groups that have been identified can be classified. The distance between the corresponding device positions in the user ID is less than the preset distance threshold, and the time series used by the application in the plurality of user IDs in the first preset time period are classified into the same user ID Class brush amount group.

The time series used by the application is the data used by the APP that substitutes the time label, that is, a time series label is recorded for each operation of the APP, which is used to record the operation time of the APP. Because group users will concentrate on a certain period of time, the APP time training of users of the same group has a high degree of similarity. The embodiment of the present application can classify the user ID of the user according to the distance between the corresponding device position in the user ID of the user and the similarity of the time series used by the application in the user ID of the user, thereby improving the classification of the user ID of the user. Accuracy.

102. If there are N swiping groups that have been identified, the server obtains the input characteristics of the target user ID. The input characteristics include user location characteristics, user APP usage characteristics, user equipment usage characteristics, and user click-through rate CTR characteristics.

In the embodiment of the present application, the server acquiring the input feature of the target user ID may specifically be: the server extracting the input feature of the target user ID from the historical behavior data of the target user ID.

The historical behavior data of the target user ID may include the location information of the device logged in by the target user ID within a preset time period, the APP usage information of the target user ID, the usage information of the device logged in by the target user ID, and the target user ID CTR characteristics.

Among them, taking into account the location aggregation type of the users who swipe and the preference and less mobility of some locations, the embodiment of the present application adds the location characteristics of the user as one of the considerations. When the user uses the device to swipe, the simulation will be similar to the terminal operation, but because of the swipe task, the open rate and use time of the swipe task content will be longer, but the use of other apps will be shorter. Based on this situation, this embodiment of the application will examine the frequency and duration of use of commonly used APPs by the terminal and the time distribution of the entire terminal using APPs, so the user APP use characteristics are added as one of the considerations. At the same time, because the user swipes with a specific purpose, the operation behavior of the terminal will also be different, such as whether there is a call history, whether a card is inserted, whether there is a terminal use behavior such as short message reception, so the user terminal use characteristics are added as One of the considerations. Since the indicator of the final success of the brush volume is the exposure click rate or download rate or the success rate of a certain behavior, the click volume of tasks related to CTR will be higher and more significant than other users, so the user CTR feature is also used as One of the considerations.

The user location feature includes the location feature of the device where the target user ID is logged in (including the location of the device when the user ID is logged in, the range of change in the location of the device, etc.). Generally speaking, when the user ID is logged in, the smaller the change in the device's position, the greater the possibility that the user ID is a swiping user.

The user APP usage characteristics include the usage time of the target APP logged in by the user ID, the usage frequency of the target APP, and the usage time distribution of the target APP. Generally speaking, the longer the use time of the target APP logged in by the user ID, the higher the use frequency of the target APP, and the more concentrated the use time distribution of the target APP, the greater the probability that the user ID is a swiping user.

The user equipment usage characteristics include the usage characteristics of the device logged in by the target user ID (for example, whether the device has a call record, whether a card is inserted, whether there is a short message reception, etc. during the login process of the target user ID). Generally speaking, if the device has no call history, no card inserted, and no short message reception during the login process of the target user ID, the greater the possibility that the user ID is a swiping user.

CTR refers to searching after entering keywords in a search engine, and then sorting out relevant web pages in order according to factors such as bidding, and then users will choose the websites they are interested in and click into them; the total number of searches for a website is taken as the total The number of times, the ratio of the number of times a user clicks and enters the website to the total number of times is called click-through rate. Generally speaking, if the CTR of the target user ID is higher, the possibility that the user ID is a swiping user is greater.

103. The server recognizes the similarity between the target user ID and each of the N brush groups based on the input characteristics of the target user ID.

In the embodiments of the present application, each of the N brushing groups will have group-common characteristics. The common characteristics of the group include similar group positions and similar time series used by group applications.

The server can calculate the similarity between the user location feature of the target user ID and the location feature of each of the N brush groups, and calculate the time series used by the application of the target user ID and the N brush groups The time similarity of the time series used by the group application of each brush group in the group; according to the location feature similarity of the group location characteristics of each brush group in the N brush groups and each brush group in the N brush groups The time similarity of the time series used by the group application of the quantity group determines the similarity between the target user ID and each of the N quantity groups.

104. If there is a scraping group whose similarity with the target user ID is greater than a preset similarity threshold among the N scraping groups, the server determines the target user ID as the scraping user ID.

In the embodiment of this application, if there is a brush group whose similarity to the target user ID is greater than the preset similarity threshold in the N brush amount groups, it indicates that the target user ID belongs to the N brush amount groups with the greatest similarity to the target user ID. For the target swiping group, the target user ID is classified into the target swiping group, and the target user ID is determined as the swiping user ID.

In the embodiment of the present application, when the target user ID is identified, the target user can be identified with the identified swiping group. If the similarity is greater than the preset similarity threshold, the target user ID can be directly identified In order to swipe the user ID, since the swiping users often have the characteristics of the swiping group, the similarity recognition with the swiping group can quickly and accurately determine whether the target user is the swiping user ID, thereby improving the recognition of the swiping user Accuracy.

Please refer to FIG. 2, which is a schematic flowchart of another user identification method disclosed in an embodiment of the present application. Fig. 2 is obtained by further optimization on the basis of Fig. 1. As shown in Fig. 2, the user identification method includes the following steps.

201. When the target user ID needs to be identified, the server determines whether there are identified N scraping groups. The N scraping groups are classified according to group user rules, and any one of the N scraping groups is scraped. The number of brush user IDs included in the group is greater than the preset number threshold, and N is a positive integer.

202. If there are N swiping groups that have been identified, the server obtains the input characteristics of the target user ID. The input characteristics include user location characteristics, user APP usage characteristics, user equipment usage characteristics, and user click-through rate CTR characteristics.

203. The server identifies the similarity between the target user ID and each of the N brush groups based on the input characteristics of the target user ID.

204. If there is a scraping group whose similarity with the target user ID is greater than a preset similarity threshold among the N scraping groups, the server determines that the target user ID is the scraping user ID.

For the specific implementation of step 201 to step 204 in the embodiment of the present application, reference may be made to the description of step 101 to step 104 shown in FIG. 1, which will not be repeated here.

205. If there is no brush group whose similarity with the target user ID is greater than the preset similarity threshold among the N brush groups, the server inputs the input characteristics of the target user ID into the trained binary classification model to obtain the target user ID Enter the preliminary classification result of the feature.

206. The server inputs the preliminary classification results into the trained classifier for calculation to obtain intermediate calculation results, and inputs the intermediate calculation results into the trained neural network model for training, to obtain the identification result of the target user ID.

In the embodiment of this application, if there is no brush group whose similarity to the target user ID is greater than the preset similarity threshold among the N brush groups, it indicates that the target user ID does not belong to any of the N brush groups. It is necessary to use the trained binary classification model, the trained classifier and the trained neural network model to identify the target user ID.

Among them, the two-classification model can adopt a multi-algorithm fusion method. For example, the two-classification model can specifically include k-Nearest Neighbor (KNN) classification algorithm, logistic regression (LR) algorithm, and support vector machine (Support Vector (Machine, SVM) algorithm of one or more combinations of two classification models.

The classifier may include an extreme gradient boosting (eXtreme Gradient Boosting, XGboost) classifier or a random forest classifier.

For example, please refer to FIG. 3, which is a schematic flow chart of an algorithm for recognizing a swiping user disclosed in an embodiment of the present application. As shown in Figure 3, the input features of the target user are first input into the two classifiers. The KNN classification algorithm, LR algorithm, and SVM algorithm in the two classifiers are single algorithms, which are used to classify the input features of the target user; The intermediate results of the classifier classification are input to the classifier. The XGboost and random forest in the classifier are fusion algorithms used for preliminary calculation of the intermediate results output by the two classifiers; then the intermediate results of the classifier classification are input to the neural network model After training, the recognition result of the target user is finally obtained. There are only two types of recognition results for target users, that is, whether it is a scalping user or not a scalping user.

The identification process of the target user ID in the embodiments of this application successively adopts a single algorithm, a fusion algorithm, and a neural network. A single algorithm can preliminarily classify input features and reduce the computational complexity of subsequent fusion algorithms. The fusion algorithm takes into account the number of users who brush This possibility can ensure the accuracy of the calculation results of the fusion algorithm. Finally, the neural network model is used for training to reduce the possibility of misjudgment, thereby improving the accuracy of the recognition result of the target user ID.

Optionally, before step 205 is performed, the following steps may be performed:

(11) The server extracts the input feature of the first user ID, the first user ID is any one of M user IDs to be identified, and M is a positive integer;

(12) The server uses the single-user rule to identify the ID of the user who is credited and the ID of the user that is not of the M user IDs to be identified, and P is a positive integer less than or equal to M;

(13) The server inputs the input features of the M to-be-identified user IDs into the initial binary classification model for training, and obtains M training results;

(14) When the accuracy of the M training results reaches a first preset accuracy threshold, the server determines that the initial two-classification model after training is a trained two-classification model.

Among them, the M user IDs to be identified can be identified by the single user rule. The M user IDs to be identified can all be used to identify whether they are credit users through a single user rule. Single user rules can include the following rules: (1) The same user ID can log in on multiple terminals (for example, mobile phones) in a short time; (2) There are multiple user IDs on one terminal for registration and login at the same time; (3) One terminal Continuous access to the same URL or the number of visits far exceeds that of ordinary users.

Each of the M user IDs to be identified either satisfies the above three single user rules at the same time, or does not satisfy the above single user rules. Among the M to-be-identified user IDs, the user ID that meets the above three single-user rules at the same time is the swiping user ID, and the user ID in the M to-be-identified user IDs that does not meet any of the three single-user rules is the non-swiping user ID . That is, the user IDs among the M user IDs to be identified can all be identified by the single user rule as to whether they are credit users. The user IDs of the users who want to be identified as the black samples of the two-classification model training, and the non-user IDs of the M user IDs to be identified are the white samples for the training of the two-class model to ensure the initial training of the two-class model The accuracy of the data improves the training effect of the two-class model. In order to improve the training effect of the two-class model, the value of M can be as large as possible.

This embodiment of the application provides a method for training a two-class model. First, a single-user rule is used to identify users who scribbled, and some more accurate users with scribbling are identified as black samples, and other normal users are used as white samples. As a two-classification problem, make predictions and count the accuracy of the prediction results. When the training results are wrong, the two-classification model will be adjusted accordingly so that the two-classification model will not have the same error next time. Until the accuracy of the two-classification model reaches the first preset accuracy threshold, the training is stopped, and the initial two-classification model after training is determined to be the trained two-classification model.

Optionally, before step 206 is performed, the following steps may be performed:

(21) The server inputs the M training results into the initial classifier for calculation, and obtains M intermediate calculation results;

(22) When the accuracy of the M intermediate calculation results reaches a second preset accuracy threshold, the server determines that the trained initial classifier is a trained classifier.

The embodiment of the present application provides a method for training a classifier. According to the previously identified users who are more accurate and use as black samples, and other normal users are trained as white samples, a classifier with higher accuracy can be obtained.

Optionally, before step 206 is performed, the following steps may be performed:

(31) The server inputs the M intermediate calculation results into the initial neural network model for training, and obtains M recognition results;

(32) When the accuracy of the M recognition results reaches the third preset accuracy threshold, the server determines that the trained initial neural network model is a trained neural network model.

The embodiment of the application provides a method for training a neural network model. According to the previously identified users with more accurate brushing as black samples and other normal users as white samples for training, a neural network with higher accuracy can be obtained. model.

Please refer to FIG. 4, which is a schematic flowchart of another user identification method disclosed in an embodiment of the present application. Figure 4 is further optimized on the basis of Figure 2. As shown in Figure 4, the user identification method includes the following steps.

401. When the target user ID needs to be identified, the server determines whether there are identified N scraping groups. The N scraping groups are classified according to group user rules, and any one of the N scraping groups is scraped. The number of brush user IDs included in the group is greater than the preset number threshold, and N is a positive integer.

402. If there are N swiping groups that have been identified, the server obtains the input characteristics of the target user ID. The input characteristics include user location characteristics, user APP usage characteristics, user equipment usage characteristics, and user click-through rate CTR characteristics.

403. The server recognizes the similarity between the target user ID and each of the N brush groups based on the input characteristics of the target user ID.

404. If there is a scraping group whose similarity with the target user ID is greater than a preset similarity threshold among the N scraping groups, the server determines that the target user ID is the scraping user ID.

405. If there is no brushing group whose similarity with the target user ID is greater than the preset similarity threshold among the N brushing groups, the server inputs the input characteristics of the target user ID into the trained binary classification model to obtain the target user ID Enter the preliminary classification result of the feature.

406. The server inputs the preliminary classification results into the trained classifier for calculation to obtain intermediate calculation results, and inputs the intermediate calculation results into the trained neural network model for training, to obtain the identification result of the target user ID.

Among them, the specific implementation of step 401 to step 406 can refer to step 201 to step 206 shown in FIG. 2, which will not be repeated here.

407. If there are no identified N swipe groups, the server determines whether there are multiple identified swipe user IDs.

408. If there are multiple credit user IDs that have been identified, the server identifies the similarity between the target user ID and the identified multiple credit user IDs.

409. If there is a swipe user ID whose similarity to the target user ID is greater than the preset similarity threshold among the multiple swipe user IDs, the server adds the swipe user-related feature to the input features of the target user ID; and step 405 is executed The middle server inputs the input features of the target user ID into the trained two-classification model to obtain the preliminary classification results of the input features of the target user ID.

In the embodiment of the present application, if there is no identified swiping group, the target user ID and a single identified swiping user ID can be calculated for similarity. When it is recognized that a user is associated with a single user who is credited, the similarity analysis algorithm can be used to judge the user, and the similarity between the target user ID and the credit user ID can be calculated to increase the input characteristics of the target user ID, thereby improving The accuracy of the identification of the target user ID further determines whether the target user is a real credit user.

Optionally, the embodiment of the present application may also use an unsupervised algorithm to complete group swipe identification, and use a clustering algorithm or a lonely forest algorithm to identify abnormal users in the group.

The foregoing mainly introduces the solution of the embodiment of the present application from the perspective of the execution process on the method side. It can be understood that, in order to realize the above-mentioned functions, the server includes hardware structures and/or software modules corresponding to each function. Those skilled in the art should easily realize that in combination with the units and algorithm steps of the examples described in the embodiments disclosed herein, the present invention can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed by hardware or computer software-driven hardware depends on the specific application and design constraint conditions of the technical solution. Professionals and technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered as going beyond the scope of the present invention.

The embodiment of the present application may divide the server side into functional units according to the foregoing method examples. For example, each functional unit may be divided corresponding to each function, or two or more functions may be integrated into one processing unit. The above-mentioned integrated unit can be implemented in the form of hardware or software functional unit. It should be noted that the division of units in the embodiments of the present application is illustrative, and is only a logical function division, and there may be other division methods in actual implementation.

Please refer to FIG. 5, which is a schematic structural diagram of a user identification device disclosed in an embodiment of the present application. As shown in FIG. 5, the user identification device 500 includes a first determination unit 501, an acquisition unit 502, an identification unit 503, and a second determination unit 504, wherein:

The first determining unit 501 is configured to determine whether there are identified N swipe groups when the target user ID needs to be identified. The N swipe groups are classified according to group user rules. Any one of the N brush groups contains brush user IDs greater than the preset number threshold, and N is a positive integer;

The acquiring unit 502 is configured to acquire the input characteristics of the target user ID when the first determining unit 501 determines that there are N swipe groups that have been identified, and the input characteristics include user location characteristics, user APP usage characteristics, user equipment usage characteristics, and user click-through rate CTR characteristics;

The identification unit 503 is configured to identify the similarity between the target user ID and each of the N brush amount groups based on the input characteristics of the target user ID;

The second determining unit 504 is configured to determine when the identifying unit 503 recognizes that there is a brushing group whose similarity with the target user ID is greater than a preset similarity threshold among the N brushing groups The target user ID is the ID of the user who swipes.

Optionally, the user identification device 500 may further include a processing unit 505505.

The processing unit 505 is configured to: if the identification unit 503 recognizes that there is no brush group whose similarity with the target user ID is greater than a preset similarity threshold among the N brush groups, The input feature of the target user ID is input into a trained binary classification model to obtain a preliminary classification result of the input feature of the target user ID;

The processing unit 505 is further configured to input the preliminary classification result into the trained classifier for calculation to obtain an intermediate calculation result, and input the intermediate calculation result into the trained neural network model for training to obtain the target user ID recognition result.

Optionally, the processing unit 505 is further configured to input the input feature of the target user ID into the trained binary classification model, and extract the first user ID before obtaining the preliminary classification result of the input feature of the target user ID The first user ID is any one of the M user IDs to be identified, and M is a positive integer; the single-user rule is used to identify the swipe user ID and the non-swipe amount among the M user IDs to be identified User ID; input the input features of the M user IDs to be identified into the initial binary classification model for training, and obtain M training results; when the accuracy of the M training results reaches the first preset accuracy threshold, determine The initial two-classification model after training is a trained two-classification model.

Optionally, the processing unit 505 is further configured to input the preliminary classification result into the trained classifier for calculation, and before obtaining the intermediate calculation result, input the M training results into the initial classifier for calculation to obtain M Intermediate calculation results; when the accuracy of the M intermediate calculation results reaches a second preset accuracy threshold, it is determined that the initial classifier after training is a trained classifier.

Optionally, the processing unit 505 is further configured to input the intermediate calculation results into the trained neural network model for training, and before the identification result of the target user ID is obtained, input the M intermediate calculation results into the initial The neural network model is trained to obtain M recognition results;

When the accuracy of the M recognition results reaches the third preset accuracy threshold, the initial neural network model after training is determined to be the trained neural network model.

Optionally, the group user rule is determined based on the location of the device corresponding to the user ID and the time series used by the application corresponding to the user ID, and the processing unit 505 is further configured to determine whether there is an existing user ID in the first determining unit 501. Before the identified N swiping groups, the distance between the corresponding device positions in the multiple swiping user IDs that have been identified is smaller than the preset distance threshold, and the time series of application usage in the multiple swiping user IDs The user IDs of the users who swiped during the first preset time period are classified into the first type of swipe groups.

Optionally, the processing unit 505 is further configured to determine whether there are multiple identified user IDs when the first determining unit 501 determines that there are no identified N crediting groups; if There is the plurality of scoring user IDs that have been identified, and the similarity between the target user ID and the plurality of scoring user IDs that have been identified; The target user ID similarity is greater than the preset similarity threshold for the swipe user ID, add swipe user-related features in the input features of the target user ID; input the target user ID input features into the trained two categories The model obtains the preliminary classification result of the input feature of the target user ID.

Wherein, the first determining unit 501, acquiring unit 502, identifying unit 503, second determining unit 504, and processing unit 505 in FIG. 5 may be processors.

Implementing the user identification device shown in Figure 5, when the target user ID is user identification, the target user can be identified with the identified brush group for similarity. If the similarity is greater than the preset similarity threshold, it can be directly identified The target user ID is the user ID of the scouring user. Since the scouring user often has the characteristics of the scouring group, the identification of the similarity with the scouring group can quickly and accurately determine whether the target user is the scouring user ID, thereby improving the scoring Measure the user’s recognition accuracy.

Please refer to FIG. 6. FIG. 6 is a schematic structural diagram of a server disclosed in an embodiment of the present application. As shown in FIG. 6, the server 600 includes a processor 601 and a memory 602. The server 600 may also include a bus 603. The processor 601 and the memory 602 may be connected to each other through the bus 603. The bus 603 may be a peripheral component. Connect the standard (Peripheral Component Interconnect, referred to as PCI) bus or extended industry standard architecture (Extended Industry Standard Architecture, referred to as EISA) bus, etc. The bus 603 can be divided into an address bus, a data bus, a control bus, and so on. For ease of representation, only one thick line is used in FIG. 6, but it does not mean that there is only one bus or one type of bus. The server 600 may further include an input communication interface 604, and the communication interface 604 may obtain data from an external device (for example, other servers or databases). The memory 602 is used to store one or more programs containing instructions; the processor 601 is used to call the instructions stored in the memory 602 to execute some or all of the method steps in FIGS. 1 to 4.

Implementing the server shown in Figure 6, when the target user ID is identified, the target user can be identified with the identified brush group for similarity. If the similarity is greater than the preset similarity threshold, it can be directly identified The target user ID is the user ID of the scalping user. Since the scalping user often has the characteristics of the scalping group, the similarity recognition with the scalping group can quickly and accurately determine whether the target user is the scalping user ID, thereby increasing the scalping volume The accuracy of user recognition.

An embodiment of the present application also provides a computer storage medium, wherein the computer storage medium stores a computer program for electronic data exchange, and the computer program causes the computer to execute any part of the user identification method described in the above method embodiment Or all steps.

The embodiments of the present application also provide a computer program product. The computer program product includes a non-transitory computer-readable storage medium storing a computer program. The computer program is operable to cause a computer to execute any of the methods described in the foregoing method embodiments. Part or all of the steps of a user identification method.

It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described sequence of actions. Because according to the present invention, certain steps can be performed in other order or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the involved actions and modules are not necessarily required by the present invention.

In the above-mentioned embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in an embodiment, reference may be made to related descriptions of other embodiments.

In the several embodiments provided in this application, it should be understood that the disclosed device may be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other divisions in actual implementation, for example, multiple units or components may be combined or may be Integrate into another system, or some features can be ignored or not implemented. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical or other forms.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.

In addition, the functional units in the various embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or software functional unit.

If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable memory. Based on this understanding, the technical solution of the present invention essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a memory, A number of instructions are included to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned memory includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other various media that can store program codes.

Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above-mentioned embodiments can be completed by instructing relevant hardware through a program. The program can be stored in a computer-readable memory, and the memory can include: flash disk , Read-only memory (English: Read-Only Memory, abbreviation: ROM), random access device (English: Random Access Memory, abbreviation: RAM), magnetic disk or optical disc, etc.

The embodiments of the present application are described in detail above, and specific examples are used in this article to illustrate the principles and implementation of the present invention. The descriptions of the above embodiments are only used to help understand the methods and core ideas of the present invention; Those of ordinary skill in the art, based on the idea of the present invention, will have changes in the specific implementation and the scope of application. In summary, the content of this specification should not be construed as limiting the present invention.

Claims

A user identification method, characterized in that it comprises:

When user identification of the target user ID is required, determine whether there are identified N brush groups, the N brush groups are classified according to group user rules, any one of the N brush groups The ID of the users who brush the amount included in the group is greater than the preset number threshold, and N is a positive integer;

If it exists, obtain the input characteristics of the target user ID, the input characteristics including user location characteristics, user APP usage characteristics, user equipment usage characteristics, and user click-through rate CTR characteristics;

Identifying the similarity between the target user ID and each of the N brushing groups based on the input characteristics of the target user ID;

If there is a brushing group whose similarity with the target user ID is greater than a preset similarity threshold among the N brushing groups, it is determined that the target user ID is a brushing user ID.
The method of claim 1, wherein the method further comprises:

If there is no brush group whose similarity with the target user ID is greater than the preset similarity threshold in the N brush groups, input the input characteristics of the target user ID into the trained binary classification model to obtain the The preliminary classification result of the input characteristics of the target user ID;

The preliminary classification result is input into a trained classifier for calculation to obtain an intermediate calculation result, and the intermediate calculation result is input into a trained neural network model for training to obtain the identification result of the target user ID.
The method according to claim 2, characterized in that, before the input feature of the target user ID is input into a trained binary classification model, and the preliminary classification result of the input feature of the target user ID is obtained, the method Also includes:

Extract the input feature of the first user ID, where the first user ID is any one of M user IDs to be identified, and M is a positive integer;

Using a single-user rule to identify the user ID of the amount-swiping user and the ID of the non-swiping user among the M user IDs to be identified;

Input the input features of the M to-be-identified user IDs into the initial binary classification model for training, and obtain M training results;

When the accuracy of the M training results reaches a first preset accuracy threshold, it is determined that the initial binary classification model after training is a trained binary classification model.
The method according to claim 3, characterized in that said inputting said preliminary classification result into a trained classifier for calculation and before obtaining intermediate calculation result, said method further comprises:

Input the M training results into the initial classifier for calculation to obtain M intermediate calculation results;

When the accuracy of the M intermediate calculation results reaches a second preset accuracy threshold, it is determined that the initial classifier after training is a trained classifier.
The method according to claim 4, characterized in that, before inputting the intermediate calculation result into a trained neural network model for training, and obtaining the identification result of the target user ID, the method further comprises:

Input the M intermediate calculation results into the initial neural network model for training, and obtain M recognition results;

When the accuracy of the M recognition results reaches the third preset accuracy threshold, the initial neural network model after training is determined to be the trained neural network model.
The method according to any one of claims 1 to 5, wherein the group user rule is determined based on the location of the device corresponding to the user ID and the time series used by the application corresponding to the user ID, and the determination of whether there is an identified Before the N brushing groups of, the method also includes:

The distance between the corresponding device positions in the identified multiple swiping user IDs is less than the preset distance threshold, and the time series used by the application in the multiple swiping user IDs is within the first preset time period. The user ID of the amount of brushing is classified into the first type of brushing group.
The method according to any one of claims 2 to 6, wherein the method further comprises:

If there are no identified N swipe groups, determine whether there are multiple identified swipe user IDs;

If there are multiple identified user IDs for scoring, identifying the similarity between the target user ID and the identified multiple scoring user IDs;

If there is a swipe user ID whose similarity to the target user ID is greater than a preset similarity threshold among the plurality of swipe user IDs, add swipe user-related features in the input features of the target user ID;

The step of inputting the input feature of the target user ID into a trained two-classification model is performed to obtain a preliminary classification result of the input feature of the target user ID.
A user identification device, characterized in that the user identification device includes a first determination unit, an acquisition unit, an identification unit and a second determination unit, wherein:

The first determining unit is configured to determine whether there are identified N brushing groups when the target user ID needs to be identified. The N brushing groups are classified according to group user rules. The user ID contained in any one of the two groups is greater than the preset number threshold, and N is a positive integer;

The acquiring unit is configured to acquire the input characteristics of the target user ID when the first determining unit determines that there are N swiping groups that have been identified, and the input characteristics include user location characteristics and user APP usage Features, user equipment usage features and user click-through rate CTR features;

The identification unit is configured to identify the similarity between the target user ID and each of the N brush amount groups based on the input characteristics of the target user ID;

The second determining unit is configured to determine the brush group that has a similarity with the target user ID greater than a preset similarity threshold among the N brush groups The target user ID is the ID of the user who swipes.
A server is characterized by comprising a processor and a memory, the memory is used to store one or more programs, the one or more programs are configured to be executed by the processor, and the programs include Perform the method according to any one of claims 1-7.
A computer-readable storage medium, characterized in that the computer-readable storage medium is used to store a computer program for electronic data exchange, wherein the computer program causes a computer to execute any one of claims 1-7 method.