User identification method, device and equipment
By establishing a judder with multi-dimensional user data, the problem of low accuracy of user judgment in the prior art is solved, and more efficient and accurate user identification is achieved.
Patent Information
- Application Number
- CN202111192236.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-10-13
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2041-10-13
AI Technical Summary
The existing video brush volume user discrimination technology has a single data dimension, which leads to the problem of low accuracy of discrimination results.
A multi-dimensional judgmenter is established through multi-dimensional user data, and the user data is comprehensively processed to identify target users, improving identification efficiency and accuracy.
It effectively overcomes the inaccurate judgment caused by single-dimensional data analysis and empirical judgment, and improves the accuracy and efficiency of flash user recognition.
Smart Images

Figure CN113947139B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data analysis, and in particular to a method, device and equipment for identifying a user. Background Art
[0002] There are currently two commonly used technologies for identifying video brushing users. One is to obtain the user's unique identifier, such as IP, cookie, IMEI, and match it with a known brushing user database. If it can be matched, it is considered a brushing user, otherwise it is not; the known brushing user database is generally obtained by purchasing a third-party database, which is costly. The other is to collect user behavior data (such as login, playback, exposure, etc.), analyze the user's usage time, playback time, number of playbacks, click-through rate, etc. through several dimensions, and use experience to determine whether the user is brushing.
[0003] These two methods have the following problems: the data dimension is single and the accuracy of the results of brush volume identification is low. Summary of the invention
[0004] In view of the above problems, embodiments of the present invention are proposed to provide a method, apparatus and device for identifying a user that overcomes the above problems or at least partially solves the above problems.
[0005] According to one aspect of an embodiment of the present invention, a method for identifying a user is provided, the method comprising:
[0006] Determine a first type of user and a second type of user according to user data of at least two dimensions;
[0007] generating at least two evaluators according to the user data of at least two dimensions of the first type of users;
[0008] The user data of at least two dimensions of the second type of user is input into the at least two evaluators for processing to obtain a judgment result of whether the second type of user is a target type of user.
[0009] According to another aspect of an embodiment of the present invention, a user identification device is provided, the device comprising:
[0010] A determination module, configured to determine a first type of user and a second type of user based on user data of at least two dimensions;
[0011] The processing module is used to generate at least two evaluators according to the user data of at least two dimensions of the first type of users; input the user data of at least two dimensions of the second type of users into the at least two evaluators for processing, and obtain a judgment result of whether the second type of users are target type users.
[0012] According to another aspect of an embodiment of the present invention, there is provided a computing device, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus;
[0013] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the above-mentioned method for identifying users who have increased their traffic.
[0014] According to another aspect of the embodiments of the present invention, there is provided a computer storage medium, wherein the storage medium stores at least one executable instruction, and the executable instruction enables a processor to execute operations corresponding to the method for identifying users who have increased their traffic as described above.
[0015] According to the solution provided by the above-mentioned embodiment of the present invention, based on the input multi-dimensional user data, evaluators corresponding to the multi-dimensional data are established. Based on multiple evaluators, target type users, that is, brushing users, can be accurately identified, thereby solving the problems of inaccurate judgment caused by single-dimensional user data analysis and experience judgment, and improving recognition efficiency and accuracy.
[0016] The above description is only an overview of the technical solution of the embodiment of the present invention. In order to more clearly understand the technical means of the embodiment of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiment of the present invention more obvious and easy to understand, the specific implementation method of the embodiment of the present invention is specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the embodiments of the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:
[0018] Figure 1 A flow chart of a method for identifying a user provided by an embodiment of the present invention is shown;
[0019] Figure 2 A specific implementation flow chart of the user identification method provided by an embodiment of the present invention is shown;
[0020] Figure 3 A schematic diagram showing the structure of a user identification device provided by an embodiment of the present invention;
[0021] Figure 4 A schematic diagram of the structure of a computing device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION
[0022] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.
[0023] Figure 1 FIG. 1 is a flow chart showing a method for identifying a user provided by an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:
[0024] Step 11, determining a first type of user and a second type of user according to user data of at least two dimensions;
[0025] Step 12, generating at least two evaluators according to the user data of at least two dimensions of the first type;
[0026] Step 13: input the at least two dimensional user data of the second type of user into the at least two evaluators for processing to obtain a judgment result of whether the second type of user is a target type of user.
[0027] In this embodiment, the user data of at least two dimensions may include: user basic data, playback data, content data and other latitudes of user video playback feature data, and these feature data are combined to provide a data basis for the implementation of subsequent steps; at the same time, the user data of at least two dimensions avoids the misjudgment caused by the use of single-dimensional data analysis and experience judgment, and increases the cost of fake volume; in the user data of at least two dimensions, the first type of users and the second type of users are screened according to certain rules, and the first type of users and related playback feature data are used to generate at least two judges; the user data of at least two dimensions of the second type of users (such as playback feature data) are input into the at least two judges for processing, and the output results of the at least two judges are combined to finally obtain the judgment result of whether the second type of user is the target type of user. The results of the at least two judges can be used to deal with different scenes and different types of videos, and dynamically adjusted for use.
[0028] In an optional embodiment of the present invention, in step 11, the first type of user is a user whose user data satisfies a first condition; the first condition includes: a first sub-condition for determining that the first type of user is a normal user and a second sub-condition for determining that the first type of user is a brushing user; the second type of user is a user whose user data satisfies the second condition.
[0029] In this embodiment, the first type of users may include normal users and users who have been inflated, and the second type of users may include suspected users who have been inflated. The first sub-condition may be to determine that the user corresponding to the first data of interactive interaction in the at least two dimensional data is a normal user; the second sub-condition may be to determine that the user corresponding to the second data is a user who has been inflated according to the second data that meets the mandatory user inflated rules in the at least two dimensional data; the second condition may be the third data in the multi-dimensional user data except the first data and the second data, and determine that the user corresponding to the third data is a suspected user who has been inflated.
[0030] The interactive data in the at least two dimensional data is used as the first data, for example, the interactive data such as user payment and participation in marketing activities, and the user screening rules corresponding to the first data are formulated through business data analysis and playback business process analysis, such as participating in the video interactive function during playback, or the playback is derived from participation in marketing activities, or the paid playback has social behaviors such as forwarding, collection, commenting, etc., and the user corresponding to the first data is determined as a normal user based on this rule;
[0031] The data that meets the mandatory user rules for brushing volume in the at least two dimensional data is used as the second data, for example, data such as the same playback duration of different videos of a single user (accurate to milliseconds), and user screening and determination rules corresponding to the second data with strong business characteristics are formulated through data analysis of user playback life cycle data, such as obvious brushing behaviors such as excessive daily playback duration, excessively high video playback frequency, and certain commonalities in video playback duration, and the user corresponding to the second data is determined as a brushing volume user based on this rule;
[0032] The remaining user data after filtering by the above rules is determined as the third data, and the user corresponding to the third data is identified as a suspected brushing user. By formulating user data filtering rules and inputting multi-dimensional data for selection and judgment, the accuracy of filtering is improved.
[0033] Preferably, the user data of at least two dimensions can be obtained according to the following steps:
[0034] Step S1, statistics are obtained for the daily playback duration and number of playbacks of a single user in different dimensions (such as different network status, different versions, different time periods, different payment status, different video types, and different priority video content). Network status such as 4G, WIFI; version information such as a specified version, low version, high version, etc.; payment status such as various consumption scenarios such as membership; video types such as movies, TV series, variety shows, etc.; different priority information such as the quality level of the video content itself, etc. Statistics are collected for the playback-related data of users in various dimensions to form high-dimensional feature data of users and restore the real playback of users.
[0035] Step S2, calculate the user's access behavior data on a specific page on that day. Calculate the number of visits, stay time and other basic access and activity related data of the user on key pages such as the homepage and VIP page. Basic access data covers the user's basic operation behavior and provides data support for the subsequent determination of the process of brushing users.
[0036] Step S3, calculate the number and duration of marketing activities, SMS activities, social interaction information and video interaction that the user has participated in on the day. Obtain the user's operation behavior data in the non-brush state to provide data support for the subsequent determination of normal user processes.
[0037] Step S4, count all the data of the user in step 111, step 112, and step 113 in the past N days, perform data summation, maximum value, minimum value, average value and other operations according to different data categories, and generate user data for the past N days. A single piece of data contains 67 dimensions, which truly reflects the actual operation characteristics of the user.
[0038] It should be noted that there is no necessary order restriction for steps S1 to S3 here, and one or more steps can be implemented separately or in combination.
[0039] In an optional embodiment of the present invention, the step 12 includes:
[0040] Step 121 : generating, based on the user data of each dimension of at least two dimensions of the user data of the first type of users, an evaluator corresponding to the user data of each dimension.
[0041] In this embodiment, based on the high-dimensional feature data of the normal users and the brushing users among the first type of users, a plurality of brushing evaluators of different dimensions can be established. Since each evaluator selects different feature data, the judgment threshold of each evaluator is different. Accordingly, when filtering data, the filtering range is wider.
[0042] In an optional embodiment of the present invention, the step 121 may include:
[0043] Step 1211, based on the occurrence probabilities of normal users and determined brushing users in the first type of users determined by the user data of each dimension in at least two dimensions of the user data of the first type of users, generate an evaluator corresponding to the user data of each dimension.
[0044] In this embodiment, based on the occurrence probabilities of normal users and brushing users among the first type of users, evaluators corresponding to the user data of each dimension are generated, thereby further improving the accuracy of data screening.
[0045] In an optional embodiment of the present invention, the step 1211 may include:
[0046] Step 12111: calculating the information entropy of the normal users and the brushing users according to the user data of each dimension of at least two dimensions of the user data of the first type of users;
[0047] Step 12112, obtaining the credibility of the evaluator corresponding to the user data of the at least two dimensions respectively according to the information entropy;
[0048] Step 12113: Generate an evaluator corresponding to the dimension based on the credibility and the user data of the dimension corresponding to the credibility.
[0049] In this embodiment, the information entropy of the occurrence probability of the normal users and the users who make a lot of money can be calculated to screen the user data of the corresponding feature dimension; according to the information entropy, the credibility of the evaluator corresponding to the user data of the feature dimension is obtained, and then according to the credibility and the user data of the feature dimension corresponding to the credibility, the evaluator corresponding to the user data of the feature dimension is generated. Since each evaluator selects different user data of the feature dimension, each evaluator has a different judgment threshold, which further improves the accuracy of data screening.
[0050] Preferably, step 12111 may include:
[0051] According to the formula: H = -∑ i p(i)log2p(i), calculate the information entropy of the normal users and the determined brushing users.
[0052] In this embodiment, H is information entropy, i=first value, indicating a normal user, i=second value, indicating a user who is brushing traffic;
[0053] When i=the first value, pi is the ratio of the number of normal users identified from the total number of normal users and brushing users according to a threshold to the total number of users;
[0054] When i=the second value, pi is the ratio of the number of users who have increased the amount of traffic identified from the total number of normal users and users who have increased the amount of traffic to the total number of users according to the threshold.
[0055] For example, taking the feature (A1) of the number of times a TV series is played as an example, the information entropy of the feature is calculated, assuming that a total of D1=400 normal users and a total of D2=50 users who brush the volume are selected.
[0056] A value of the feature is randomly selected as the identification threshold, and the total number of normal users and inflated users are identified according to the threshold: normal users C'1=410, of which C1=370 are correctly identified normal users; inflated users C'2=40, of which C2=30 are correctly identified inflated users.
[0057] First, according to the information entropy calculation formula H = -∑ i p(i)log2P(i), calculate the known user set information entropy H=0.5033.
[0058] i=0 indicates a normal user, and pi is 370 / 450;
[0059] i=1 indicates a user who increases the amount of traffic, and pi is 30 / 450.
[0060] Preferably, step 12113 may include:
[0061] According to the credibility and the user data of the dimension corresponding to the credibility, according to the formula Obtain the judgement device corresponding to the dimension.
[0062] Wherein, L(i) is the identifier of the judge corresponding to the dimension, To define the model function, γ is the credibility, Z=WX+b is the linear transformation function, X is the user data of the dimension corresponding to the credibility, y is the user identifier, W and b are the coefficients of the linear function, which are preset values.
[0063] Taking the construction of content dimension evaluator as an example, assuming that the selected user content dimension data includes the number of plays of TV series, movies, variety shows, key TV series, key movies, key variety shows, etc., forming the input data X = (x1, x2, x3, x4, x5, x6), the user identifier forms y = (y1, y2...yn), where W = (w1, w2, w3, w4, w5, w6), according to the maximization objective function Generate a content dimension judge.
[0064] The content dimension evaluator outputs an identification of suspected fake playback. Based on the input feature credibility γ of the evaluator, the threshold t of the evaluator is defined as t=-γlog(P / P+N), where P is the number of users confirmed to have faked the playback and N is the number of normal users.
[0065] The content dimension playback feature data of the suspected brushing user is used as the judge input, and the judge output value is compared with the threshold t to determine the qualitative identification {0, 1} of whether the playback is brushing. Based on the credibility and the user data of the dimension corresponding to the credibility, the judge corresponding to the dimension is established, and the judge threshold is dynamically adjusted to optimize the judge parameter setting.
[0066] In an optional embodiment of the present invention, the step 12112 may include:
[0067] Step 121121, calculating the conditional entropy of the normal users and the brushing users according to the user data of at least one dimension of the normal users and the brushing users;
[0068] Step 121122: Obtain feature information gain of the user data of the at least one dimension according to the conditional entropy and the information entropy;
[0069] Step 121123, obtaining the credibility of the evaluator according to the feature information gain.
[0070] In this embodiment, the information entropy of the normal users and the brushing users calculated in the above embodiment is used to calculate the conditional entropy of the user set corresponding to the feature identification according to the information entropy and the conditional entropy calculation formula, and then the feature information gain of the user data of the at least one dimension is obtained according to the conditional entropy and the information entropy, and finally the credibility of the evaluator is obtained according to the feature information gain and the credibility calculation formula.
[0071] The feature information gain value of the user data is calculated by the data importance of normal users and brushing users in at least one dimension, and the reliability of each feature is accurately calculated using the calculated feature information gain value; the strong correlation between information entropy, conditional entropy feature information gain, feature information gain and credibility avoids the problem of inaccurate judgment caused by the method of using artificial experience to judge and set evaluation conditions in the existing method, thereby improving the accuracy of judgment.
[0072] In step 121121, according to the user data of at least one dimension of the normal users and the brushing users, according to the formula The conditional entropy of the normal users and the users who make fake traffic is calculated.
[0073] Among them, H(A1) is the conditional entropy, C’j is the number of users with fake data determined according to a threshold, Cj is the number of correct users with fake data in C’j identified according to a threshold, D is the total number of normal users and users with fake data, and D i is the number of normal users or the number of users with fake data.
[0074] In step 121122, according to the conditional entropy and the information entropy, based on the formula G(A1) = H - H(A1), the feature information gain of the user data of the at least one dimension is obtained; where G(A1) is the feature information gain.
[0075] In step 121123, according to the feature information gain, based on the formula the credibility of the discriminator is obtained, where γ k is the credibility, k is the kth feature, and G(K) is the feature information gain of the kth feature.
[0076] In an optional embodiment of the present invention, step 13 may include:
[0077] Step 131, input the user data of at least two dimensions of the second type of users into the discriminators corresponding to the user data of each dimension for processing, and obtain the processing results of each discriminator;
[0078] Step 132, perform weighted operation processing on the processing results of each discriminator according to the weight of each discriminator, and obtain the judgment result on whether the second type of users are users with fake data.
[0079] In this embodiment, the user data of at least two dimensions in the third data corresponding to the second type of users is input into the corresponding discriminator for calculating information entropy, conditional entropy, feasibility, etc., to obtain the discriminator identification. According to the formula L = ∑ n Q(n)L(n) to obtain the judgment result on whether the suspected users with fake data are users with fake data, where Q(n) is the weight of the nth discriminator (0 < Q < 1), L(n) ∈ {0, 1}, L(n) is the identification of the nth discriminator, and the identification L (0 < L < 1) is a quantitative measure of the user's play being a play with fake data.
[0080] For example, according to data such as activity period, number of participants in the activity, and activity rewards, an activity quality regression model is established. Combining the activity quality coefficient output by the model, according to the current activity quality, the fake data determination threshold is adjusted again, and comparing with the identification L output in the above embodiment, the final fake data identification {0, 1} is output.
[0081] The following combines Figure 2 the specific implementation process module diagram to illustrate the specific implementation process of the above method:
[0082] S201, obtain statistics on the daily playback time and number of times of a single user in different dimensions, and calculate the number of times and playback time (in seconds) of video content of different network status, different versions, different time periods, different payment status, different video types, and different priorities of the user. Network status is such as: 4G, WIFI, etc.; version information is such as: a certain version, low version, high version, etc.; payment status is such as: various consumption scenarios such as membership; video types are such as: movies, TV series, variety shows, etc.; different priority information is such as: the quality level of the video content itself, etc. Statistics are collected on the playback-related data of the user in each dimension to form high-dimensional feature data of the user and restore the user's real playback.
[0083] S202, calculate the user's access behavior data on a specific page on that day. Calculate the number of visits, stay time and other basic access and activity related data of the user on key pages such as the home page and VIP page. The basic access data covers the user's basic operation behavior and provides data support for the subsequent determination of the process of brushing users.
[0084] S203, calculate the number and duration of marketing activities, SMS activities, social interaction information and video interaction of the user on the day, etc. Obtain the user's operation behavior data in the non-brush state to provide data support for the subsequent determination of normal user processes.
[0085] S204, collects statistics of all data of users S201, S202, and S203 in the past N days, and performs operations such as data summation, maximum value, minimum value, and average value according to different data categories to form user feature data that truly reflects the actual operation characteristics of the users.
[0086] S205, simulate the real operation of user playback, and filter out some normal playback users with strong business characteristics. Through business data analysis and playback business process analysis, formulate N normal user screening rules, such as participating in the video interactive function during playback, or the playback comes from participating in marketing activities, or paid playback and there are social behaviors such as forwarding, collection, and commenting.
[0087] S206, through data analysis of user playback life cycle data, formulate rules for determining users with strong business characteristics, such as excessive daily playback time, too high video playback frequency, certain commonality in video playback time, etc., which are obvious brushing users. Users who do not have the characteristics of S205 are defined as obvious brushing users.
[0088] S207, determine the users suspected of inflating the volume of playback. After S205, the normal playback users with strong characteristics are screened out, and S206, the inflating users with strong characteristics are screened out, and the remaining users are classified as suspected inflating the volume of playback users.
[0089] S208, combine the user high-dimensional playback feature data of S205, S206, and S204 to establish multiple different-dimensional brushing evaluators. When each evaluator outputs a label, it dynamically adjusts the judgment threshold according to the credibility of the input feature of the evaluator and outputs the judgment result. Since each evaluator selects different data features, the judgment threshold of each evaluator is different. The main steps of establishing an evaluator include:
[0090] Step i), calculating the information entropy of each feature, which is used to screen the feature dimension, adjust the judgment threshold of the judge, and calculate the weight coefficient of each judge;
[0091] Step ii), based on the existing abnormal playback and normal playback data, minimize the error function, solve the feature vector quantization parameters of each dimension, and generate an evaluator.
[0092] S209, taking the construction of a content dimension judge as an example, the content judge outputs an identification of suspected brushing playback. Based on the credibility of the judge's input features, the threshold of the judge is defined. The content dimension playback feature data of the suspected brushing user is used as the input of the judge, and the judge output value is compared with the threshold to determine the qualitative identification {0,1} of whether the playback is brushing.
[0093] S210, taking the construction of a network dimension judge as an example, the network judge outputs an identification of suspected fake playback. Similar to the process of S209, the judgment threshold of the network judge is calculated, and the user playback feature data of the network dimension is used as the input of the judge, and the judge outputs a qualitative identification {0,1} of whether the playback is fake.
[0094] S211, taking the construction of a version dimension judge as an example, the version judge outputs an identification of suspected fake playback. Similar to the process of S209, the judgment threshold of the version judge is calculated, and the user playback feature data of the version dimension is used as the input of the judge, and the judge outputs a qualitative identification {0,1} of whether the playback is fake.
[0095] S212, weight all judges and output the final identification. For each judge, according to the feature credibility calculated in S208, assign different weights to the judge to obtain the final judgment value. Output the quantitative measurement of whether the user's play is a fake play.
[0096] S213, establish an activity quality regression model based on activity time period, number of activity participants, activity rewards and other data. Combined with the activity quality coefficient output by the model, adjust the brushing volume determination threshold again according to the current activity quality, compare the user brushing volume value output in S212, and output the final brushing volume flag {0,1}.
[0097] The above-mentioned embodiment of the present invention takes multi-dimensional data such as user basic attributes, payment behavior, social behavior, playback behavior, video content attributes and playback life cycle as input, and establishes an automatic judge through strong rule recognition and model training. It avoids the inaccurate judgment caused by the use of single-dimensional data analysis and experience judgment, increases the cost of fake volume, and reduces the possibility of bypassing the volume judgment system. Multi-dimensional data more realistically restores the user's playback scene, and combines business data analysis with the comprehensive judgment of multiple judges to make more accurate identification. Based on the credibility of data features, an evaluator is established, and the thresholds and weights of each evaluator are dynamically adjusted. The evaluator and data features are closely combined, the parameter setting process of the evaluator is optimized, the dependence on the experience of analysts is reduced, and the degree of automation in the evaluator generation process is further improved. The above-mentioned embodiment can not only output the qualitative results of the volume playback, but also quantify the judgment results, which can more conveniently deal with different business scenarios.
[0098] Figure 3 FIG. 2 shows a schematic diagram of a user identification device provided by an embodiment of the present invention. Figure 3 As shown, the device comprises:
[0099] A determination module 31, configured to determine a first type of user and a second type of user according to user data of at least two dimensions;
[0100] The processing module 32 is used to determine the first type of users and the second type of users according to the user data of the at least two dimensions, and to generate at least two evaluators according to the user data of the at least two dimensions of the first type of users, and to input the user data of the at least two dimensions of the second type of users into the at least two evaluators for processing to obtain a judgment result of whether the second type of users are target type users.
[0101] Optionally, the determination module 31 determines that the first type of user is a user whose user data satisfies a first condition; the first condition includes: a first sub-condition for determining that the first type of user is a normal user and a second sub-condition for determining that the first type of user is a brushing user;
[0102] The second type of users are users whose user data meets the second condition.
[0103] Optionally, the processing module 32 may generate at least two evaluators according to the user data of at least two dimensions of the first type of users, including:
[0104] Based on the occurrence probabilities of normal users and brushing users in the first type of users determined according to the user data of each dimension in at least two dimensions of the user data of the first type of users, an evaluator corresponding to the user data of each dimension is generated.
[0105] Optionally, the determination module 32 determines the occurrence probability of normal users and brushing users in the first type of users according to the user data of each dimension of at least two dimensions of the user data of the first type of users, and generates an evaluator corresponding to the user data of each dimension, including:
[0106] Calculate the information entropy of the occurrence probability of the normal users and the brushing users according to the user data of each dimension of at least two dimensions of the user data of the first type of users;
[0107] According to the information entropy, obtaining the credibility of the evaluator corresponding to the user data of the at least two dimensions respectively;
[0108] According to the credibility and the user data of the dimension corresponding to the credibility, an evaluator corresponding to the user data of the dimension is generated.
[0109] Optionally, the determining module 32 obtains the credibility of the evaluator corresponding to the user data of the at least two dimensions respectively according to the information entropy, including:
[0110] Calculate the conditional entropy of the normal users and the inflated users according to the user data of at least two dimensions of the normal users and the inflated users;
[0111] Obtaining feature information gain of the user data of the at least two dimensions according to the conditional entropy and the information entropy;
[0112] The credibility of the judge is obtained according to the feature information gain.
[0113] Optionally, the determination module 32 inputs the user data of at least two dimensions of the second type of user into the at least two evaluators for processing, and obtains a judgment result of whether the second type of user is a target type of user, including:
[0114] Inputting the user data of at least two dimensions of the second type of users into the evaluator corresponding to each dimension of the user data for processing, respectively, to obtain a processing result of each evaluator;
[0115] The processing result of each evaluator is weighted according to the weight of each evaluator to obtain a judgment result of whether the second type of user is a confirmed brushing user.
[0116] The device is a device corresponding to the above method, and all implementation methods in the above method embodiments are applicable to the embodiments of the device and can achieve the same technical effects.
[0117] An embodiment of the present invention provides a non-volatile computer storage medium, wherein the computer storage medium stores at least one executable instruction, and the computer executable instruction can execute the method for identifying users who increase or decrease the amount of traffic in any of the above method embodiments.
[0118] Figure 4 A schematic diagram of the structure of a computing device provided by an embodiment of the present invention is shown, and the specific embodiment of the present invention does not limit the specific implementation of the computing device.
[0119] like Figure 4 As shown, the computing device may include: a processor, a communications interface, a memory, and a communication bus.
[0120] Wherein: the processor, the communication interface, and the memory communicate with each other via a communication bus. The communication interface is used to communicate with network elements of other devices such as a client or other servers. The processor is used to execute a program, and specifically can execute the relevant steps in the embodiment of the method for identifying a user who has been swiping the amount of data on a computing device.
[0121] Specifically, the program may include program codes including computer operation instructions.
[0122] The processor may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. The one or more processors included in the computing device may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.
[0123] The memory is used to store programs. The memory may include a high-speed RAM memory, and may also include a non-volatile memory, such as at least one disk memory.
[0124] The program can be specifically used to enable the processor to execute the method for identifying a user who has swiped the volume in any of the above-mentioned method embodiments. The specific implementation of each step in the program can refer to the corresponding descriptions in the corresponding steps and units in the above-mentioned method for identifying a user who has swiped the volume, and will not be repeated here. A person skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described devices and modules can refer to the corresponding process description in the above-mentioned method embodiments, and will not be repeated here.
[0125] The algorithm or display provided herein is not inherently related to any particular computer, virtual system or other device. Various general purpose systems can also be used together with the teachings based on this. According to the above description, it is obvious that the structure required for constructing this type of system. In addition, the embodiment of the present invention is not directed to any specific programming language either. It should be understood that various programming languages can be utilized to realize the content of the embodiment of the present invention described herein, and the description of the above specific language is to disclose the best mode of the embodiment of the present invention.
[0126] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.
[0127] Similarly, it should be understood that in order to streamline the embodiments of the present invention and to aid in understanding one or more of the various inventive aspects, in the above description of the exemplary embodiments of the present invention, the various features of the embodiments of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, the disclosed method should not be interpreted as reflecting the following intention: that the claimed embodiments of the present invention require more features than the features explicitly recited in each claim. More specifically, as reflected in the claims below, the inventive aspects lie in less than all the features of the single embodiment disclosed above. Therefore, the claims that follow the specific embodiment are hereby expressly incorporated into the specific embodiment, wherein each claim itself serves as a separate embodiment of the present invention.
[0128] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and in addition they may be divided into a plurality of submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive, all features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner may be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.
[0129] In addition, those skilled in the art will appreciate that, although some embodiments herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present invention and form different embodiments. For example, in the claims below, any one of the claimed embodiments may be used in any combination.
[0130] The various component embodiments of the present invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. It should be understood by those skilled in the art that a microprocessor or digital signal processor (DSP) may be used in practice to implement some or all of the functions of some or all of the components according to the embodiments of the present invention. The embodiments of the present invention may also be implemented as a device or apparatus program (e.g., a computer program and a computer program product) for executing part or all of the methods described herein. Such a program implementing an embodiment of the present invention may be stored on a computer-readable medium, or may have the form of one or more signals. Such a signal may be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.
[0131] It should be noted that the above embodiments illustrate rather than limit the present invention, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be construed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising a number of different elements and by means of a suitably programmed computer. In a unit claim listing a number of devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be understood as limitations on the order of execution.
Claims
1. A method for identifying a user, characterized in that: The method comprises: Determine a first type of user and a second type of user according to user data of at least two dimensions; Calculate the information entropy of the occurrence probability of normal users and brushing users according to the user data of each dimension of at least two dimensions of the user data of the first type of users; Obtaining, according to the information entropy, the credibility of at least two evaluators corresponding to the user data of the at least two dimensions respectively; generating, according to the credibility and the user data of the dimension corresponding to the credibility, an evaluator corresponding to the user data of the dimension; The user data of at least two dimensions of the second type of user is input into the at least two evaluators for processing to obtain a judgment result of whether the second type of user is a target type of user.
2. The user identification method according to claim 1, characterized in that: The first type of user is a user whose user data satisfies a first condition; the first condition includes: a first sub-condition for determining that the first type of user is a normal user and a second sub-condition for determining that the first type of user is a confirmed brushing user; The second type of users are users whose user data meets the second condition.
3. The user identification method according to claim 1, characterized in that: Obtaining the credibility of at least two evaluators corresponding to the user data of the at least two dimensions respectively according to the information entropy, including: Calculate the conditional entropy of the normal users and the inflated users according to the user data of at least two dimensions of the normal users and the inflated users; Obtaining feature information gain of the user data of the at least two dimensions according to the conditional entropy and the information entropy; According to the feature information gain, the credibility of at least two evaluators is obtained.
4. The user identification method according to claim 1, characterized in that: Inputting the user data of at least two dimensions of the second type of user into the at least two evaluators for processing, and obtaining a judgment result of whether the second type of user is a target type of user, includes: Inputting the user data of at least two dimensions of the second type of users into the evaluator corresponding to each dimension of the user data for processing, respectively, to obtain a processing result of each evaluator; The processing result of each evaluator is weighted according to the weight of each evaluator to obtain a judgment result of whether the second type of user is a brushing user.
5. A user identification device, characterized in that: The device comprises: A determination module, which determines a first type of user and a second type of user according to user data of at least two dimensions; The processing module is used to calculate the information entropy of the occurrence probability of normal users and brushing users based on the user data of each dimension of at least two dimensions of the user data of the first type of users; obtain the credibility of at least two evaluators corresponding to the user data of the at least two dimensions according to the information entropy; generate the evaluator corresponding to the user data of the dimension according to the credibility and the user data of the dimension corresponding to the credibility; input the user data of at least two dimensions of the second type of users into the at least two evaluators for processing, and obtain the judgment result of whether the second type of users are target type users.
6. A computing device comprising: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the method for identifying users who have swiped their volume as described in any one of claims 1-4.
7. A computer storage medium, wherein at least one executable instruction is stored in the storage medium, and the executable instruction enables a processor to execute an operation corresponding to the method for identifying a user who has increased the amount of traffic as described in any one of claims 1 to 4.
Citation Information
Patent Citations
Risk identification method and device based on user behaviors as well as storage medium
CN107483500A
Data processing method and device and electronic equipment
CN111242182A
Plug-in identification method and device, computer equipment and readable storage medium
CN111417021A
Method and device for acquiring explainable reason of abnormal user
CN113255929A
User identification method and related product
CN113383362A
