Traffic fraud behavior identification method, device, storage medium and electronic device

The method enhances traffic fraud identification by analyzing user click data to detect clustering behavior and optimize time period division, addressing the limitations of current methods in identifying group cheating and periodic behaviors.

CN113592036BActive Publication Date: 2024-12-03BEIJING WODONG TIANJUN INFORMATION TECH CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202110981015.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-08-25
Publication Date
2024-12-03
Estimated Expiration
2041-08-25

AI Technical Summary

Technical Problem

Current methods for identifying traffic fraud are inadequate in detecting group traffic cheating behaviors and are based on empirical data, leading to errors and inability to identify periodic cheating behaviors across various time periods.

Method used

A method that analyzes user click data to identify clustering behavior among users by calculating click behavior similarity, adding users with similar patterns to a traffic cheating group, and optimizing the time period division for improved accuracy.

Benefits of technology

Effectively identifies group traffic cheating behaviors by reducing misjudgment and improving detection accuracy through user click behavior analysis and optimized time period division.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113592036B_ABST
    Figure CN113592036B_ABST
Patent Text Reader

Abstract

The embodiments of the present invention propose a method, device, storage medium and electronic device for identifying traffic cheating behavior. The method includes: obtaining user click data of users visiting a web page; extracting user click data of users visiting a first link within a first time period from the user click data; extracting user click data of each user from the user click data of users visiting the first link within the first time period; extracting the click behavior characteristics of each user from the user click data of each user; according to the click behavior characteristics of each user, determining whether there is clustering behavior between users, and if so, adding users with clustering behavior to the traffic cheating group. The embodiments of the present invention realize the identification of group traffic cheating behavior.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of Internet access technology, and in particular to a method and device for identifying traffic fraud behavior, a readable storage medium, and an electronic device. Background Art

[0002] In the current Internet era of information explosion, traffic is of vital importance in the Internet world. False traffic has become a common behavior on the Internet. Currently, there have been group traffic cheating behaviors that use a large number of manual or machine simulations to generate traffic.

[0003] The current method of identifying traffic fraud is to identify a single fraud device. This has the following disadvantages:

[0004] 1. Unable to identify group traffic cheating behavior;

[0005] Second, the identification methods are all based on empirical data and rules, which have certain errors and can only identify traffic cheating behaviors in a short period of time. For periodic traffic cheating behaviors with monthly, quarterly, annual, etc. periods, when arbitrarily selecting a period to identify traffic cheating behaviors, the existing identification methods cannot identify them. Summary of the invention

[0006] The embodiments of the present invention provide a method and device for identifying traffic cheating behavior, a readable storage medium and an electronic device to identify group traffic cheating behavior.

[0007] The technical solution of the embodiment of the present invention is achieved as follows:

[0008] A method for identifying traffic fraud behavior, the method comprising:

[0009] Get user click data of web pages visited by users;

[0010] Extracting user click data that accesses the first link within a first time period from the user click data;

[0011] Extracting user click data of each user from the click data of users who accessed the first link during the first time period;

[0012] Extracting click behavior features of each user from the user click data of each user;

[0013] Based on the click behavior characteristics of each user, determine whether there is clustering behavior among users. If so, the users with clustering behavior will be added to the traffic cheating group.

[0014] Determining whether there is a grouping behavior among users includes:

[0015] According to the click behavior characteristics of each user, the click behavior similarity between two users is calculated. If any click behavior similarity is greater than a preset similarity threshold, it is determined that clustering behavior exists between the two users corresponding to the click behavior similarity.

[0016] The user click data includes: user identification information, the web page link identification clicked by the user and the user click time;

[0017] The extracting the click behavior feature of each user from the user click data of each user comprises: dividing the first time period into at least one sub-time period;

[0018] For each sub-time period, obtaining user click data of each user who accesses the first link in the sub-time period;

[0019] For each sub-time period, all users who access the first link in the sub-time period are fully paired in pairs;

[0020] For each user pair that accesses the first link in each sub-time period, based on the user click data of the two users in the user pair in the current sub-time period, the number of days that the two users in the user pair simultaneously access the first link in the current sub-time period is counted, which is regarded as the first type of days, and the number of days that only one of the two users in the user pair accesses the first link in the current sub-time period is counted, which is regarded as the second type of days;

[0021] The first type of days and the second type of days are used as the click behavior features of the two users in the user pair in the current sub-time period.

[0022] Determining whether there is a grouping behavior among users includes:

[0023] For each user pair that accesses the first link in each sub-time period, calculate the sum of the first-category days and the second-category days corresponding to the user pair, and then divide the first-category days by the sum to obtain an initial value of the click behavior similarity between the two users in the user pair; divide the first-category days by the total number of days in the current sub-time period to obtain a weight; multiply the weight by the initial value of the click behavior similarity to obtain the click behavior similarity between the two users in the user pair; if the click behavior similarity is greater than a preset similarity threshold, it is determined that there is a clustering behavior between the two users in the user pair;

[0024] The aforementioned adding users with grouping behavior into the traffic fraud gang includes:

[0025] The two users in the user pair are added to the traffic cheating group in the current sub-time period.

[0026] After adding the two users in the user pair to the traffic cheating gang in the current sub-time period, the method further includes:

[0027] When the traffic cheating gang sets of all sub-time periods within the first time period are obtained, users who have appeared in each traffic cheating gang set are selected;

[0028] Delete the users who have appeared in each traffic cheating gang set from each traffic cheating gang set, and obtain the updated traffic cheating gang set of each sub-time period;

[0029] For each user who has appeared in each traffic fraud gang set, calculate the similarity of the user's click behavior with each traffic fraud gang set, select the traffic fraud gang set with the greatest click behavior similarity, and add the user to the selected traffic fraud gang set.

[0030] The calculation of the click behavior similarity between the user and each traffic fraud gang set includes:

[0031] For each traffic fraud gang set, the click behavior similarity between the user and each user in the traffic fraud gang set is calculated respectively, and the smallest click behavior similarity is selected as the click behavior similarity between the user and the traffic fraud gang set.

[0032] The step of dividing the first time period into at least one sub-time period includes:

[0033] Dividing the first time period into at least one sub-time period according to a preset initial length of the sub-time period, or according to the length of the current sub-time period and the adjustment step;

[0034] Furthermore, after adding the two users in the user pair to the traffic cheating group in the current sub-time period, the method further includes:

[0035] After obtaining the traffic cheating gang set of all current sub-time periods, determine whether the length of the current sub-time period reaches the maximum length of the preset sub-time period;

[0036] If not, returning to the action of dividing the first time period into at least one sub-time period according to the length of the current sub-time period and the adjustment step;

[0037] If so, according to the principle that the more members in the detected traffic cheating group set, the better the detection result, the best at least one traffic cheating group set is selected as the final detection result from the at least one traffic cheating group set detected each time.

[0038] The step of dividing the first time period into at least one sub-time period includes:

[0039] The first time period is divided into a plurality of sub-time periods, and two adjacent sub-time periods overlap for a preset second time length.

[0040] After extracting the user click data of each user from the click data of users who accessed the first link within the first time period and before extracting the click behavior feature of each user from the user click data of each user, the method further includes:

[0041] For each user, based on the user click data of the user, the maximum duration of the user's continuous access to the first link within the first time period is counted; if the maximum duration is less than the preset first duration, it is determined that the user is not a member of the traffic cheating gang, and the user click data of the user is deleted from the user click data of each user who accessed the first link within the first time period.

[0042] After adding the users with grouping behavior into the traffic cheating gang, the method further includes:

[0043] For each user who has grouped together, determine whether the user meets the following conditions. If so, delete the user from the traffic fraud group:

[0044] The user's stay time on the webpage of the first link is greater than a preset time threshold, or / and, the user's access depth to the first link is greater than a preset depth threshold, or / and, the number of item types purchased by the user on the first link is greater than a preset number.

[0045] A traffic fraud behavior identification device, the device comprising:

[0046] A click behavior feature extraction module is used to obtain user click data of users visiting a webpage; extract user click data of users visiting a first link within a first time period from the user click data; extract user click data of each user from the user click data of users visiting the first link within the first time period; extract the click behavior feature of each user from the user click data of each user;

[0047] The identification module is used to determine whether there is clustering behavior among users based on the click behavior characteristics of each user. If so, the users with clustering behavior will be added to the traffic fraud group.

[0048] A non-transitory computer-readable storage medium storing instructions, which, when executed by a processor, causes the processor to perform the steps of the method for identifying traffic fraud behavior as described in any one of the above items.

[0049] An electronic device includes the non-transitory computer-readable storage medium as described above, and the processor capable of accessing the non-transitory computer-readable storage medium.

[0050] In an embodiment of the present invention, the click data of users who access the first link within a first time period is extracted from the user click data, and then the click behavior characteristics of each user are extracted from the user click data of each user. According to the click behavior characteristics of each user, it is determined whether there is clustering behavior between users. If so, the users with clustering behavior are added to the traffic cheating group, thereby realizing the identification of group traffic cheating behavior. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0052] Figure 1 A flow chart of a method for identifying traffic fraud behavior provided by an embodiment of the present invention;

[0053] Figure 2 This is an application example of the present invention;

[0054] Figure 3 A flow chart of a method for identifying traffic fraud behavior provided by another embodiment of the present invention;

[0055] Figure 4 This is an application example of the sliding time window of the present invention;

[0056] Figure 5 A schematic diagram of the structure of a device for identifying traffic fraud behavior provided by an embodiment of the present invention;

[0057] Figure 6 An exemplary structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein, for example. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or devices.

[0060] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0061] The embodiment of the present invention proposes a method for identifying traffic cheating behavior, in which the user click data of users visiting a web page is obtained; the user click data of users visiting a first link within a first time period is extracted from the user click data; the user click data of each user is extracted from the user click data of users visiting the first link within the first time period; the click behavior characteristics of each user are extracted from the user click data of each user; according to the click behavior characteristics of each user, it is determined whether there is clustering behavior between users, and if so, the users with clustering behavior are added to the traffic cheating group. The embodiment of the present invention realizes the identification of group traffic cheating behavior.

[0062] Figure 1 This is a flow chart of a method for identifying traffic fraud behavior provided by an embodiment of the present invention, and the specific steps are as follows:

[0063] Step 101: Obtain user click data of a user visiting a web page.

[0064] In actual applications, user click data of users visiting web pages can be obtained from a webmaster group statistics table, a wake-up UUID (Universally Unique Identifier) ​​allocation table, a device browsing information table, and the like.

[0065] The user click data at least includes: user identification information, the web page link identification clicked by the user and the user click time, the web page link identification is such as: URL (Uniform Resource Locator, Uniform Resource Locator).

[0066] User identification information such as: user ID, user device ID, browser ID, login account, or any combination thereof.

[0067] The user click data may also include: user transaction information such as: the type of items purchased by the user, the user's GMV (Gross Merchandise Volume, total transaction amount), etc., or a combination thereof.

[0068] Step 102: extracting the user click data of users who accessed the first link within the first time period from the acquired user click data.

[0069] The first time period can be selected as needed.

[0070] Step 103: extracting the user click data of each user from the user click data of those who access the first link within the first time period.

[0071] Step 104: Extracting the click behavior characteristics of each user from the user click data of each user.

[0072] Step 105: Determine whether there is clustering behavior among users based on the click behavior characteristics of each user. If so, add the users with clustering behavior to the traffic cheating group.

[0073] In the above embodiment, by extracting the click data of users who accessed the first link within the first time period from the user click data, and then extracting the click behavior characteristics of each user from the user click data of each user, it is determined whether there is clustering behavior between users according to the click behavior characteristics of each user. If so, the users with clustering behavior are added to the traffic cheating group, thereby realizing the identification of group traffic cheating behavior.

[0074] In an optional embodiment, in step 105, determining whether there is clustering behavior between users includes: calculating the click behavior similarity between two users based on the click behavior characteristics of each user, and if any click behavior similarity is greater than a preset similarity threshold, determining that there is clustering behavior between the two users corresponding to the click behavior similarity.

[0075] In the above embodiment, the clustering behavior between users is judged by calculating the similarity of the click behaviors between two users.

[0076] In an optional embodiment, in step 104, extracting the click behavior characteristics of each user from the user click data of each user includes: dividing the first time period into at least one sub-time period; for each sub-time period, obtaining the user click data of each user who accesses the first link within the sub-time period; for each sub-time period, completely pairing all users who access the first link within the sub-time period; for each user pair who accesses the first link within each sub-time period, based on the user click data of the two users in the user pair in the current sub-time period, counting the number of days on which the two users in the user pair simultaneously access the first link within the current sub-time period, which are regarded as first-category days, and counting the number of days on which one and only one of the two users in the user pair accesses the first link within the current sub-time period, which are regarded as second-category days; and using the first-category days and the second-category days as the click behavior characteristics of the two users in the user pair in the current sub-time period.

[0077] Here, "complete pairing" means that for each user who accesses the first link in the sub-time period, the user is paired with each other user. For example, there are 4 users who access the first link in the sub-time period, namely: a, b, c, d, and the pairing results are: ab, ac, ad, bc, bd, cd. When there are m users who access the first link in a sub-time period, the number of paired user pairs is: 1+2+3+...+(m-1)=m*(m-1) / 2.

[0078] In the above embodiment, for each user pair, the number of days on which the two users in the user pair simultaneously access the first link in the current sub-time period is counted, which is regarded as the first type of days, and the number of days on which only one of the two users in the user pair accesses the first link in the current sub-time period is counted, which is set as the second type of days; the first type of days and the second type of days are used as the click behavior features of the two users in the user pair in the current sub-time period, thereby realizing the extraction of user click behavior features.

[0079] Considering that: when the number of days that two users have not visited the first link at the same time is relatively large, if the click behavior similarity algorithm used is inappropriate, the click behavior similarity of the two users calculated in this case will be high, which is inconsistent with the actual situation. In view of this situation, the embodiment of the present invention provides the following click behavior similarity algorithm: In an optional embodiment, in step 105, determining whether there is clustering behavior between users includes: for each user pair that visits the first link in each sub-time period, calculating the sum of the first category days and the second category days corresponding to the user pair, and then dividing the first category days by the sum to obtain an initial value of the click behavior similarity between the two users in the user pair; dividing the first category days by the total number of days in the current sub-time period to obtain a weight; multiplying the weight by the initial value of the click behavior similarity to obtain the click behavior similarity between the two users in the user pair, and if the click behavior similarity is greater than the preset similarity threshold, it is determined that there is clustering behavior between the two users in the user pair;

[0080] Furthermore, in step 105, adding users with grouping behavior to the traffic cheating group includes: adding two users in the user pair to the traffic cheating group in the current sub-time period.

[0081] In the above embodiment, by using the quotient obtained by dividing the first category of days by the total number of days in the current sub-time period as a weight, the occurrence of a situation where two users in a user pair have not visited the first link for a large number of days, resulting in a high similarity in the click behaviors of the two users, is reduced, thereby ultimately reducing the misjudgment of group traffic cheating behavior.

[0082] For example, there are four users who visit the first link in a sub-time period, namely a, b, c, and d. The user pairs are ab, ac, ad, bc, bd, and cd. If the similarity of the click behaviors of users ab and bd is greater than the preset similarity threshold, users a, b, and d are added to the traffic cheating group in the sub-time period.

[0083] In an optional embodiment, after adding the two users in the user pair to the traffic cheating gang set in the current sub-time period, it further includes: when obtaining the traffic cheating gang sets of all sub-time periods in the first time period, selecting users who have appeared in each traffic cheating gang set; deleting the users who have appeared in each traffic cheating gang set from each traffic cheating gang set respectively, to obtain an updated traffic cheating gang set for each sub-time period; for each user who has appeared in each traffic cheating gang set, calculating the click behavior similarity between the user and each traffic cheating gang set, selecting the traffic cheating gang set with the greatest click behavior similarity, and adding the user to the selected traffic cheating gang set.

[0084] In the above embodiment, for users who appear repeatedly in all traffic fraud gang sets, they are finally divided into a traffic fraud gang set based on the similarity of their click behavior with each traffic fraud gang set, thereby further improving the identification accuracy of the traffic fraud gang and facilitating the management of the traffic fraud gang.

[0085] In an optional embodiment, the click behavior similarity between the user and each traffic cheating gang set is calculated, including: for each traffic cheating gang set, respectively calculating the click behavior similarity between the user and each user in the traffic cheating gang set, and selecting the smallest click behavior similarity as the click behavior similarity between the user and the traffic cheating gang set.

[0086] The above embodiment provides a specific solution for determining the similarity between the click behaviors of a user and a set of traffic fraud gangs.

[0087] Figure 2 This is an application example of the present invention. In this example, the first time period is divided into 4 sub-time periods, each sub-time period is 4 days, and the situation of users a~m accessing the first link every day in the first time period is as follows: Figure 2 As shown, a value of 1 indicates that the user has visited the first link on that day, and a value of 0 indicates that the user has not visited the first link on that day.

[0088] In each sub-time period, the click behavior similarity between two users is calculated respectively, and the users whose click behavior similarity is greater than the preset similarity threshold are added to the traffic cheating group set of the corresponding sub-time period, and the result is:

[0089] The set of traffic cheating groups in sub-time period 1: X1 = {a, b, f, h, k};

[0090] The set of traffic cheating groups in sub-time period 2: X2 = {a, b, d, e, h};

[0091] The set of traffic cheating groups in sub-time period 3: X3 = {a, b, i, l};

[0092] The set of traffic cheating groups in sub-time period 4: X4 = {a, b, g, j, l}.

[0093] Then: the users who have appeared in each traffic fraud group are: Φ=X1∩X2∩X3∩X4={a,b};

[0094] Then update X1=X1-Φ={f,h,k}, X2=X2-Φ={d,e,h}, X3=X3-Φ={i,l}, X4=X4-Φ={g,j,l}.

[0095] For a, the click behavior similarities between a and X1, X2, X3, and X4 are calculated respectively, and Xn (n=1, 2, 3, or 4) corresponding to the largest click behavior similarity is selected as the traffic fraud group set that a belongs to. For example, when calculating the click behavior similarity between a and X1, the click behavior similarities between a and f, h, and k in X1 are calculated respectively, and the smallest click behavior similarity is selected as the click behavior similarity between a and X1.

[0096] Using the same method as a, select one of X1, X2, X3, and X4 as the traffic cheating group set to which b belongs.

[0097] Considering that the length of the sub-time period may affect the recognition accuracy of group traffic cheating behavior, the embodiment of the present invention provides the following optimization solution for this situation:

[0098] In an optional embodiment, dividing the first time period into at least one sub-time period includes: dividing the first time period into at least one sub-time period according to a preset initial length of the sub-time period, or according to the length of the current sub-time period and the adjustment step size;

[0099] Moreover, after adding the two users in the user pair to the traffic cheating gang set in the current sub-time period, it further includes: after obtaining the traffic cheating gang set of all current sub-time periods, judging whether the length of the current sub-time period reaches the maximum length of the preset sub-time period; if not, returning to the action of dividing the first time period into at least one sub-time period according to the length of the current sub-time period and the adjustment step; if so, according to the principle that the more members in the detected traffic cheating gang set, the better the detection result, in the at least one traffic cheating gang set detected each time, selecting the optimal at least one traffic cheating gang set as the final detection result.

[0100] In the above embodiment, multiple groups of traffic cheating gang sets are calculated by changing the length of the sub-time period (wherein, a corresponding group of traffic cheating gang sets will be calculated each time the sub-time period is divided), and based on the principle that the more members in the detected traffic cheating gang set, the better the detection result, a group of optimal traffic cheating gang sets are selected from the multiple groups of detected traffic cheating gang sets as the final detection result, thereby improving the recognition accuracy of gang traffic cheating behavior.

[0101] In an optional embodiment, dividing the first time period into at least one sub-time period includes: dividing the first time period into a plurality of sub-time periods, and two adjacent sub-time periods overlap by a preset second time length.

[0102] Considering that normal users will not continuously access the same link for a long period of time, in order to reduce the workload of identifying group traffic cheating and speed up the identification of group traffic cheating, the embodiment of the present invention provides the following optimization scheme:

[0103] In an optional embodiment, after step 103 and before step 104, it further includes: for each user, based on the user click data of the user, counting the maximum duration of the user's continuous access to the first link within the first time period; if the maximum duration is less than the preset first duration, it is determined that the user is not a member of the traffic cheating gang, and the user click data of the user is deleted from the user click data of each user who accessed the first link within the first time period.

[0104] In an optional embodiment, in step 105, after adding users with grouping behavior to the traffic cheating group set, it further includes: for each user with grouping behavior, determining whether the user meets the following conditions, if so, deleting the user from the traffic cheating group set: the user's stay time on the web page of the first link is greater than a preset time threshold, or / and, the user's visit depth to the first link is greater than a preset depth threshold, or / and, the number of item types purchased by the user on the first link is greater than a preset number.

[0105] The above embodiment further confirms whether the user is a member of a traffic cheating gang by analyzing the length of time the user stays on the web page of the first link, and / or the depth of the user's visit to the first link, and / or the type of items purchased by the user on the first link, thereby avoiding misjudgment of traffic cheating behavior and improving the accuracy of identifying traffic cheating behavior.

[0106] Figure 3 A flow chart of a method for identifying traffic fraud behavior provided by another embodiment of the present invention, wherein the specific steps are as follows:

[0107] Step 301: Obtain user click data of users visiting web pages.

[0108] In actual applications, user click data of users visiting web pages can be obtained from the webmaster group statistics table, the wake-up UUID allocation table, the device browsing information table, etc.

[0109] The user click data at least includes: user identification information, the web page link identification clicked by the user and the user click time, the web page link identification is such as: URL (Uniform Resource Locator, Uniform Resource Locator).

[0110] User identification information such as: user ID, user device ID, browser ID, login account, or any combination thereof.

[0111] The user click data may also include: user transaction information such as: the type of items purchased by the user, the user's GMV (Gross Merchandise Volume, total transaction amount), etc., or a combination thereof.

[0112] In practical applications, if there is missing data in a piece of information in the user click data, the missing value can be filled by using methods such as the nearest average, Bayesian formalization method, or decision tree induction. In addition, noise filtering can also be performed on the information in the user click data, for example: filtering according to the user click time, or filtering out the empty value data of the webmaster ID.

[0113] Step 302: Select a first time period, extract the user click data of users who accessed the first link within the first time period from the user click data obtained in step 301, and set the initial window length, initial sliding step, window length adjustment step, sliding step adjustment step, maximum window length and maximum sliding step of the sliding time window.

[0114] In practical applications, for example, in a traffic activation scenario, all users to be activated can be assigned to multiple traffic activation executors. Each traffic activation executor activates users by placing links of interest to users on one or more web pages to attract users to click on the links. For this scenario, the first link can be a link placed by a traffic activation executor.

[0115] In practical applications, it is considered that normal users will not continuously access the same link for a long time. Therefore, in step 302, after extracting the click data of users who accessed the first link within the first time period, the click data of users whose maximum duration of continuous access to the first link is less than the first duration are filtered out.

[0116] For example: assuming the first time period is 30 days, then based on the click data of all users who access the first link within the first time period, for each user, query the maximum time each user continuously accesses the first link within the first time period. For a certain user, if the maximum time his or her continuous access to the first link does not exceed 30 days, it is considered that the user is definitely not a member of the traffic cheating gang, and the user's click data is filtered out and does not participate in the subsequent process.

[0117] Step 303: according to the current window length and the current sliding step length, the sliding time window is sequentially slid over the first time period.

[0118] In actual applications, if no user clicks on the first link in the first q (q is a positive integer) days within the first time period, the sliding time window may start from the q+1 day. Figure 4As shown, assuming that the first time period is the entire year of 2020, and no user clicks on the first link in the first 58 days of 2020, the sliding time window starts from the 59th day, where Figure 4 In the example, the window length of each time window is 120 days and the sliding step length is 10 days.

[0119] Step 304: For each time window, among all the user click data that accessed the first link within the first time period, select all the user click data that accessed the first link within the time window.

[0120] Step 305: Pair all users who access the first link within the time window completely. For each user pair, calculate the similarity of the click behaviors of the two users based on the number of days that the two users in the user pair simultaneously access the first link within the time window and the number of days that only one user accesses the first link.

[0121] Here, pairing all users who access the first link within the time window means that for any user a, user a is paired with each other user. For example, there are 4 users who access the first link within the time window, namely a, b, c, and d. The pairing results are ab, ac, ad, bc, bd, and cd. When there are m users who access the first link within a time window, the number of paired user pairs is 1+2+3+...+(m-1)=m*(m-1) / 2.

[0122] The calculation formula for click behavior similarity is as follows:

[0123]

[0124] Among them, δ is the click behavior similarity; n is the number of days in each time window; k is the kth day in the current time window;

[0125] α k The meaning is: if on the kth day, two users in the current user pair visit the first link at the same time, then α k =1, otherwise, α k =0;

[0126] β k The meaning is: if on the kth day, there is only one user in the current user pair who has visited the first link (that is, either user a or user b has visited), then β k =1, otherwise, β k =0.

[0127] That is, Refers to the total number of days that two users in the current user pair simultaneously accessed the first link in the current time window;

[0128] Refers to the total number of days that one of the two users in the current user pair accesses the first link within the current time window.

[0129] For example, if n=120 days, within a certain time window of 120 days, the number of days that users a and b in the user pair simultaneously accessed the first link was 80 days in total, the number of days that they did not access the first link was 10 days in total, and the number of days that only one user accessed the first link (that is, either user a or user b) was 30 days in total. Then the similarity of the click behaviors of users a and b is:

[0130]

[0131] It should be noted that in formula (1) It is essentially a weight, and its function is to reduce the high similarity of the click behaviors of two users in a user pair when the two users have not visited the first link for a long time.

[0132] Step 306: According to the click behavior similarity of each user pair in the time window, select each user pair whose click behavior similarity is greater than a preset similarity threshold, and add all users in each selected user pair to the traffic cheating gang set of the time window.

[0133] For example, there are four users who access the first link within the time window, namely a, b, c, and d. The user pairs are ab, ac, ad, bc, bd, and cd. If the similarity of the click behaviors of users ab and bd is greater than the preset similarity threshold, users a, b, and d are added to the traffic cheating group of the time window.

[0134] Step 307: According to the traffic cheating gang set of each time window in the first time period, select a user who has appeared in the traffic cheating gang set in each time window.

[0135] Step 308: Delete the users who have appeared in the traffic cheating group set in each time window from the traffic cheating group set in each time window, and obtain an updated traffic cheating group set in each time window.

[0136] Step 309: For each user who has appeared in the traffic cheating group set in each time window, calculate the click behavior similarity between the user and the traffic cheating group set in each time window, select the largest click behavior similarity among them, and add the user to the traffic cheating group set corresponding to the largest click behavior similarity.

[0137] Among them, the specific method of calculating the click behavior similarity between the user and the traffic cheating gang set in each time window is as follows: for the traffic cheating gang set in each time window, the click behavior similarity between the user and each user in the traffic cheating gang set is calculated respectively, and the smallest click behavior similarity is selected as the click behavior similarity between the user and the traffic cheating gang set.

[0138] Step 310: Determine whether the window length and sliding step length of each current time window have reached the maximum window length and maximum sliding step length respectively. If so, execute step 312; otherwise, execute step 311.

[0139] Step 311: adjust the window length and / or sliding step length of the sliding time window according to the preset window length adjustment step length and / or sliding step length adjustment step length, and return to step 303.

[0140] Step 312: According to the principle that the more traffic cheating gang members are detected in each time window, the better the detection result is, an optimal set of traffic cheating gangs is selected from the multiple sets of traffic cheating gangs detected, and the selected optimal set of traffic cheating gangs is used as the final detection result.

[0141] Among them, a set of traffic cheating gangs will be calculated for each set of window length + sliding step size.

[0142] In practical applications, the embodiments of the present invention can be executed periodically to update the traffic cheating group set according to the latest user click data. The identification method can be based on real-time traffic click behavior, thereby improving the identification accuracy. Moreover, for periodic group traffic cheating behavior, the period can be arbitrarily selected for identification, and the scope of application is wide.

[0143] Figure 5 A schematic diagram of the structure of a traffic fraud behavior identification device provided by an embodiment of the present invention, the device mainly includes:

[0144] The click behavior feature extraction module 51 is used to obtain user click data of users visiting web pages; extract user click data of users visiting the first link within a first time period from the user click data; extract user click data of each user from the user click data of users visiting the first link within the first time period; and extract the click behavior feature of each user from the user click data of each user.

[0145] The identification module 52 is used to determine whether there is clustering behavior among users based on the click behavior characteristics of each user extracted by the click behavior characteristic extraction module 51. If so, the users with clustering behavior are added to the traffic cheating group.

[0146] In an optional embodiment, the identification module 52 determines whether there is clustering behavior between users by calculating the click behavior similarity between two users based on the click behavior characteristics of each user, and if any click behavior similarity is greater than a preset similarity threshold, it is determined that clustering behavior exists between the two users corresponding to the click behavior similarity.

[0147] In an optional embodiment, the user click data acquired by the click behavior feature extraction module 51 includes: user identification information, the web page link identification clicked by the user, and the user click time;

[0148] The click behavior feature extraction module 51 extracts the click behavior feature of each user from the user click data of each user, including: dividing the first time period into at least one sub-time period; for each sub-time period, obtaining the user click data of each user who accesses the first link within the sub-time period; for each sub-time period, completely pairing all users who access the first link within the sub-time period; for each user pair who accesses the first link within each sub-time period, based on the user click data of the two users in the user pair in the current sub-time period, counting the number of days when the two users in the user pair simultaneously access the first link in the current sub-time period, which are regarded as the first type of days, and counting the number of days when one and only one of the two users in the user pair accesses the first link in the current sub-time period, which are regarded as the second type of days; the first type of days and the second type of days are used as the click behavior features of the two users in the user pair in the current sub-time period.

[0149] In an optional embodiment, the identification module 52 determines whether there is a clustering behavior between users, including: for each user pair that accesses the first link in each sub-time period, calculating the sum of the first-category days and the second-category days corresponding to the user pair, and then dividing the first-category days by the sum to obtain an initial value of the click behavior similarity between the two users in the user pair; dividing the first-category days by the total number of days in the current sub-time period to obtain a weight; multiplying the weight by the initial value of the click behavior similarity to obtain the click behavior similarity between the two users in the user pair, and if the click behavior similarity is greater than a preset similarity threshold, it is determined that there is a clustering behavior between the two users in the user pair;

[0150] The identification module 52 adds the users with grouping behavior to the traffic cheating group set, including: adding two users in the user pair to the traffic cheating group set in the current sub-time period.

[0151] In an optional embodiment, after the identification module 52 adds the two users in the user pair to the traffic cheating gang set in the current sub-time period, it further includes: when the traffic cheating gang sets of all sub-time periods in the first time period are obtained, select users who have appeared in each traffic cheating gang set; delete the users who have appeared in each traffic cheating gang set from each traffic cheating gang set respectively, and obtain the updated traffic cheating gang set of each sub-time period; for each user who has appeared in each traffic cheating gang set, calculate the similarity of the click behavior of the user with each traffic cheating gang set, select the traffic cheating gang set with the largest click behavior similarity, and add the user to the selected traffic cheating gang set.

[0152] In an optional embodiment, the identification module 52 calculates the click behavior similarity between the user and each traffic fraud gang set, including: for each traffic fraud gang set, respectively calculating the click behavior similarity between the user and each user in the traffic fraud gang set, and selecting the smallest click behavior similarity as the click behavior similarity between the user and the traffic fraud gang set.

[0153] In an optional embodiment, the click behavior feature extraction module 51 divides the first time period into at least one sub-time period, including: dividing the first time period into at least one sub-time period according to a preset initial length of the sub-time period, or according to the length of the current sub-time period and the adjustment step;

[0154] Moreover, after the identification module 52 adds the two users in the user pair to the traffic cheating gang set in the current sub-time period, it further includes: after obtaining the traffic cheating gang set of all current sub-time periods, judging whether the length of the current sub-time period reaches the maximum length of the preset sub-time period; if not, notifying the click behavior feature extraction module 51 to return the action of dividing the first time period into at least one sub-time period according to the length of the current sub-time period and the adjustment step; if so, according to the principle that the more members in the detected traffic cheating gang set, the better the detection result, the optimal at least one traffic cheating gang set is selected as the final detection result in the at least one traffic cheating gang set detected each time.

[0155] In an optional embodiment, the click behavior feature extraction module 51 divides the first time period into at least one sub-time period, including: dividing the first time period into multiple sub-time periods, and two adjacent sub-time periods overlap for a preset second time length.

[0156] In an optional embodiment, after the click behavior feature extraction module 51 extracts the user click data of each user from the user click data that accesses the first link within the first time period and before extracting the click behavior features of each user from the user click data of each user, it further includes: for each user, based on the user click data of the user, counting the maximum duration of the user's continuous access to the first link within the first time period; if the maximum duration is less than the preset first duration, it is determined that the user is not a member of the traffic cheating gang, and the user click data of the user is deleted from the user click data of each user who accesses the first link within the first time period.

[0157] In an optional embodiment, after the identification module 52 adds users with grouping behavior to the traffic fraud gang set, it further includes: for each user with grouping behavior, determining whether the user meets the following conditions, and if so, deleting the user from the traffic fraud gang set: the user's stay time on the web page of the first link is greater than a preset time threshold, or / and, the user's visit depth to the first link is greater than a preset depth threshold, or / and, the number of item types purchased by the user on the first link is greater than a preset number.

[0158] The embodiment of the present application also provides a computer-readable storage medium, which stores instructions, and the instructions, when executed by a processor, can execute the steps in the above-mentioned method for identifying traffic fraud behavior. In practical applications, the computer-readable medium can be included in each device / apparatus / system in the above-mentioned embodiments, or it can exist independently without being assembled into the device / apparatus / system. Among them, instructions are stored in a computer-readable storage medium, and the instructions stored therein can execute the steps in the above-mentioned method for identifying traffic fraud behavior when executed by a processor.

[0159] According to the embodiments disclosed in the present application, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, it may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof, but it is not used to limit the scope of protection of the present application. In the embodiments disclosed in the present application, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, apparatus, or device.

[0160] like Figure 6 As shown, an embodiment of the present invention also provides an electronic device. Figure 6 As shown, it shows a schematic diagram of the structure of an electronic device involved in an embodiment of the present invention, specifically:

[0161] The electronic device may include a processor 61 with one or more processing cores, a memory 62 with one or more computer-readable storage media, and a computer program stored in the memory and executable on the processor. When executing the program in the memory 62, the above-mentioned traffic fraud behavior identification method may be implemented.

[0162] Specifically, in practical applications, the electronic device may further include components such as a power supply 63, an input / output unit 64, etc. Those skilled in the art may understand that Figure 6 The structure of the electronic device shown in the figure does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0163] The processor 61 is the control center of the electronic device. It uses various interfaces and lines to connect various parts of the entire electronic device. It executes various functions of the server and processes data by running or executing software programs and / or modules stored in the memory 62 and calling data stored in the memory 62, thereby monitoring the electronic device as a whole.

[0164] The memory 62 can be used to store software programs and modules, that is, the above-mentioned computer-readable storage medium. The processor 61 executes various functional applications and data processing by running the software programs and modules stored in the memory 62. The memory 62 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function, etc.; the data storage area may store data created according to the use of the server, etc. In addition, the memory 62 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices. Accordingly, the memory 62 may also include a memory controller to provide the processor 61 with access to the memory 62.

[0165] The electronic device also includes a power supply 63 for supplying power to various components, which can be logically connected to the processor 61 through a power management system, so as to manage charging, discharging, power consumption and other functions through the power management system. The power supply 63 can also include any components such as one or more DC or AC power supplies, recharging systems, power failure detection circuits, power converters or inverters, power status indicators, etc.

[0166] The electronic device may also include an input and output unit 64, which may be used to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control. The input unit output 64 may also be used to display information input by the user or information provided to the user and various graphical user interfaces, which may be composed of graphics, text, icons, videos and any combination thereof.

[0167] The flow chart and block diagram in the accompanying drawings of the present application show the possible architecture, function and operation of the system, method and computer program product according to the various embodiments disclosed in the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the above-mentioned module, program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the box can also occur in the order of the standards in different figures. For example, the boxes represented by two connections can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0168] Those skilled in the art will appreciate that the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations or combinations are not explicitly described in the present application. In particular, without departing from the spirit and teachings of the present application, the features described in the various embodiments and / or claims of the present application may be combined and / or combined in a variety of ways, and all of these combinations and / or combinations fall within the scope disclosed in the present application.

[0169] Specific embodiments are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas, and is not used to limit the present application. For those skilled in the art, changes can be made in the specific implementation methods and application scopes according to the ideas, spirits and principles of the present invention, and any modifications, equivalent substitutions, improvements, etc. made therein should be included in the scope of protection of this application.

Claims

1. A method for identifying traffic fraud behavior, characterized in that: The method includes: Get user click data of web pages visited by users; Extracting user click data that accesses the first link within a first time period from the user click data; Extracting user click data of each user from the click data of users who accessed the first link during the first time period; Extracting click behavior features of each user from the user click data of each user; Based on the click behavior characteristics of each user, determine whether there is clustering behavior among users. If so, add the users with clustering behavior to the traffic fraud group. Extracting the click behavior feature of each user from the click data of each user includes: dividing the first time period into at least one sub-time period; For each sub-time period, obtaining user click data of each user who accesses the first link in the sub-time period; For each sub-time period, all users who access the first link in the sub-time period are completely paired with each other; For each user pair that accesses the first link in each sub-time period, based on the user click data of the two users in the user pair in the current sub-time period, the number of days that the two users in the user pair simultaneously access the first link in the current sub-time period is counted, which is regarded as the first type of days, and the number of days that only one of the two users in the user pair accesses the first link in the current sub-time period is counted, which is regarded as the second type of days; The first type of days and the second type of days are used as the click behavior features of the two users in the user pair in the current sub-time period.

2. The method according to claim 1, characterized in that Determining whether there is a grouping behavior among users includes: According to the click behavior characteristics of each user, the click behavior similarity between two users is calculated. If any click behavior similarity is greater than a preset similarity threshold, it is determined that clustering behavior exists between the two users corresponding to the click behavior similarity.

3. The method according to claim 1, characterized in that The user click data includes: user identification information, the web page link identification clicked by the user and the user click time.

4. The method according to claim 3, characterized in that Determining whether there is a grouping behavior among users includes: For each user pair that accesses the first link in each sub-time period, calculate the sum of the first-category days and the second-category days corresponding to the user pair, and then divide the first-category days by the sum to obtain an initial value of the click behavior similarity between the two users in the user pair; divide the first-category days by the total number of days in the current sub-time period to obtain a weight; multiply the weight by the initial value of the click behavior similarity to obtain the click behavior similarity between the two users in the user pair; if the click behavior similarity is greater than a preset similarity threshold, it is determined that there is a clustering behavior between the two users in the user pair; The aforementioned adding users with grouping behavior into the traffic fraud gang includes: The two users in the user pair are added to the traffic cheating group in the current sub-time period.

5. The method according to claim 4, characterized in that After adding the two users in the user pair to the traffic cheating gang in the current sub-time period, the method further includes: When the traffic cheating gang sets of all sub-time periods within the first time period are obtained, users who have appeared in each traffic cheating gang set are selected; Delete the users who have appeared in each traffic cheating gang set from each traffic cheating gang set, and obtain the updated traffic cheating gang set of each sub-time period; For each user who has appeared in each traffic fraud gang set, calculate the similarity of the user's click behavior with each traffic fraud gang set, select the traffic fraud gang set with the greatest click behavior similarity, and add the user to the selected traffic fraud gang set.

6. The method according to claim 5, characterized in that The calculation of the click behavior similarity between the user and each traffic fraud gang set includes: For each traffic fraud gang set, the click behavior similarity between the user and each user in the traffic fraud gang set is calculated respectively, and the smallest click behavior similarity is selected as the click behavior similarity between the user and the traffic fraud gang set.

7. The method according to claim 4, characterized in that The step of dividing the first time period into at least one sub-time period includes: Dividing the first time period into at least one sub-time period according to a preset initial length of the sub-time period, or according to the length of the current sub-time period and the adjustment step; Furthermore, after adding the two users in the user pair to the traffic cheating group in the current sub-time period, the method further includes: After obtaining the traffic cheating gang set of all current sub-time periods, determine whether the length of the current sub-time period reaches the maximum length of the preset sub-time period; If not, returning to the action of dividing the first time period into at least one sub-time period according to the length of the current sub-time period and the adjustment step; If so, according to the principle that the more members in the detected traffic cheating group set, the better the detection result, the best at least one traffic cheating group set is selected as the final detection result from the at least one traffic cheating group set detected each time.

8. The method according to claim 3, characterized in that The step of dividing the first time period into at least one sub-time period includes: The first time period is divided into a plurality of sub-time periods, and two adjacent sub-time periods overlap for a preset second time length.

9. The method according to claim 1, characterized in that: After extracting the user click data of each user from the click data of users who accessed the first link within the first time period and before extracting the click behavior feature of each user from the user click data of each user, the method further includes: For each user, based on the user click data of the user, the maximum duration of the user's continuous access to the first link within the first time period is counted; if the maximum duration is less than the preset first duration, it is determined that the user is not a member of the traffic cheating gang, and the user click data of the user is deleted from the user click data of each user who accessed the first link within the first time period.

10. The method according to claim 1, characterized in that After adding the users with grouping behavior into the traffic cheating gang, the method further includes: For each user who has grouped together, determine whether the user meets the following conditions. If so, delete the user from the traffic fraud group: The user's stay time on the webpage of the first link is greater than a preset time threshold, or / and, the user's access depth to the first link is greater than a preset depth threshold, or / and, the number of item types purchased by the user on the first link is greater than a preset number.

11. A device for identifying traffic fraud behavior, characterized in that: The device includes: A click behavior feature extraction module is used to obtain user click data of users visiting a webpage; extract user click data of users visiting a first link within a first time period from the user click data; extract user click data of each user from the user click data of users visiting the first link within the first time period; extract the click behavior feature of each user from the user click data of each user; The identification module is used to determine whether there is clustering behavior among users based on the click behavior characteristics of each user. If so, the users with clustering behavior will be added to the traffic fraud group. Click behavior feature extraction module, specifically used for: The first time period is divided into at least one sub-time period; for each sub-time period, the user click data of each user who accesses the first link within the sub-time period is obtained; for each sub-time period, all users who access the first link within the sub-time period are completely paired with each other; for each user pair who accesses the first link within each sub-time period, the number of days on which the two users in the user pair simultaneously access the first link within the current sub-time period is counted according to the user click data of the two users in the user pair within the current sub-time period, which are regarded as the first type of days, and the number of days on which only one of the two users in the user pair accesses the first link within the current sub-time period is counted, which are regarded as the second type of days; the first type of days and the second type of days are used as the click behavior characteristics of the two users in the user pair within the current sub-time period.

12. A non-transitory computer-readable storage medium storing instructions, characterized in that: When the instruction is executed by a processor, the processor executes the steps of the traffic fraud behavior identification method as described in any one of claims 1 to 10.

13. An electronic device, characterized in that: Includes the non-transitory computer-readable storage medium of claim 12, and the processor that can access the non-transitory computer-readable storage medium.