Method, device, electronic device and storage medium for identifying cheating traffic media

Abstract

The embodiments of the present application provide a method, device, electronic device and storage medium for identifying cheating traffic media, and relate to the field of network security technology. The method includes: obtaining the historical click logs of each traffic media; obtaining the click volume of each click position under each traffic media according to the historical click logs, obtaining the similarity of the click volume of each click position under any two traffic media in each traffic media, and calculating the overall similarity between the two traffic media according to the similarity of the click volume of each click position under any two traffic media, and constructing an undirected graph according to the overall similarity between any two traffic media; determining the maximum clique in the undirected graph, and identifying cheating traffic media according to the traffic media corresponding to the maximum clique. The embodiments of the present application can more accurately identify cheating traffic media, protect the interests of advertisers, and optimize the ecology and reputation of the mobile advertising alliance.

Description

Technical Field

[0001] The present application relates to the field of network security technology. Specifically, the present application relates to a method, device, electronic device and storage medium for identifying cheating traffic media. Background Art

[0002] The CPC (cost per click) advertising charging model is a model where advertisers only pay for users clicking on their ads, and no longer pay for ad exposure. For advertisers, CPC advertising avoids the risk of exposure without clicks, and is currently one of the mainstream advertising charging methods.

[0003] Traffic media is a media that can provide traffic and is also a carrier for displaying recommended information (such as advertisements). It usually includes media, websites, and applications. Traffic media can participate in the profit sharing of recommended information. Under the same exposure of recommended information, the higher the click-through rate, the higher the profit share. Driven by the profit of recommended information income, some traffic media will cheat to forge recommended information clicks and defraud advertisers of fees. Traffic media that use the same means or applications to cheat are called cheating traffic media gangs.

[0004] The existing method for identifying cheating traffic media gangs is mainly to calculate the user overlap between traffic media pairs, and to identify traffic media with particularly high user overlap as gangs. However, this method has some problems. For example, when the number of users controlled by the cheating traffic media gang is large, different users can be assigned to different traffic media, and the user overlap is not high. Therefore, the performance of the existing technology solution in detecting cheating traffic gangs is low. Summary of the invention

[0005] Embodiments of the present invention provide a method, device, electronic device, and storage medium for identifying cheating traffic media that overcome the above problems or at least partially solve the above problems.

[0006] In a first aspect, a method for identifying cheating traffic media is provided, the method comprising:

[0007] Obtain the historical click logs of each traffic media. The historical click logs include the click information generated by each click on the recommended information during the historical period, and the click location is recorded in the click information;

[0008] The click volume of each click position under each traffic media is obtained according to the historical click log, the similarity of the click volume of each click position under any two traffic media in each traffic media is obtained, and the overall similarity between any two traffic media is calculated according to the similarity of the click volume of each click position under any two traffic media;

[0009] According to the overall similarity between any two traffic media, an undirected graph is constructed, wherein the vertices of the undirected graph are traffic media, and when the overall similarity between any two traffic media is greater than a second predetermined threshold, two vertices corresponding to any two traffic media are connected by an edge;

[0010] Determine the maximal clique in the undirected graph, and identify the cheating traffic media based on the traffic media corresponding to the maximal clique.

[0011] In a possible implementation, the click information also includes a unique identifier of the recommended information;

[0012] Identify cheating traffic media by using the traffic media corresponding to the largest cluster, including:

[0013] Obtain the click volume of at least two pieces of recommended information at each click position under the traffic media according to the historical click log of the traffic media;

[0014] The traffic media corresponding to the maximum cluster is regarded as the suspected cheating traffic media, and the similarity between the click positions of the at least two recommended information in the suspected cheating traffic media and the click positions of the at least two recommended information in all the suspected cheating traffic media is obtained according to the click volume of each click position of the at least two recommended information in all the suspected cheating traffic media;

[0015] For any suspected cheating traffic media, if the similarity between the click positions of at least two recommended information under the suspected cheating traffic media and the click positions of at least two recommended information under all suspected cheating traffic media is less than a first preset threshold, the suspected cheating traffic media is determined to be cheating traffic media.

[0016] In a possible implementation, obtaining the similarity of the click volume of each click position under any two traffic media in each traffic media includes:

[0017] The click volume of each click position under the traffic media is used as the element value of the preset dimension in the vector corresponding to the traffic media;

[0018] The distance between any two vectors is calculated as the similarity of the click volume of the traffic media corresponding to the two vectors at each click position.

[0019] In a possible implementation, obtaining the similarity of the click positions of at least two pieces of recommendation information in suspected cheating traffic media according to the click volume of each click position of at least two pieces of recommendation information in all traffic media includes:

[0020] For any one of the at least two pieces of recommended information, count the number of clicks on each click position of the recommended information in the suspected cheating traffic media and the total number of clicks on each click position of the recommended information in all traffic media;

[0021] Obtain the similarity between the number of clicks at each click position of the recommended information in the suspected cheating traffic media and the total number of clicks at each click position of the recommended information in all traffic media, as the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media;

[0022] For any one of the at least two recommendation information, weighted sum is performed on the position similarity of the recommendation information in the suspected cheating traffic media relative to all traffic media according to the number of clicks on the recommendation information in the suspected cheating traffic media, so as to obtain the similarity between the click positions of the at least two recommendation information in the suspected cheating traffic media and the click positions of the at least two recommendation information in all suspected cheating traffic media.

[0023] In a possible implementation, obtaining the similarity between the number of clicks at each click position of the recommended information in the suspected cheating traffic media and the total number of clicks at each click position of the recommended information includes:

[0024] The number of clicks on each click position of the recommended information in the suspected cheating traffic media is used as the element value of the preset dimension in the first vector; the total number of clicks on the recommended information at each click position is used as the element value of the preset dimension in the second vector;

[0025] The distance between the first vector and the second vector is calculated as the similarity between the click volume of the recommended information at each click position in the suspected cheating traffic media and the total click volume of the recommended information at each click position.

[0026] In a possible implementation, weighted summing of the position similarities of the recommended information in the suspected cheating traffic media relative to all traffic media is performed according to the number of clicks on the recommended information in the suspected cheating traffic media, and obtaining the similarities between the click positions of at least two recommended information in the suspected cheating traffic media and the click positions of at least two recommended information in all suspected cheating traffic media includes:

[0027] The weight of the recommendation information is determined based on the ratio of the number of clicks on the recommended information in the suspected cheating traffic media to the number of clicks on at least two recommended information in the suspected cheating traffic media;

[0028] The weighted result of the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media is obtained by multiplying the weight value of the recommended information by the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media;

[0029] The weighted results of the position similarities of all recommended information in suspected cheating traffic media relative to all traffic media are added together to obtain the similarity between the click positions of at least two recommended information in suspected cheating traffic media and the click positions of at least two recommended information in all suspected cheating traffic media.

[0030] In a possible implementation, the click information also includes a unique identifier of the traffic media;

[0031] According to the traffic media corresponding to the largest group, the cheating traffic media is identified, and then it also includes:

[0032] A blacklist is constructed based on the unique identifier of the cheating traffic media; the click information collected by each traffic media in real time is obtained. If the unique identifier of the traffic media recorded in the click information is in the blacklist, it is determined that the click information is a cheating click, and the cheating click is filtered out.

[0033] In a second aspect, a device for identifying cheating traffic media is provided, including:

[0034] The click information acquisition module is used to obtain the historical click logs of each traffic media. The historical click logs include the click information generated by each click on the recommended information in the historical period, and the click position is recorded in the click information;

[0035] The overall similarity calculation module is used to obtain the click volume of each click position under each traffic media according to the historical click log, obtain the similarity of the click volume of each click position under any two traffic media in each traffic media, and calculate the overall similarity between any two traffic media according to the similarity of the click volume of each click position under any two traffic media;

[0036] An undirected graph construction module, used to construct an undirected graph according to the overall similarity between any two traffic media, wherein the vertices of the undirected graph are traffic media, and when the overall similarity between any two traffic media is greater than a second predetermined threshold, two vertices corresponding to any two traffic media are connected by an edge;

[0037] The cheating traffic media identification module is used to determine the maximum cluster in the undirected graph and identify the cheating traffic media according to the traffic media corresponding to the maximum cluster.

[0038] In a possible implementation, the click information also includes a unique identifier of the recommended information;

[0039] The fraud traffic media identification module includes:

[0040] A first click volume statistics unit is used to obtain the click volume of at least two pieces of recommended information at each click position in the traffic media according to the historical click log of the traffic media;

[0041] The recommendation information relevance determination unit is used to classify the traffic media corresponding to the maximum cluster as suspected cheating traffic media, and obtain the similarity between the click positions of the at least two recommended information in the suspected cheating traffic media and the click positions of the at least two recommended information in all suspected cheating traffic media according to the click volume of each click position of the at least two recommended information in all suspected cheating traffic media;

[0042] The screening unit is used to determine, for any suspected cheating traffic media in each traffic media, the suspected cheating traffic media as cheating traffic media if the similarity between the click positions of at least two recommended information under the suspected cheating traffic media and the click positions of at least two recommended information under all suspected cheating traffic media is less than a first preset threshold.

[0043] In a possible implementation, the overall similarity calculation module includes:

[0044] A first vector generation submodule is used to use the click volume of each click position under the traffic media as an element value of a preset dimension in a vector corresponding to the traffic media;

[0045] The first vector distance submodule is used to calculate the distance between any two vectors as the similarity of the click volume of the traffic media corresponding to the any two vectors at each click position.

[0046] In a possible implementation, the recommendation information relevance determination unit includes:

[0047] A click volume statistics subunit is used to count the click volume of each click position of the recommended information in the suspected cheating traffic media and the total click volume of each click position of the recommended information in all traffic media for any one of the at least two recommended information;

[0048] The position similarity calculation subunit is used to obtain the similarity between the click volume of each click position of the recommended information in the suspected cheating traffic media and the total click volume of each click position of the recommended information in all traffic media, as the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media;

[0049] The recommendation information relevance determination subunit is used to perform weighted summation of the position similarities of each recommendation information in the suspected cheating traffic media relative to all traffic media for any one of the at least two recommendation information according to the number of clicks on the recommended information in the suspected cheating traffic media, so as to obtain the similarity between the click positions of the at least two recommendation information in the suspected cheating traffic media and the click positions of the at least two recommendation information in all suspected cheating traffic media.

[0050] In a possible implementation, the location similarity calculation subunit includes:

[0051] The second vector generation submodule is used to use the number of clicks of the recommended information at each click position in the suspected cheating traffic media as the element value of the preset dimension in the first vector; and use the total number of clicks of the recommended information at each click position as the element value of the preset dimension in the second vector;

[0052] The second vector distance submodule is used to calculate the distance between the first vector and the second vector as the similarity between the click volume of the recommended information at each click position in the suspected cheating traffic media and the total click volume of the recommended information at each click position.

[0053] In a possible implementation, the recommendation information relevance determination subunit includes:

[0054] A weight determination submodule, used to use the ratio of the number of clicks on the recommended information in the suspected cheating traffic media to the number of clicks on at least two recommended information in the suspected cheating traffic media as the weight value of the recommended information;

[0055] A weighting submodule, for multiplying the weight value of the recommended information by the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media, to obtain a weighted result of the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media;

[0056] The weighted summation submodule is used to add the weighted results of the position similarity of at least two recommended information under suspected cheating traffic media relative to all traffic media, so as to obtain the similarity between the click positions of at least two recommended information under suspected cheating traffic media and the click positions of at least two recommended information under all suspected cheating traffic media.

[0057] In a possible implementation, the click information also includes a unique identifier of the traffic media;

[0058] The device for identifying cheating traffic media also includes:

[0059] The real-time judgment module is used to build a blacklist based on the unique identifier of the cheating traffic media; obtain the click information collected by each traffic media in real time. If the unique identifier of the traffic media recorded in the click information is in the blacklist, it is determined that the click information is a cheating click and the cheating click is filtered out.

[0060] In a third aspect, an embodiment of the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the method provided in the first aspect are implemented.

[0061] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method provided in the first aspect.

[0062] The method, device, electronic device and storage medium for identifying cheating traffic media provided by the embodiments of the present invention obtain the historical click logs of each traffic media, and the historical click logs record the click position of each click on the recommended information. This utilizes the characteristic that cheating traffic media does not care about the specific content of the recommended information when forging the click position, but has a small difference in click positions. Compared with the prior art method of utilizing user information and recommended information features, this method has a higher accuracy rate. By obtaining the click volume of each click position under each traffic media, calculating the similarity of the click volume of each click position under two types of traffic media, constructing an undirected graph, and determining the maximum clique in the undirected graph to identify cheating traffic media, cheating traffic media that use the same cheating means in the advertising alliance can be identified, thereby avoiding invalid recommendation information delivery, saving the delivery resources of the recommendation information delivery system, and saving the recommendation information delivery costs of advertisers. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in describing the embodiments of the present application are briefly introduced below.

[0064] Figure 1 An application scenario diagram provided for an embodiment of the present application;

[0065] Figure 2 A flowchart of a method for identifying cheating traffic media provided in an embodiment of the present application;

[0066] Figure 3 A distribution diagram of click positions of different traffic media for a recommended information instance provided in an embodiment of the present application;

[0067] Figure 4 A logic diagram of a method for identifying cheating traffic media according to another embodiment of the present application;

[0068] Figure 5 A schematic diagram of a process for identifying cheating traffic media through a large group provided in an embodiment of the present application;

[0069] Figure 6 A schematic diagram of an undirected graph provided in an embodiment of the present application;

[0070] Figure 7 A schematic diagram of a process for obtaining similarities between click positions of at least two pieces of recommendation information in suspected cheating traffic media and click positions of at least two pieces of recommendation information in all suspected cheating traffic media provided by an embodiment of the present application;

[0071] Figure 8 This is a flowchart of identifying cheating traffic media according to another embodiment of the present application;

[0072] Fig. 9 A schematic diagram of the structure of a device for identifying cheating traffic media provided in an embodiment of the present application;

[0073] Fig.10 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0074] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present invention.

[0075] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "an" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be an intermediate element. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0076] The method, device, electronic device and storage medium for identifying cheating traffic media provided in this application are intended to solve the above technical problems in the prior art.

[0077] Figure 1 The application scenario diagram provided by the embodiment of the present application is as follows: Figure 1 As shown, the mobile advertising alliance 11 is an alliance composed of multiple traffic media. The mobile advertising alliance in the figure has two traffic media. It should be noted that the number of traffic media in the embodiment of the present application is not specifically limited. It is set to two traffic media here for the convenience of description. The background servers of the two traffic media are traffic media server 111 and traffic media server 112. When the terminal 21 accesses the traffic media server, the traffic media server provides traffic services. The traffic media server is connected to the cheating traffic identification server 31 through the network. The cheating traffic identification server 31 is used to execute the method of identifying cheating traffic media in the following embodiments.

[0078] The traffic media server and the cheating traffic identification server can be independent physical servers, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.

[0079] The execution method of the server of the embodiment of the present application can be completed in the form of cloud computing. Cloud computing is a computing model that distributes computing tasks on a resource pool composed of a large number of computers, so that various application systems can obtain computing power, storage space and information services as needed. The network that provides resources is called a "cloud". The resources in the "cloud" are infinitely expandable in the eyes of users, and can be obtained at any time, used on demand, expanded at any time, and paid for by use.

[0080] As a provider of basic cloud computing capabilities, a cloud computing resource pool (referred to as a cloud platform, generally referred to as an IaaS (Infrastructure as a Service) platform) will be established, and various types of virtual resources will be deployed in the resource pool for external customers to choose to use. The cloud computing resource pool mainly includes: computing devices (virtualized machines, including operating systems), storage devices, and network devices.

[0081] According to the logical function division, the PaaS (Platform as a Service) layer can be deployed on the IaaS (Infrastructure as a Service) layer, and the SaaS (Software as a Service) layer can be deployed on the PaaS layer. SaaS can also be deployed directly on IaaS. PaaS is a platform for software operation, such as databases, web containers, etc. SaaS is a variety of business software, such as web portals, SMS mass senders, etc. Generally speaking, SaaS and PaaS are upper layers relative to IaaS.

[0082] 1. Advertiser 41 places advertisements in mobile advertising alliance 11.

[0083] 2. The traffic media server displays the recommended information. When a user (may be a real person or a computer program) accesses the traffic media server through a terminal, if the user clicks on the recommended information displayed on the traffic media, the traffic media server collects the information of this click operation in real time.

[0084] 3. The traffic media server generates click information and sends it to the cheating traffic identification server 31. The click information records the click position of the recommended information.

[0085] 4. The cheating traffic identification server 31 receives click information sent by traffic media within a certain period of time, and forms a historical click log. The historical click log records the click position of each click on the recommended information within a certain period of time, and identifies the cheating traffic media in the mobile advertising alliance 11 through the historical click log. The cheating traffic identification server 31 can inform the mobile advertising alliance 11 of the cheating traffic media so that the mobile advertising alliance removes the cheating mobile platform, and can also determine the traffic media to which the subsequent real-time click information belongs. If the click information belongs to the cheating traffic media, the click information is filtered out, and if the click information does not belong to the cheating traffic media, the click information is sent to the mobile advertising alliance 11.

[0086] 5. When the recommendation information contract period expires, the mobile advertising alliance 11 sends the number of clicks on each (normal) traffic media during the contract period to the advertiser 41, and the advertiser 41 pays the recommendation information fee to each traffic media based on the number of clicks.

[0087] Furthermore, since the click information received by the advertiser 41 has excluded the data of cheating traffic media, the conversion analysis of recommended information based on the click information by the advertiser 41 will be more accurate, and the recommended information can be accurately delivered by comparing the conversion effects of recommended information of each traffic media.

[0088] Optionally, the cheating traffic identification server in the above embodiment may also be used as a module in the mobile advertising alliance 11.

[0089] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0090] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0091] Figure 2 A flow chart of a method for identifying cheating traffic media provided in an embodiment of the present application is shown as follows: Figure 2 As shown, including:

[0092] S101. Obtain historical click logs of each traffic media.

[0093] The historical click log includes the click information generated by each click on the recommended information within the historical period, and the click location is recorded in the click information.

[0094] Optionally, the traffic media in the embodiment of the present application may be electronic media, websites, and applications.

[0095] Optionally, the recommendation information in the embodiment of the present application may be a static advertisement, such as a picture, or a dynamic advertisement, such as a gif picture, a short video, and the like.

[0096] Optionally, the length of the historical period in the embodiment of the present application may be one day, one week, one month, etc.

[0097] Traffic media that implement CPC billing model will count click information. For normal traffic media, when traffic media displays recommended information, generally only when real people click on the recommended information displayed on the traffic media, the traffic media will generate click information.

[0098] The specific information to be collected needs to be determined in advance. For existing traffic media, user information or recommended information features are usually collected, so as to judge cheating traffic media based on whether the user groups overlap or whether the recommended information features are clustered. However, in fact, both methods have the above defects.

[0099] In order to overcome the problem of inaccurate basis for selecting and identifying cheating traffic media in the prior art, the embodiment of the present application found through research that real users will click on different positions when clicking on recommended information due to differences in palm size and different postures of holding the terminal. For example, if there is a button in the recommended information, the click position is often concentrated near the button. Since the cheating traffic media is a forged click position, it has no perception of the specific content of the recommended information, so there is no difference in the click position of the recommended information. Therefore, the embodiment of the present application chooses to use the click position of each click as the basis for identifying cheating traffic media.

[0100] It is understandable that different recommended information may have different sizes. For example, some recommended information is displayed in full screen, while some recommended information is only displayed in part of the screen. In addition, even if the same recommended information is displayed in display areas of different sizes, the display size may be different. Therefore, in order to uniformly process the click position of the recommended information, the embodiment of the present application first determines the coordinates of the click position in the display area, and then converts the above coordinates into the standard display area according to the ratio of the display area to the preset standard display area to obtain the click position of the recommended information.

[0101] For example, if the display area of ​​a terminal on which the traffic media runs is 1000*2000, and the display range of the recommended information on the display area is an area consisting of four coordinates (0,0), (0,1024), (330,0) and (330,1024). If the user clicks at the coordinate (100,100), it can be determined that the user clicked on the recommended information, and the traffic media converts the coordinates into the click position. Of course, it can also be performed by the execution subject of the embodiment of the present application. The embodiment of the present application does not make specific limitations on this.

[0102] For example, if the display area of ​​the terminal is 1000*2500, the coordinates are converted to (100*1000 / 1000, 100*2500 / 2000) or (100, 125) according to the ratio.

[0103] The embodiment of the present application uniformly converts the actual location of the user's click into an area of ​​a standard size, so that the subsequent judgment of cheating traffic based on the click location is more accurate.

[0104] S102. Obtain the click volume of each click position under the traffic media according to the historical click log, obtain the similarity of the click volume of each click position under any two traffic media in each traffic media, and calculate the overall similarity between any two traffic media according to the similarity of the click volume of each click position under any two traffic media.

[0105] Since the click position of each click on the recommended information is recorded in the historical click log, the click volume of each click position can be counted. For example, if the click positions of 3 clicks are recorded in the historical click log, which are position 1, position 2 and position 1 respectively, then it can be known that the click volume of position 1 is 2 and the click volume of position 2 is 1. After obtaining the click volume of each click position under each traffic media, the similarity of the click volume of each click position under any two traffic media can be calculated. The embodiment of the present application does not specifically limit the method for calculating the overall similarity. For example, for each traffic media, each click position can be sorted according to the size of the click volume, and then the similarity of the sorting result can be used as the overall similarity.

[0106] S103. Construct an undirected graph according to the overall similarity between any two traffic media, wherein the vertices of the undirected graph are traffic media, and when the overall similarity between any two traffic media is greater than a second predetermined threshold, the two vertices corresponding to the any two traffic media are connected by an edge.

[0107] S104. Determine the maximum clique in the undirected graph, and identify cheating traffic media based on the traffic media corresponding to the maximum clique.

[0108] After obtaining the overall similarity between any two traffic media, the embodiment of the present application uses a maximal clique mining algorithm to identify cheating traffic media. A clique is a complete subgraph of an undirected graph. A complete subgraph means that each pair of vertices in the subgraph is connected by an edge. If a clique is not included in any other clique, that is, it is not a true subset of any other clique, then the clique is called a maximal clique of an undirected graph, which is an independent complete subset in the undirected graph.

[0109] The maximum clique mining algorithm of the embodiment of the present application may be the Bron–Kerbosch algorithm, which is a recursive backtracking algorithm. The algorithm mainly constructs three sets, assuming that:

[0110] The R set records the points that have been added to the current maximal clique.

[0111] The P set records the points that may be added (that is, the points that may have edges with all the points in the R set).

[0112] The X set records the points that have completed the maximum group counting (the function is to determine the duplication)

[0113] P∪X is the set of all points that may form a maximal cluster with the R set (although it is known that the points in X can no longer participate in the formation of a maximal cluster), that is, a part of the points connected to the last point added to the R set.

[0114] The basic Born_Kerbosch algorithm adds v to the set R for each point v in the set P, and searches for the next point that may be added to the set R in the part of the set that is connected to the point v. When backtracking, v is removed from the set P and added to the set X, which means that the maximum clique containing the point v has been calculated in the current state.

[0115] When the R set is a maximal clique, it must satisfy that both the P set and the X set are empty. The P set stores the points that may be added to the R set. The P set is empty, which means that no points can be added to the R set. The X set stores the points that have completed the maximal clique counting, and the points in the X set must have edges with all the points in the R set (because each time the depth-first search is performed downward, the P set and the X set are respectively taken to be adjacent to the R set to ensure this). That is, the points in the X set must be able to form a maximal clique with the R set. If the X set is not empty, it means that the maximal clique in the R set has been calculated when the maximal clique containing the points in the X set was calculated before. Therefore, R is a maximal clique if and only if P and X are both empty sets.

[0116] The embodiment of the present application constructs an undirected graph based on each traffic media and the overall similarity between the traffic media. In the undirected graph, the traffic media is used as the vertices in the undirected graph. If the overall similarity between two traffic media is greater than a preset threshold, the vertices corresponding to the two types of traffic media are connected. Based on this, the vertices in the maximal cluster mined from the undirected graph are all traffic media whose overall similarity to each other is greater than the preset threshold, and the cheating traffic media can be obtained.

[0117] The method for identifying cheating traffic media in the embodiment of the present application obtains the historical click logs of each traffic media, and the historical click logs record the click positions of each click on the recommended information. This method utilizes the fact that cheating traffic media does not care about the specific content of the recommended information when forging the click position, but has a small difference in click positions. This method has a higher accuracy rate than the prior art method of utilizing user information and recommended information features. This method obtains the number of clicks at each click position under each traffic media, calculates the similarity of the number of clicks at each click position under two types of traffic media, constructs an undirected graph, and uses a maximum clique mining algorithm to mine the maximum clique in the undirected graph to identify cheating traffic media. This method can more accurately identify cheating traffic media, protect the interests of advertisers, and optimize the ecology and reputation of the mobile advertising alliance.

[0118] In graph theory, maximal clique and maximum clique are two completely different concepts. A maximal clique is a clique that cannot be contained by a larger clique. In other words, there is no longer an edge between a point and any vertex in the clique. There are often more than one maximal clique, and the maximum clique is the maximal clique with the largest size (measured by the number of vertices in the clique) in an undirected graph, that is, there is only one maximal clique. It should be understood that in practical applications, more than one maximal clique can be mined through the maximal clique mining algorithm. If each maximal clique found is used as a cheating traffic media, there may be a problem of overdoing it. In addition, when the recommended information on two traffic media is the same, it may also lead to the problem of similar click positions under the traffic media.

[0119] See also Figure 3 , Figure 3 The recommended information in the example is a recommendation information of a mobile phone memory cleaning program. The recommendation information displays a one-click cleaning button for the user to click. Generally, when the user sees the recommendation information, most of them will click on the button area. Figure 3 A, B, and C in the figure are the display effects after the historical click logs of the recommendation information in three different traffic media are mapped to the same display area. Each point in A, B, and C represents a click. Figure 3 It can be seen that the click position of the user for the recommended information has little to do with the traffic media, and the distribution similarity of the click positions is very high.

[0120] From the above analysis, it can be seen that after the embodiment of the present application obtains the maximum cluster by using the maximum cluster mining algorithm, the vertices in the maximum cluster can also be screened, thereby further increasing the recognition accuracy of cheating traffic media.

[0121] like Figure 3 When the similarity of recommended information on multiple traffic media is extremely high, it may lead to a high similarity of the user's click position, so that the multiple traffic media may form a maximum cluster in the undirected graph, causing misjudgment. Therefore, although the method of obtaining the maximum cluster as the cheating traffic media using the maximum cluster mining algorithm can identify the cheating gang, there may be a small amount of noise in this case. In some embodiments, the maximum cluster mining algorithm can be used to determine more than one maximum cluster as suspected cheating traffic media in the undirected graph, calculate the similarity of recommended information in each maximum cluster, and identify the traffic media corresponding to the maximum cluster with a recommended information similarity lower than a preset threshold as cheating traffic media. The preset threshold can be obtained by analyzing historical data samples. It can be understood that there are multiple methods for calculating the similarity of recommended information in each maximum cluster. In a faster method, the number of recommended information shared by each traffic media in the maximum cluster can be extracted, and the number of shared recommended information divided by the total number of recommended information of each traffic media is used as the similarity of recommended information.

[0122] Figure 4 This is a logic diagram of a method for identifying cheating traffic media according to another embodiment of the present application. Figure 4 As shown, in addition to the click position (x, y) of each click on the recommended information, the historical click log of the embodiment of the present application also includes the unique identification (AD ID) of the recommended information corresponding to each click on the recommended information and the unique identification (APP ID) of the traffic media. By adding the information of the unique identification of the recommended information to the operation of clicking the recommended information each time, it is beneficial to perform refined statistics on the display of each recommended information on the traffic media. In addition, the embodiment of the present application finds that since there is no correlation between different recommended information, there should be no correlation between the click positions of the user when clicking at least two recommended information. Then, after the cheating traffic is mined out through the maximum group, it is further detected whether there is a correlation between the click position of at least two recommended information in a suspected cheating traffic media and the click of at least two recommended information in all suspected cheating traffic media, and the cheating traffic is screened twice to determine the final cheating traffic media, and then the unique identification of the cheating traffic media is sent to the anti-cheating online system, and the anti-cheating online system judges each real-time click in real time according to the unique identification of the cheating traffic. If the click comes from the cheating traffic media, it is judged as a cheating click and the cheating click is filtered out.

[0123] Based on the above embodiments, as an optional embodiment, cheating traffic media is identified by mining the maximum clique in the undirected graph, see Figure 5 ,include:

[0124] S201. Obtain the number of clicks on at least two pieces of recommended information at each click position in the traffic media according to the historical click logs of the traffic media.

[0125] It is understandable that, since the historical click log records the unique identifier of the recommended information and the click position corresponding to each click, the number of clicks on each recommended information at each click position can be counted.

[0126] S202. The traffic media corresponding to the maximum cluster is regarded as suspected cheating traffic media. According to the click volume of each click position of at least two recommended information under all suspected cheating traffic media, the similarity between the click positions of at least two recommended information under the suspected cheating traffic media and the click positions of at least two recommended information under all suspected cheating traffic media is obtained.

[0127] The traffic media corresponding to the vertices in the maximum cluster are regarded as suspected cheating traffic media, laying the foundation for filtering out normal traffic media from suspected cheating traffic media in the future.

[0128] For each suspected cheating traffic media, the embodiment of the present application will obtain the similarity of the click positions of at least two recommended information under the suspected cheating traffic media according to the click volume of each click position of at least two recommended information under all suspected cheating traffic media. The purpose of doing so is to consider the distribution characteristics of a recommended information under one traffic media and all suspected traffic media, and further accurately judge the difference of at least two recommended information under each suspected cheating traffic media, thereby further improving the accuracy of judging cheating traffic media.

[0129] Specifically, for any recommended information, first, based on the click positions of the recommended information in different traffic media, the similarity between the click positions of the recommended information in the suspected cheating traffic media and the click positions of the recommended information in all traffic media is obtained. If the similarity is high, it means that the credibility of the click positions of the recommended information recorded by the traffic media is high. Further expanding to all recommended information, the similarity between the click positions of at least two recommended information in the suspected cheating traffic media and the click positions of at least two recommended information in all suspected cheating traffic media can be obtained.

[0130] S203. For any suspected cheating traffic media, if the similarity between the click positions of at least two recommended information under the suspected cheating traffic media and the click positions of at least two recommended information under all suspected cheating traffic media is less than a first preset threshold, the suspected cheating traffic media is determined to be cheating traffic media.

[0131] Since the click positions of different recommended information are theoretically irrelevant, the similarity of the click positions of different recommended information under the same traffic media should be low. If the similarity between the click positions of at least two recommended information under suspected cheating traffic media and the click positions of at least two recommended information under all suspected cheating traffic media is less than the first preset threshold, it means that the click positions of these recommended information do not meet the irrelevance, and it can be determined that they are forged click positions. Therefore, the suspected cheating traffic media can be judged as cheating traffic media. Correspondingly, if the similarity between the click positions of at least two recommended information under suspected cheating traffic media and the click positions of at least two recommended information under all suspected cheating traffic media is greater than the first preset threshold, the suspected cheating traffic media is judged as normal traffic media.

[0132] After using the maximum clique mining algorithm to mine out suspected cheating traffic media, the embodiment of the present application further uses the irrelevance of the click positions of different recommended information to obtain the similarity between the click positions of at least two recommended information under the suspected cheating traffic media and the click positions of at least two recommended information under all suspected cheating traffic media, and determines the suspected cheating traffic media with a similarity less than a first preset threshold as cheating traffic media, thereby further improving the accuracy of identifying cheating traffic media.

[0133] Based on the above embodiments, as an optional embodiment, obtaining the similarity of the click volume of each click position under any two traffic media includes:

[0134] S301. The click volume of each click position under the traffic media is used as the element value of a preset dimension in the vector corresponding to the traffic media.

[0135] Specifically, the embodiment of the present application corresponds each click position to a dimension in the vector, that is, the total number of click positions is the number of dimensions of the vector. For example, there are 4 click positions, namely click position 1, click position 2, click position 3 and click position 4. If the click volumes of the 4 click positions under the traffic media are 500, 200, 30 and 0 respectively, then the vector corresponding to the traffic media is (500, 200, 30, 0). By vectorizing the click volume, it is convenient for the subsequent calculation of similarity.

[0136] S302 , calculating the distance between any two vectors as the similarity of click volumes of the traffic media corresponding to the two vectors at each click position.

[0137] The embodiment of the present application measures the similarity of traffic media by the distance between vectors. Specifically, it can be measured by Euclidean distance, Manhattan distance, cosine, Hamming distance, etc.

[0138] Next, taking the cosine of the angle as an example, the specific method of calculating the distance between any two vectors in the embodiment of the present application is explained.

[0139] Specifically, if the number of click positions is n, the click volume of traffic media 1 at these n click positions is {x1, x2, …, x n}, the click volume of traffic media 2 at these n click positions is {y1,y2,…,y n}, then the vector corresponding to traffic medium 1 can be expressed as (x1, x2, …, x n ), the vector corresponding to traffic medium 2 can be expressed as (y1,y2,…,y n ), the cosine distance between two vectors is:

[0140]

[0141] Based on the above embodiments, as an optional embodiment, an undirected graph is constructed according to each flow medium and the overall similarity, including:

[0142] The traffic media are regarded as vertices in an undirected graph. If the overall similarity between any two traffic media is greater than a second preset threshold, the vertices corresponding to any two traffic media are connected by an edge.

[0143] Next, a specific example is used to illustrate the process of constructing an undirected graph in the embodiment of the present application. There are 6 traffic media, namely, traffic media 0 to 5, where:

[0144] The overall similarity between the traffic media 0 and 1, and between 4 and 5 is greater than a second preset threshold;

[0145] The overall similarity between traffic media 1 and 0, 2, and 4 is greater than a second preset threshold;

[0146] The overall similarity between the traffic media 2 and 1 and 4 is greater than a second preset threshold;

[0147] The overall similarity between the traffic media 3 and 4 is greater than a second preset threshold;

[0148] The overall similarity between traffic media 4 and 0, 1, 2, 3, 5 is greater than a second preset threshold;

[0149] The overall similarity between the traffic media 5 and 0 and 4 is greater than the second preset threshold.

[0150] Construct an undirected graph based on the overall similarity between traffic media and pairwise traffic media, see Figure 6 , where vertices 0 to 5 correspond to traffic media 0 to 5 respectively, vertex 0 is connected to vertices 1, 4 and 5, vertex 1 is connected to vertices 0, 2 and 4, vertex 2 is connected to vertices 1 and 4, vertex 3 is connected to vertex 4, vertex 4 is connected to vertices 0 and 2, and vertex 5 is connected to vertices 0 and 4.

[0151] Further, we can find the maximal cliques from the undirected graph: {0,4,5}, {0,1,4}, {1,2,4}, {3,4}. In this example, we can find that these six traffic media are all suspected traffic media.

[0152] Based on the above embodiments, as an optional embodiment, see Figure 7 , according to the click amount of each click position of at least two recommended information in all traffic media, obtaining the similarity between the click position of at least two recommended information in suspected cheating traffic media and the click position of at least two recommended information in all suspected cheating traffic media, including:

[0153] S401. For any one of the at least two pieces of recommendation information, count the number of clicks on each click position of the recommendation information in the suspected cheating traffic media and the total number of clicks on each click position of the recommendation information in all traffic media.

[0154] It should be understood that since the unique identifier of the recommended information corresponding to each click is recorded in the historical click log of the traffic media, the number of clicks on the recommended information at each click position under each traffic media can be counted, including the number of clicks on the recommended information under the suspected cheating traffic media. In addition, after counting the number of clicks on the recommended information at each click position under each traffic media, the total number of clicks on the recommended information at each click position under all traffic media can be obtained by aggregating the data of all traffic media.

[0155] For example, if the recommended information has 300 clicks on position 1 under traffic media 1, 400 clicks on position 1 under traffic media 2, and 200 clicks on position 1 under traffic media 3, then the recommended information has 900 clicks on position 1 under all traffic media (assuming there are 3 traffic media in total).

[0156] S402. Obtain the similarity between the number of clicks at each click position of the recommended information in the suspected cheating traffic media and the total number of clicks at each click position of the recommended information as the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media.

[0157] Before obtaining the similarity of click positions of at least two recommended information on a suspected cheating platform, the embodiment of the present application needs to first obtain the position similarity of each recommended information on a suspected cheating traffic medium relative to all traffic media. The position similarity is used to characterize the similarity between the click position distribution of the recommended information on a traffic medium and the click position distribution request of the recommended information on all traffic media. For example, the click positions of a recommended information on a certain traffic medium are mainly concentrated in the upper left corner of the display area, while the click positions of the recommended information on all traffic media are mainly concentrated in the middle of the display area. It can be roughly obtained that the position similarity of the recommended information on a certain traffic medium relative to all traffic media is low, and it can be further inferred that the click volume of the recommended information on a certain traffic medium has a low credibility. The embodiment of the present application can qualitatively analyze the credibility of the click volume of each recommended information on the suspected cheating traffic media through step S402.

[0158] S403. For any one of the at least two recommendation information, based on the number of clicks on the recommendation information in the suspected cheating traffic media, weighted sum is performed on the position similarity of each recommendation information in the suspected cheating traffic media relative to all traffic media to obtain the similarity of the click positions of the at least two recommendation information in the suspected cheating traffic media.

[0159] Since step S402 is equivalent to obtaining the credibility of the number of clicks on each recommended information in the suspected cheating traffic media, in order to obtain the credibility of the number of clicks on all recommended information by the suspected cheating traffic media, the embodiment of the present application uses the number of clicks on each recommended information in the suspected cheating traffic media to perform weighted summation on the position similarity of the corresponding recommended information, and obtains the similarity of the click positions of at least two recommended information in the suspected cheating traffic media and the click positions of at least two recommended information in all suspected cheating traffic media. The similarity of the click positions of different recommended information in the suspected cheating traffic media can characterize the difference in the click positions of different recommended information in the suspected cheating traffic media. The higher the similarity, the greater the difference, and the more likely it is to be a normal traffic media.

[0160] Based on the above embodiments, as an optional embodiment, obtaining the similarity between the number of clicks at each click position of the recommended information in the suspected cheating traffic media and the total number of clicks at each click position of the recommended information includes:

[0161] The number of clicks on the recommended information at each click position in the suspected cheating traffic media is used as the element value of the preset dimension in the first vector; the total number of clicks on the recommended information at each click position is used as the element value of the preset dimension in the second vector.

[0162] Similar to step S301, the embodiment of the present application corresponds each click position to a dimension in the vector, that is, the total number of click positions is the number of dimensions of the vector. For example, there are 4 click positions, namely click position 1, click position 2, click position 3 and click position 4. If the click volumes of the recommended information at the 4 click positions under the suspected cheating traffic media are 200, 300, 400 and 500 respectively, and the click volumes of the recommended information at the 4 click positions under all traffic media are 500, 800, 1400 and 900 respectively, then the first vector is (200, 300, 400, 500) and the second vector is (500, 800, 1400, 900).

[0163] The distance between the first vector and the second vector is calculated as the similarity between the click volume of the recommended information at each click position in the suspected cheating traffic media and the total click volume of the recommended information at each click position.

[0164] The embodiment of the present application uses the distance between vectors to measure the similarity between the number of clicks at each click position of the recommended information in the suspected cheating traffic media and the total number of clicks at each click position of the recommended information. Specifically, it can be measured by Euclidean distance, Manhattan distance, angle cosine, Hamming distance, etc.

[0165] Based on the above embodiments, as an optional embodiment, weighted summation is performed on the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media according to the number of clicks on the recommended information in the suspected cheating traffic media, and the similarity between the click positions of at least two recommended information in the suspected cheating traffic media and the click positions of at least two recommended information in all suspected cheating traffic media is obtained, including:

[0166] S501. A ratio of the number of clicks on the recommended information in the suspected cheating traffic media to the number of clicks on at least two recommended information in the suspected cheating traffic media is used as a weight value of the recommended information.

[0167] It can be understood that the number of clicks on at least two recommended information under the suspected cheating traffic media, that is, the total number of clicks recorded in the historical click log of the suspected cheating traffic media, can be determined by dividing the number of clicks on a single recommended information under the suspected cheating traffic media by the total number of clicks, and the proportion of clicks on the single recommended information under the suspected cheating traffic media can be used as the weight value of the unit recommendation information.

[0168] S502, multiplying the weight value of the recommended information by the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media, to obtain a weighted result of the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media;

[0169] S503: Add weighted results of the position similarities of at least two pieces of recommendation information in suspected cheating traffic media relative to at least two traffic media to obtain the similarities of click positions of at least two pieces of recommendation information in suspected cheating traffic media.

[0170] Specifically, the similarity between the click positions of at least two pieces of recommended information under the suspected cheating traffic media can be calculated using the following formula:

[0171]

[0172] Relevance represents the similarity between the click positions of at least two recommended information in the suspected cheating traffic media, m represents the number of unique identifiers of recommended information recorded in the historical click logs of the suspected cheating traffic media, and C i Indicates the number of clicks on the recommended information i in the suspected cheating traffic media, Similarity i It represents the position similarity of the recommended information i relative to all traffic media under suspected cheating traffic media.

[0173] On the basis of the above embodiments, as an optional embodiment, after identifying cheating traffic media according to the traffic media corresponding to the maximum cluster, it also includes: obtaining click information collected by each traffic media in real time, if the click information comes from the cheating traffic media, it is determined that the click information is a cheating click, and the click information is filtered out.

[0174] After the cheating traffic media is determined, the embodiment of the present application determines whether the click information obtained in real time is from the cheating traffic media. If so, the click information is filtered out. If not, the click information is sent to the mobile advertising alliance for statistics. In this way, it can be ensured that the final billing of the mobile advertising alliance is authentic and reliable.

[0175] Figure 8 This is a flow chart of another embodiment of the present application for identifying cheating traffic media. Figure 8As shown, the traffic media is specifically APP (Application), and the historical period is one week. The process is divided into two parts, namely the offline mining part and the online application part. The offline mining part refers to the use of the historical click logs of the traffic media to determine the cheating traffic media, and the online application part refers to the real-time judgment of the click information obtained in real time according to the source of the click information, to determine whether the click information is a cheating click or a normal click.

[0176] Specifically, in the offline mining stage, the overall similarity between each APP is calculated based on the click position (click coordinates) of each APP in the past week, and an undirected graph of the APP is constructed. In the undirected graph, the APP is used as a vertex in the undirected graph. If the overall similarity of two APPs is greater than a second preset threshold, the vertices corresponding to the two APPs are connected by an edge, and the suspected cheating traffic media is mined out through the Bron-Kerbosch algorithm; further, based on the click position of each recommended information, the position similarity of the recommended information under the suspected cheating traffic media relative to all traffic media is calculated, and the correlation of at least two recommended information in the suspected cheating traffic media is further obtained based on the position similarity, and the cheating traffic media is determined from the suspected cheating traffic media based on the correlation, and then the cheating traffic media is written into the APP blacklist.

[0177] When used online, the click information sent by each traffic media is obtained in real time. For example, the APP blacklist will make online judgments on the click information. If the APP from which the click information comes is in the APP blacklist, the click information will be judged as a cheating click. If the APP from which the click information comes is not in the APP blacklist, the click information will be judged as a normal click. If it is a normal click, the click information will be aggregated to the mobile recommendation information platform, which will generate the cost of promoting the recommendation information based on the number of normal clicks.

[0178] The present application embodiment provides a device for identifying cheating traffic media, such as Fig. 9 As shown, the device may include: a click information acquisition module 101, an overall similarity calculation module 102, an undirected graph construction module 103 and a cheating traffic media identification module 104. Specifically:

[0179] The click information acquisition module 101 is used to acquire the historical click logs of each traffic media, the historical click logs include the click information generated by each click on the recommended information in the historical period, and the click position is recorded in the click information;

[0180] The overall similarity calculation module 102 is used to obtain the click volume of each click position under the traffic media according to the historical click log, obtain the similarity of the click volume of each click position under any two traffic media in each traffic media, and calculate the overall similarity between any two traffic media according to the similarity of the click volume of each click position under any two traffic media;

[0181] The undirected graph construction module 103 is used to construct an undirected graph according to the overall similarity between any two traffic media, wherein the vertices of the undirected graph are the traffic media, and when the overall similarity between any two traffic media is greater than a second predetermined threshold, the two vertices corresponding to the any two traffic media are connected by an edge;

[0182] The cheating traffic media identification module 104 is used to determine the maximum clique in the undirected graph and identify the cheating traffic media according to the traffic media corresponding to the maximum clique.

[0183] The device for identifying cheating traffic media provided by the embodiment of the present invention specifically executes the above method embodiment process. For details, please refer to the content of the above method embodiment for identifying cheating traffic media, which will not be repeated here. The device for identifying cheating traffic media provided by the embodiment of the present invention obtains the historical click logs of each traffic media. The historical click logs record the click position of each click on the recommended information. It utilizes the characteristic that cheating traffic media does not care about the specific content of the recommended information when forging the click position, but has a small difference in click positions. Compared with the prior art method of utilizing user information and recommended information features, it has a higher accuracy rate. By obtaining the click volume of each click position under each traffic media, calculating the similarity of the click volume of each click position under the two types of traffic media, constructing an undirected graph, and using the maximum clique mining algorithm to tap the maximum clique in the undirected graph to identify cheating traffic media, it can identify cheating traffic media, protect the interests of advertisers, and optimize the ecology and reputation of the mobile advertising alliance.

[0184] Based on the above embodiments, as an optional embodiment, the click information also includes a unique identifier of the recommended information;

[0185] Fraudulent traffic media identification module:

[0186] A first click volume statistics unit is used to obtain the click volume of at least two pieces of recommended information at each click position in the traffic media according to the historical click log of the traffic media;

[0187] The recommendation information relevance determination unit is used to classify the traffic media corresponding to the maximum cluster as suspected cheating traffic media, and obtain the similarity between the click positions of the at least two recommended information in the suspected cheating traffic media and the click positions of the at least two recommended information in all suspected cheating traffic media according to the click volume of each click position of the at least two recommended information in all suspected cheating traffic media;

[0188] The screening unit is used to determine, for any suspected cheating traffic media in each traffic media, the suspected cheating traffic media as cheating traffic media if the similarity of the click positions of at least two recommended information under the suspected cheating traffic media to any suspected cheating traffic media in each traffic media is less than a first preset threshold.

[0189] Based on the above embodiments, as an optional embodiment, the overall similarity calculation module includes:

[0190] A first vector generation submodule is used to use the click volume of each click position under the traffic media as an element value of a preset dimension in a vector corresponding to the traffic media;

[0191] The first vector distance submodule is used to calculate the distance between any two vectors as the similarity of the click volume of the traffic media corresponding to the any two vectors at each click position.

[0192] Based on the above embodiments, as an optional embodiment, the recommendation information relevance determination unit includes:

[0193] A click volume statistics subunit is used to count the click volume of each click position of the recommended information in the suspected cheating traffic media and the total click volume of each click position of the recommended information in all traffic media for any one of the at least two recommended information;

[0194] The position similarity calculation subunit is used to obtain the similarity between the click volume of each click position of the recommended information in the suspected cheating traffic media and the total click volume of each click position of the recommended information in all traffic media, as the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media;

[0195] The recommendation information relevance determination subunit is used to perform weighted summation of the position similarities of each recommendation information in the suspected cheating traffic media relative to all traffic media for any one of the at least two recommendation information according to the number of clicks on the recommended information in the suspected cheating traffic media, so as to obtain the similarity between the click positions of the at least two recommendation information in the suspected cheating traffic media and the click positions of the at least two recommendation information in all suspected cheating traffic media.

[0196] Based on the above embodiments, as an optional embodiment, the position similarity calculation subunit includes:

[0197] The second vector generation submodule is used to use the number of clicks of the recommended information at each click position in the suspected cheating traffic media as the element value of the preset dimension in the first vector; and use the total number of clicks of the recommended information at each click position as the element value of the preset dimension in the second vector;

[0198] The second vector distance submodule is used to calculate the distance between the first vector and the second vector as the similarity between the click volume of the recommended information at each click position in the suspected cheating traffic media and the total click volume of the recommended information at each click position.

[0199] Based on the above embodiments, as an optional embodiment, the recommendation information relevance determination subunit includes:

[0200] A weight determination submodule, used to use the ratio of the number of clicks on the recommended information in the suspected cheating traffic media to the number of clicks on at least two recommended information in the suspected cheating traffic media as the weight value of the recommended information;

[0201] A weighting submodule, for multiplying the weight value of the recommended information by the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media, to obtain a weighted result of the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media;

[0202] The weighted summation submodule is used to add the weighted results of the position similarity of at least two recommended information under suspected cheating traffic media relative to all traffic media, so as to obtain the similarity between the click positions of at least two recommended information under suspected cheating traffic media and the click positions of at least two recommended information under all suspected cheating traffic media.

[0203] Based on the above embodiments, as an optional embodiment, the click information also includes a unique identifier of the flow media;

[0204] The device for identifying cheating traffic media also includes:

[0205] The real-time judgment module is used to build a blacklist based on the unique identifier of the cheating traffic media; obtain the click information collected by each traffic media in real time. If the unique identifier of the traffic media recorded in the click information is in the blacklist, it is determined that the click information is a cheating click and the cheating click is filtered out.

[0206] An electronic device is provided in an embodiment of the present application, and the electronic device includes: a memory and a processor; at least one program, which is stored in the memory and is used to be executed by the processor. Compared with the prior art, it can achieve: by obtaining the historical click log of each traffic media, the historical click log records the click position of each click on the recommended information, and utilizes the fact that the cheating traffic media does not care about the specific content of the recommended information when forging the click position, but has a small difference in the click position. Compared with the prior art method of utilizing user information and recommended information features, it has a higher accuracy rate, and by obtaining the click volume of each click position under each traffic media, the similarity of the click volume of each click position under the two types of traffic media is calculated, and an undirected graph is constructed. The maximum clique in the undirected graph is mined through the maximum clique mining algorithm to identify the cheating traffic media, which can more accurately identify the cheating traffic media, protect the interests of advertisers, and optimize the ecology and reputation of the mobile advertising alliance.

[0207] In an alternative embodiment, an electronic device is provided, such as Fig.10 As shown, Fig.10 The electronic device 4000 shown includes: a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, such as through a bus 4002. Optionally, the electronic device 4000 may also include a transceiver 4004. It should be noted that in actual applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present application.

[0208] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 4001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0209] The bus 4002 may include a path to transmit information between the above components. The bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 4002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig.10 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0210] The memory 4003 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

[0211] The memory 4003 is used to store the application code for executing the solution of the present application, and the execution is controlled by the processor 4001. The processor 4001 is used to execute the application code stored in the memory 4003 to implement the contents shown in the above method embodiment.

[0212] The embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer is run on a computer, the computer can execute the corresponding content in the aforementioned method embodiment. Compared with the prior art, by obtaining the historical click logs of each traffic media, the historical click logs record the click position of each click on the recommended information, and utilize the fact that the cheating traffic media does not care about the specific content of the recommended information when forging the click position, but has a small difference in the click position. Compared with the prior art method of using user information and recommended information features, it has a higher accuracy rate, and by obtaining the click volume of each click position under each traffic media, calculating the similarity of the click volume of each click position under the two types of traffic media, constructing an undirected graph, and using the maximum clique mining algorithm to dig out the maximum clique in the undirected graph to identify the cheating traffic media, it can more accurately identify the cheating traffic media, protect the interests of advertisers, and optimize the ecology and reputation of the mobile advertising alliance.

[0213] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0214] The above are only some embodiments of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for identifying cheating traffic media, characterized in that: include: Obtaining a historical click log of each traffic media, wherein the historical click log includes click information generated by each click on the recommended information within a historical period, and the click information includes a click location; Obtaining the click volume of each click position under each flow media according to the historical click log, obtaining the similarity of the click volume of each click position under any two flow media in each flow media, and calculating the overall similarity between the any two flow media according to the similarity of the click volume of each click position under the any two flow media; According to the overall similarity between the two arbitrary traffic media, an undirected graph is constructed, wherein the vertices of the undirected graph are traffic media, and when the overall similarity between the two arbitrary traffic media is greater than a second predetermined threshold, two vertices corresponding to the two arbitrary traffic media are connected by an edge; A maximal clique in the undirected graph is determined, and cheating traffic media is identified according to traffic media corresponding to the maximal clique.

2. The method for identifying cheating traffic media according to claim 1, characterized in that: The click information also includes a unique identifier of the recommended information; The step of identifying cheating traffic media according to the traffic media corresponding to the maximum cluster includes: According to the historical click logs of traffic media, the click volume of different recommended information at each click position under traffic media is obtained; The traffic media corresponding to the maximum clique is regarded as the suspected cheating traffic media, and according to the click volume of each click position of at least two recommended information under all suspected cheating traffic media, the similarity between the click positions of the at least two recommended information under the suspected cheating traffic media and the click positions of the at least two recommended information under all suspected cheating traffic media is obtained; For any suspected cheating traffic media, if the similarity between the click positions of the at least two recommendation information under the suspected cheating traffic media and the click positions of the at least two recommendation information under all suspected cheating traffic media is less than a first preset threshold, the suspected cheating traffic media is determined to be a cheating traffic media.

3. The method for identifying cheating traffic media according to claim 1, characterized in that: The obtaining of the similarity of the click volume of each click position under any two flow media in the flow media includes: The click volume of each click position under the traffic media is used as the element value of a preset dimension in the vector corresponding to the traffic media; The distance between any two vectors is calculated as the similarity of the click volumes of the two traffic media corresponding to the any two vectors at each click position.

4. The method for identifying cheating traffic media according to claim 2, characterized in that: The obtaining, according to the click amount of each click position of the at least two recommended information in all traffic media, the similarity between the click positions of the at least two recommended information in the suspected cheating traffic media and the click positions of the at least two recommended information in all suspected cheating traffic media includes: For any one of the at least two pieces of recommendation information, counting the number of clicks on each click position of the recommendation information in the suspected cheating traffic media and the total number of clicks on each click position of the recommendation information in all traffic media; Obtaining the similarity between the number of clicks at each click position of the recommended information in the suspected cheating traffic media and the total number of clicks at each click position of the recommended information in all traffic media as the position similarity of the recommended information in the suspected cheating traffic media relative to all traffic media; For any one of the at least two recommendation information, based on the number of clicks on the recommendation information under the suspected cheating traffic media, a weighted sum is performed on the position similarities of the recommendation information under the suspected cheating traffic media relative to all traffic media to obtain the similarity between the click positions of the at least two recommendation information under the suspected cheating traffic media and the click positions of the at least two recommendation information under all suspected cheating traffic media.

5. The method for identifying cheating traffic media according to claim 4, characterized in that: Obtaining the similarity between the number of clicks at each click position of the recommendation information under the suspected cheating traffic media and the total number of clicks at each click position of the recommendation information, including: The number of clicks of the recommended information at each click position under the suspected cheating traffic media is used as the element value of the preset dimension in the first vector; the total number of clicks of the recommended information at each click position is used as the element value of the preset dimension in the second vector; The distance between the first vector and the second vector is calculated as the similarity between the number of clicks of the recommendation information at each click position in the suspected cheating traffic media and the total number of clicks of the recommendation information at each click position.

6. The method for identifying cheating traffic media according to claim 5, characterized in that: The step of performing weighted summation of the position similarities of the recommended information in the suspected cheating traffic media relative to all traffic media according to the number of clicks of the recommended information in the suspected cheating traffic media to obtain the similarities between the click positions of the at least two recommended information in the suspected cheating traffic media and the click positions of the at least two recommended information in all suspected cheating traffic media includes: The ratio of the number of clicks on the recommended information in the suspected cheating traffic medium to the number of clicks on the at least two recommended information in the suspected cheating traffic medium is used as the weight value of the recommended information; According to the weight value of the recommendation information, the position similarity of the recommendation information under the suspected cheating traffic medium relative to all traffic media is multiplied to obtain a weighted result of the position similarity of the recommendation information under the suspected cheating traffic medium relative to all traffic media; Add the weighted results of the position similarities of the at least two recommendation information under the suspected cheating traffic media relative to all traffic media to obtain the similarity between the click positions of the at least two recommendation information under the suspected cheating traffic media and the click positions of the at least two recommendation information under all suspected cheating traffic media.

7. The method for identifying cheating traffic media according to any one of claims 1 to 6, characterized in that: The click information also includes a unique identifier of the traffic media; The step of identifying cheating traffic media according to the traffic media corresponding to the maximum cluster further includes: A blacklist is constructed according to the unique identifier of the cheating traffic media; click information collected in real time by each traffic media is obtained, and if the unique identifier of the traffic media recorded in the click information is in the blacklist, it is determined that the click information is a cheating click, and the cheating click is filtered out.

8. A device for identifying cheating traffic media, characterized in that: include: A click information acquisition module is used to acquire the historical click logs of each traffic media, wherein the historical click logs include click information generated by each click on the recommended information within a historical period, and the click information records the click position; An overall similarity calculation module, used to obtain the click volume of each click position under each flow media according to the historical click log, obtain the similarity of the click volume of each click position under any two flow media in each flow media, and calculate the overall similarity between the any two flow media according to the similarity of the click volume of each click position under the any two flow media; An undirected graph construction module, used to construct an undirected graph according to the overall similarity between the two arbitrary traffic media, wherein the vertices of the undirected graph are traffic media, and when the overall similarity between the two arbitrary traffic media is greater than a second predetermined threshold, two vertices corresponding to the two arbitrary traffic media are connected by an edge; The cheating traffic media identification module is used to determine the maximum clique in the undirected graph and identify the cheating traffic media according to the traffic media corresponding to the maximum clique.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method for identifying cheating traffic media as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions enable the computer to execute the steps of the method for identifying cheating traffic media as described in any one of claims 1 to 7.

Images