CN110310122B

CN110310122B - iOS charging risk control method based on graph structure

Info

Publication number: CN110310122B
Application number: CN201910567553.8A
Authority: CN
Inventors: 雷鸣
Original assignee: Shanghai Microphone Culture Media Co ltd
Current assignee: Shanghai Microphone Culture Media Co ltd
Priority date: 2019-06-27
Filing date: 2019-06-27
Publication date: 2023-09-01
Anticipated expiration: 2039-06-27
Also published as: CN110310122A

Abstract

The invention provides an iOS charging risk control method based on a graph structure, which comprises the following steps: s1, acquiring substitute charging user data through a substitute charging platform, and performing preliminary filtering and screening on the substitute charging user data; s2, carrying out graphic structure Graph processing on the user data subjected to preliminary filtering and screening, and establishing a graphic structure of a correlation edge between the user data; s3, calculating a pagerank value of each point in the user data graph structure through a pagerank algorithm, and endowing each webpage with an evaluation value indicating the importance degree of the webpage through the algorithm; s4, carrying out a process of finding the substituted charge network community according to the graph structure, and predicting the density degree of each user data in the substituted charge network community according to a semi-supervised learning method; and S5, calculating a pagerank average value of all user data in the network community by substitution, and judging substitution possibility of all user data in the network community by substitution.

Description

iOS charging risk control method based on graph structure

Technical Field

The invention relates to the field of computer network risk control, in particular to an iOS (integrated circuit) substitution risk control method based on a graph structure.

Background

With the rapid development of the mobile internet, many internet companies continuously push out various mobile phone APP, and these internet companies continuously change our lives and simultaneously provide better services for people, wherein the better services include member VIP service, purchasing payment album of content provider, recharging virtual goods of game, etc., more or less the services need to be recharged and paid by users, and the cost is convenient for users and simultaneously promotes the internet companies to push out better contents and services. The APP is mainly mobile phone software running on a mobile terminal, the two systems with the largest market share of the mobile terminal are IOS systems of android and apples, the IOS systems respectively occupy 70% and 30%, when users and the Internet are charged on the mobile terminals of the IOS systems, such as mobile phones, the APP needs to follow payment policies formulated by apple companies, however, in the process of charging in the IOS version, a black-generation charging loophole appears, namely, the users charge by a charging platform for own account to purchase virtual items, such as VIP member services, payment contents or game equipment, and the like, after the virtual items reach account numbers, the charging platform initiates refund application to apple companies, however, the virtual items cannot be cancelled and refund to related Internet companies, the users only need to pay little fees to obtain virtual items with amounts of tens of times through regular channel charging, the charging platform can benefit through charging users' payment, and the apple companies can charge more than 30% of the internet companies in each time through charging, and the related Internet companies can bear huge losses in the whole charging process.

Disclosure of Invention

The invention aims to at least solve the problem that corresponding punishment measures are made for the substitution charging holes in the prior art, mobile internet companies also find some substitution charging behaviors in real time through some rules, for example, judge whether to charge with common equipment, if not, add users into a blacklist, seal numbers for users who are frequently substituted for charging, even shut down a certain server and the like when the loss is serious, and although complex and detailed rules are formulated, the Du Juedai charging phenomenon still cannot occur. All the invention particularly and innovatively provides an iOS charging risk control method based on a graph structure.

In order to achieve the above object of the present invention, the present invention provides an iOS substitution risk control method based on a graph structure, including the steps of:

s1, acquiring substitute charging user data through a substitute charging platform, and performing preliminary filtering and screening on the substitute charging user data;

s2, carrying out graphic structure Graph processing on the user data subjected to preliminary filtering and screening, and establishing a graphic structure of a correlation edge between the user data;

s3, calculating a pagerank value of each point in the user data graph structure through a pagerank algorithm, and endowing each webpage with an evaluation value indicating the importance degree of the webpage through the algorithm;

s4, carrying out a process of finding the substituted charge network community according to the graph structure, and predicting the density degree of each user data in the substituted charge network community according to a semi-supervised learning method;

and S5, calculating a pagerank average value of all user data in the network community by substitution, and judging substitution possibility of all user data in the network community by substitution.

Preferably, the S1 includes:

s1-1, acquiring user data through network data of a substitution charging platform, performing type division according to charging types of the substitution charging platform, and extracting a behavior history log of substitution charging transaction;

s1-2, deleting noise fields contained in a behavior history log of a user substitution transaction, and reserving keyword fields, wherein the keyword fields comprise user ids, namely u_id, user client ips, namely client_ip, equipment ids used by users, namely device_id, recharging amount pay_current and recharging time pay_time;

s1-3, carrying out preliminary filtering screening on the substitute charging user data according to the keyword fields to form the substitute charging user data containing the keyword fields.

Preferably, the S2 includes:

s2-1, preliminarily filtering and screening user data to form a graph structure, wherein user nodes in the graph structure are formed by user u_id;

s2-2, selecting different recharging types of the recharging platform according to user data to form first type data, second type data, third type data and fourth type data, wherein the first type data to the fourth type data are sequentially increased;

s2-3, a user sends recharging requests to different recharging types of a substitution recharging platform, any one type of data from first type data to fourth type data is selected to establish a standard of a graphic structure, if the recharging requests are generated between a first user u_id1 and a second user u_id2 through a shared user client side_ip, and recharging amount pay_amountis greater than or equal to the selected type of data, a directed edge pointing to each other is generated between two user nodes of the first user u_id1 and the second user u_id2;

s2-4, if recharging operation occurs between the first user u_id1 and the second user u_id2 through a shared iOS device, and the recharging amount pay_amountis greater than or equal to the type of data, directional edges pointing to each other are generated between the two user nodes of the first user u_id1 and the second user u_id2; the Graph structure Graph of all recharging users can be constructed by acquiring the behavior history log of the user substitution recharging transaction.

Preferably, the S3 includes:

s3-1, calculating the pagerank value of each point in the Graph structure Graph through a pagerank algorithm, wherein when any user node performs the substitution operation, the pagerank calculates the possibility that the user node reaches a certain generation of charging platform after clicking any time of links, if the more external back-pointing links from other high-clicking-rate substitution charging platforms are possessed by a certain generation of charging platform, the larger the probability that the pointing rate of the user performing the substitution operation reaches the substitution charging platform is, otherwise, the smaller the probability that the pointing rate of the user performing the substitution charging operation reaches the substitution charging platform is;

s3-2, the pagerank algorithm endows each generation charging platform with an evaluation value for indicating the link importance degree of the generation charging platform, and an importance index formed by the high click rate generation charging platform is obtained according to the importance indexes of all other generation charging platforms pointing to the generation charging platform and the link numbers of the generation charging platforms contained in the generation charging platforms;

s3-3, calculating importance indexes, namely pagerank values, for the user u_id in each Graph structure Graph through a pagerank algorithm.

Preferably, the S4 includes:

s4-1, network community discovery is carried out on the Graph structure Graph, semi-supervised learning is carried out through a label propagation algorithm, label information of marked user nodes is utilized to predict label information of unmarked nodes, in the Graph structure Graph, connection of some user nodes is tight, and connection relation among some users is sparse;

s4-2, in the network community of the generation and filling of all user data, the part with the compact connection of the Graph structure is regarded as one network community of the generation and filling, the internal user nodes of the network community are connected more tightly, the part with the sparse connection of the Graph structure is regarded as another network community of the generation and filling, the relative connection of the network communities of the generation and filling is sparse, and the nodes in the Graph structure can be divided into a plurality of network communities of the generation and filling through LPA.

Preferably, the step S5 includes:

s5-1, calculating an average pagerank of each generation charging network community, obtaining a plurality of generation charging network communities which are closely connected respectively through an LPA algorithm after obtaining the importance degree pagerank of each user u_id in a Graph, and obtaining the importance degree label_pagerank of each generation charging network community in the Graph structure Graph by calculating the average value of the pagerank of each user u_id in the generation charging network communities;

s5-2, analyzing the captured substitution charging user data, wherein the substitution charging network community importance index is in direct proportion to the occurrence substitution charging behavior probability of all user u_ids in the substitution charging network community, and the higher the substitution charging network community importance index is, the larger the substitution charging proportion of the user u_ids contained in the substitution charging network community is, and the importance degree label_pagerank of the substitution charging network community can be used as a basis for judging the substitution charging possibility of all the user u_ids in the network community.

Preferably, the method further comprises:

s6, setting a judgment threshold value alpha in a substitution charging platform acquired by the iOS, and judging that all the user u_ids in the substitution charging network community generate substitution charging behaviors when the importance degree label_pagerank of the substitution charging network community is larger than alpha and smaller than alpha, wherein the user u_id in the substitution charging network community is a normal user.

In summary, due to the adoption of the technical scheme, the beneficial effects of the invention are as follows:

by means of the network data mining process, the abnormal account numbers are collected and analyzed, and the problematic abnormal charging account number data information is mined, so that the data analysis mining efficiency is improved, and risks are reduced for the Internet platform.

Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.

Drawings

The foregoing and/or additional aspects and advantages of the invention will become apparent and may be better understood from the following description of embodiments taken in conjunction with the accompanying drawings in which:

FIG. 1 is a general flow of prior art IOS version APP processing of a charging issue;

FIG. 2 is a diagram structure-based reverse IOS version APP substitution method;

FIG. 3 is a graph showing the distribution of various web communities of different gray colors obtained by using the LPA web community discovery algorithm according to the present invention.

Detailed Description

Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative only and are not to be construed as limiting the invention.

As shown in fig. 1 and 2, the method for mining the substitution charging behavior based on the graph structure generally has the defects of being passive, high in delay, low in accuracy and the like by formulating complex rules for filtering, has the advantages of being high in accuracy, high in initiative, capable of coping with variable substitution charging behavior and the like, and is verified in actual use. In order to achieve the above object, the following technical solutions are adopted in the embodiments of the present invention:

step 1, obtaining history logs of recharging of each package in product service through each channel by a user. And deleting useless fields contained in the user recharging log, and reserving main fields including user id (uid), user client ip (client_ip), device id (deviceid) used by a user, recharging amount paymount, recharging time and the like.

Step 2, as shown in fig. 3, a graph structure is established, wherein nodes in the graph structure are formed by user's uid, general member recharging is divided into four packages of level, for example, member recharging is divided into four packages of a, B, C and D, respectively corresponding to four sequentially raised prices of a, B, C and D, considering that the price of package is generally selected to be higher and more profit is obtained when the user finds a substitution recharging platform, here, the price of package C is selected as a standard for establishing the graph structure, if the price of package C is used for recharging through a common client_ip between user's uid1 and uid2, and if the recharging amount payamount is greater than or equal to C, directional edges which are mutually pointed to each other are generated between two nodes of uid1 and uid2, and similarly, if the recharging amount payamount is greater than or equal to C through a common device between user's uid1 and uid2, directional edges which are mutually pointed to each other are generated between two nodes; by acquiring the user recharging history log, graph structures Graph of all recharging users can be constructed.

Step 3, calculating the pagerank value of each point in the Graph through a pagerank algorithm, wherein the pagerank value is calculated by calculating the possibility that a certain person reaches a certain webpage after clicking links for any time, if the certain webpage has more external back finger links from other hot webpages, the probability that a new user clicks to reach the webpage is larger, the algorithm endows each webpage with an evaluation value for indicating the importance degree of the webpage, and the importance of the webpage is obtained according to the importance of all other webpages pointing to the webpage and the number of links contained in the webpages; also, we can calculate the importance level, i.e., the pagerank value, for the points (i.e., uid) in each Graph by the pagerank algorithm.

And 4, performing network community discovery on the Graph, wherein label information of marked nodes is utilized to predict label information of unmarked nodes by using a Graph structure semi-supervised learning method of LPA (Label Propagation, label propagation algorithm), in the Graph, connection of some nodes is tighter, connection relation between some users is sparse, in the Graph network, the part which is more tightly connected can be regarded as a network community, the internal nodes are more tightly connected, and the two network communities are relatively sparsely connected, and finally, the nodes in the Graph can be divided into a plurality of network communities by LPA.

And 5, calculating the average parerank of each network community, after obtaining the importance degree parerank of each user in the graph, obtaining a plurality of network communities which are closely connected by LPA algorithm, and obtaining the importance degree label_parerank of the network community in the graph by calculating the average value of the parerank for all the user in the network community, wherein the importance degree of the network community is in direct proportion to the probability of occurrence of the charging behavior of all the user in the network community according to the analysis found by the captured charging users, and the larger the importance degree of the network community is, the larger the proportion of occurrence of the charging of the user contained in the network community is, namely the label_parerank of the network community can be used as the basis for judging the charging possibility of all the user in the network community.

Step six, setting a threshold value, namely alpha, in actual use, when the label_pagerank of the network community is larger than alpha, the user can judge that all the uid in the network community is in substitution and charging, and when the value is smaller than alpha, the user indicates that the uid in the network community is a normal user, and the user can properly increase or decrease the size of the alpha threshold value according to the strict degree in an actual service scene.

While embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that: many changes, modifications, substitutions and variations can be made to the embodiments without departing from the spirit and principles of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. The iOS charging risk control method based on the graph structure is characterized by comprising the following steps of:

s5, judging the substitution possibility of all user data in the substitution network community by calculating a pagerank average value of all user data in the substitution network community;

the S1 comprises the following steps:

2. The iOS substitution risk control method based on the graph structure according to claim 1, wherein the S2 comprises:

3. The iOS substitution risk control method based on the graph structure according to claim 1, wherein the S3 comprises:

4. The iOS-based risk control method for replacing charging according to claim 1, wherein S4 comprises:

5. The iOS-based risk control method for replacing charging according to claim 3, wherein S5 comprises:

6. The iOS-based risk control method for replacing charging based on graph structure according to claim 1, further comprising: