Method, device, equipment and medium for screening risky users on Internet of Things platforms

By performing clustering kernel matrix processing and multi-stage clustering on the initial dataset of the IoT platform, potential risk users are screened out, solving the problems of low accuracy and low efficiency in existing technologies, and achieving efficient and accurate risk user screening.

CN117290564BActive Publication Date: 2024-11-08E SURFING IOT CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311185412.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-09-14
Publication Date
2024-11-08
Estimated Expiration
2043-09-14

AI Technical Summary

Technical Problem

Existing IoT platforms suffer from low accuracy and efficiency when screening risky users, especially due to the curse of dimensionality caused by the autocorrelation and sparsity of user-related data, which affects platform resources and user experience.

Method used

By acquiring the initial dataset, performing clustering kernel matrix processing, calculating eigenvalues ​​and eigenvectors for dimensionality reduction, and using clustering algorithms to adaptively calculate the optimal neighborhood radius and data object number threshold, multi-stage clustering and risk diagnosis are performed to screen out potential risk users.

Benefits of technology

It improves the accuracy and efficiency of risk user screening, reduces the amount of computation, and ensures data accuracy and screening efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117290564B_ABST
    Figure CN117290564B_ABST
Patent Text Reader

Abstract

The embodiment of the present invention discloses a method, device, equipment and medium for screening risky users for an Internet of Things platform. It includes: processing the initial data set of the acquired user to obtain a clustering kernel matrix; reducing the dimension of the eigenvector according to the eigenvalue of the clustering kernel matrix to obtain a data sample set; extracting M samples from the data sample set to obtain a sampled sample set, and adaptively calculating the optimal neighborhood radius and the threshold value of the number of data objects in the optimal neighborhood through a clustering algorithm according to the sampled sample set, where M≥1; inputting it into the clustering algorithm for clustering to obtain a centroid set; extracting the data in the centroid set to obtain an initial screening data set; clustering the remaining data sets in the initial data set according to the initial screening data set and the optimal neighborhood radius to obtain a target screening data set; determining the potential risk user sequence according to the target screening data set. The embodiment of the present application improves the screening accuracy and efficiency of risky users.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things, and in particular to a method, device, equipment and medium for screening risky users on an Internet of Things platform. Background Art

[0002] With the rapid development of IoT devices, a large number of devices with different network standards and protocols are connected to the IoT platform for unified management and transmission of massive data. However, during the operation and maintenance of the platform, it often happens that the platform resources are maliciously occupied due to the unreasonable operation of some risky users, which affects the formal access and use of other users, or there is a risk of user loss. Therefore, in the prior art, based on the business model of users when using the IoT platform, risky users are screened and mined, and early intervention is made to avoid their expansion of adverse effects. However, since the feature dimensions of the IoT platform user-related dimensional data such as the number of online terminals, the number of uplink and downlink message communications, and the number of application-level concurrency are all autocorrelated, they are not linearly distributed data, which leads to poor prediction effect of the user mining model and low screening accuracy; there are many feature dimensions involved in the end, cloud and other aspects, resulting in sparse distribution on each feature dimension, leading to the dimensional disaster problem, and the amount of calculation is huge when predicting the user mining model, and its screening efficiency is low. Summary of the invention

[0003] The embodiments of the present invention provide a method, device, equipment and medium for screening risky users on an Internet of Things platform, aiming to improve the accuracy and efficiency of screening risky users.

[0004] In a first aspect, an embodiment of the present invention provides a method for screening risky users for an Internet of Things platform, which includes:

[0005] Acquire an initial data set of a user, and process the initial data set to obtain a clustering kernel matrix;

[0006] Calculating the eigenvalues ​​and eigenvectors of the aggregation kernel matrix, and reducing the dimensionality of the eigenvectors according to the eigenvalues ​​to obtain a data sample set;

[0007] Extracting M samples from the data sample set to obtain a sampled sample set, and adaptively calculating an optimal neighborhood radius and a threshold value of the number of data objects in the optimal neighborhood through a clustering algorithm according to the sampled sample set, wherein M≥1;

[0008] Inputting the optimal neighborhood radius and the data object number threshold in the optimal neighborhood into a clustering algorithm for clustering to obtain a centroid set;

[0009] Performing risk diagnosis on the data in the centroid set to extract sample data to obtain an initial screening data set;

[0010] Clustering the remaining data sets in the initial data set according to the initial screening data set and the optimal neighborhood radius to obtain a target screening data set;

[0011] The risk value of each sample point in the target screening data set is calculated, and the risk values ​​are sorted to obtain a potential risk user sequence.

[0012] In a second aspect, an embodiment of the present invention further provides a risk user screening device for an Internet of Things platform, which includes:

[0013] An acquisition processing unit, used for acquiring an initial data set of a user, and processing the initial data set to obtain a clustering kernel matrix;

[0014] A dimension reduction calculation unit is used to calculate the eigenvalues ​​and eigenvectors of the aggregation kernel matrix, and reduce the dimension of the eigenvector according to the eigenvalues ​​to obtain a data sample set;

[0015] A first extraction unit is used to extract M samples from the data sample set to obtain a sample set, and adaptively calculate an optimal neighborhood radius and a threshold value of the number of data objects in the optimal neighborhood according to the sample set through a clustering algorithm, wherein M≥1;

[0016] A first clustering unit, used for inputting the optimal neighborhood radius and the data object number threshold in the optimal neighborhood into a clustering algorithm for clustering to obtain a centroid set;

[0017] A second extraction unit is used to perform risk diagnosis on the data in the centroid set to extract sample data to obtain an initial screening data set;

[0018] a second clustering unit, configured to cluster the remaining data sets in the initial data set according to the initial screening data set and the optimal neighborhood radius to obtain a target screening data set;

[0019] The calculation and sorting unit is used to calculate the risk value of each sample point in the target screening data set, and sort the risk values ​​to obtain a potential risk user sequence.

[0020] In a third aspect, an embodiment of the present invention further provides a computer device, wherein the computer device is equipped with an Internet of Things platform, the computer device comprises a memory and a processor, a computer program is stored in the memory, and the above method is implemented when the processor executes the computer program.

[0021] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the storage medium stores a computer program, and the computer program can implement the above method when executed by a processor.

[0022] The embodiment of the present invention provides a method, device, equipment and medium for screening risky users for an Internet of Things platform. The method includes: obtaining an initial data set of a user, and processing the initial data set to obtain a clustering kernel matrix; calculating the eigenvalues ​​and eigenvectors of the clustering kernel matrix, and reducing the eigenvectors according to the eigenvalues ​​to obtain a data sample set; extracting M samples from the data sample set to obtain a sampled sample set, and adaptively calculating the optimal neighborhood radius and the threshold value of the number of data objects in the optimal neighborhood through a clustering algorithm according to the sampled sample set, wherein M≥1; inputting the optimal neighborhood radius and the threshold value of the number of data objects in the optimal neighborhood into a clustering algorithm for clustering to obtain a centroid set; performing risk diagnosis on the data in the centroid set to extract sample data to obtain an initial screening data set; clustering the remaining data sets in the initial data set according to the initial screening data set and the optimal neighborhood radius to obtain a target screening data set; calculating the risk value of each sample point in the target screening data set, and sorting the risk values ​​to obtain a potential risk user sequence. The technical solution of the embodiment of the present invention can obtain a data sample set after dimensionality reduction by performing a series of processing on the initial data set, thereby reducing the dimension of the data; based on the data sample set, through two sampling operations (the first sampling is: sampling the data sample set to obtain a sampled sample set; the second sampling is: performing risk diagnosis on the data in the centroid set to extract sample data to obtain an initial screening data set) and two clustering operations (the first clustering is: clustering the centroid set through a clustering algorithm based on the adaptive calculation of the optimal neighborhood radius and the number of data objects in the optimal neighborhood; the second clustering is clustering the remaining data sets according to the initial screening data set and the optimal neighborhood radius to obtain a target screening data set) to determine the target screening data set, and determine the potential risk user sequence according to the target screening data set, which not only ensures data accuracy, but also reduces the amount of calculation, and improves the screening accuracy and efficiency of risky users. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0024] Figure 1 A schematic diagram of a process for screening risky users on an Internet of Things platform provided by an embodiment of the present invention;

[0025] Figure 2 A schematic diagram of product types provided by an embodiment of the present invention;

[0026] Figure 3A schematic diagram of a network type provided for an embodiment of the present invention;

[0027] Figure 4 A schematic diagram of the protocol type provided by an embodiment of the present invention;

[0028] Figure 5 A simplified flow chart of adaptively obtaining the optimal neighborhood radius ∈ and the threshold value min_samples of the number of data objects in the optimal neighborhood in the DBSCAN algorithm provided by an embodiment of the present invention;

[0029] Figure 6 An architectural diagram of risky user screening for an Internet of Things platform provided by an embodiment of the present invention;

[0030] Figure 7 A schematic block diagram of a risk user screening device for an Internet of Things platform provided by an embodiment of the present invention;

[0031] Figure 8 A schematic block diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0034] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0035] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0036] As used in this specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0037] See also Figure 1 , Figure 1 1 is a flow chart of a method for screening risky users for an Internet of Things platform provided by an embodiment of the present invention. The method for screening risky users for an Internet of Things platform provided by an embodiment of the present invention can be applied to the Internet of Things platform. For example, the method for screening risky users for an Internet of Things platform can be implemented by a software program configured on the Internet of Things platform. Figure 1 As shown, the method includes the following steps S100-S160.

[0038] S100: Acquire an initial data set of a user, and process the initial data set to obtain a clustering kernel matrix.

[0039] In the embodiment of the present invention, the Internet of Things platform obtains the user's initial data set n*m, where n is the data dimension (product type, network type, protocol type, number of online terminals, data push volume, number of data push exceptions, number of API calls, number of API call exceptions, number of concurrent reports per second from terminals, number of concurrent API calls per second), n=10; m is the data sample (i.e., m valid user sample data). The user's data set attribute conversion rules are defined as follows: I represents the product type, which is a binary number; N represents the network type, which is a binary number; P represents the protocol type, which is a binary number; D represents the number of online terminals of the user; M d Represents the highest daily data push volume of the user in the past 7 days; M fail Represents the number of failed data pushes for users in the past 7 days; G d Represents the highest daily API call volume of the user in the past 7 days; G fail Represents the number of abnormal API calls made by the user in the past 7 days; S tps Represents the highest southbound TPS of the user in the past 7 days; N tps Represents the highest northbound TPS of the user in the past 7 days; Figure 2 As shown, N is as follows Figure 3 As shown, P is as follows Figure 4As shown. After obtaining the initial data set, the initial data set is processed to obtain the aggregation kernel matrix. Specifically, the initial data set is cleaned to obtain the cleaned data set; the non-binary dimensions in the cleaned data set are centralized to obtain the central data set; the central data set is mapped to a high-dimensional space by a Gaussian kernel function to obtain a kernel matrix; the kernel matrix is ​​linearly transformed to obtain the aggregation kernel matrix. More specifically, the n-dimensional data set D = (x (1) ,x (2) ,…,x (m) ) is centered on the dimension of non-binary numbers, as shown in formula (1):

[0040]

[0041] The Gaussian kernel function is shown in formula (2); and based on the Gaussian kernel function, the central data set is mapped into a high-dimensional space, as shown in formula (3), and a kernel matrix K is obtained, as shown in formula (4);

[0042]

[0043]

[0044]

[0045] The kernel matrix K is made more concentrated by linear transformation, as shown in formula (5). In formula (5), l m It is an m*m matrix, and all values ​​in the matrix are 1 / m.

[0046] K′=Kl m K-Kl m +l m K m (5)

[0047] S110, calculating the eigenvalues ​​and eigenvectors of the aggregation kernel matrix, and reducing the dimensionality of the eigenvectors according to the eigenvalues ​​to obtain a data sample set.

[0048] In the embodiment of the present invention, the eigenvalue Λ and eigenvector α of the aggregation kernel matrix K′ are solved, and the eigenvector α is arranged into a matrix from top to bottom in rows according to the size of the corresponding eigenvalue Λ, and the first preset number of rows d (d=2) is taken to form a reduced-dimensional data sample set D′=(z (1) ,z (2) ), where z (i) =(u1,u2). The user data sample set D′ after dimensionality reduction is mapped to a 2D space with u1 and u2 as dimensions.

[0049] S120, extracting M samples from the data sample set to obtain a sampled sample set, and adaptively calculating an optimal neighborhood radius and a threshold value of the number of data objects in the optimal neighborhood according to the sampled sample set through a clustering algorithm, wherein M≥1.

[0050] In the embodiment of the present invention, M samples are randomly selected from the data sample set D' to form a sample set S', where M = m*s, s is a sampling factor, s∈(0,1). The distance Dis between each data point in the sample set and its pth nearest neighbor data point is calculated. p , where 1≤p≤M; calculate Dis of all data points in the sample set p The average value of the nearest neighbor distance for different p values ​​is obtained, and a candidate neighborhood distance parameter set is formed based on the average nearest neighbor distance for different p values; the threshold value of the number of data objects in the neighborhood is adaptively determined according to the candidate neighborhood distance parameter set; the optimal neighborhood radius and the threshold value of the number of data objects in the optimal neighborhood are calculated by the clustering algorithm according to the candidate neighborhood distance parameter set and the threshold value of the number of data objects in the neighborhood, wherein the clustering algorithm is the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm. Specifically, the candidate neighborhood distance and the threshold value of the number of data objects corresponding to p=i obtained from the candidate neighborhood distance parameter set and the threshold value of the number of data objects in the neighborhood are input into the DBSCAN algorithm to divide the number of clusters J corresponding to p=i i , where 1≤i≤M; if J appears for the first time i =J i-1 =J i-2 , then J i As the optimal number of clusters J opt ; i is increased by 1, and the candidate neighborhood distance and the data object number threshold corresponding to p = i obtained from the candidate neighborhood distance parameter set and the neighborhood data object number threshold are input into the DBSCAN algorithm to divide the number of clusters J corresponding to p = i i Steps until J i Not equal to J opt So far, the p value at this time is taken as the optimal p value p opt , p opt The corresponding candidate neighborhood distance and the data object number threshold are used as the optimal neighborhood radius and the data object number threshold in the optimal neighborhood.

[0051] For ease of understanding, the specific steps of adaptively calculating the optimal neighborhood radius ∈ and the threshold value min_samples of the number of data objects in the optimal neighborhood based on the sample set S' are introduced as follows:

[0052] (1) By calculating the distance Dis between each data point in the sample set S′ and its pth nearest neighbor data point p , and for all data points Dis p Find the average value, denoted as The average nearest neighbor distances for different p values ​​form a candidate neighborhood distance ∈ parameter set, as shown in formula (6):

[0053]

[0054] (2) Adaptively determine the threshold of the number of data objects in the neighborhood according to the candidate neighborhood distance ∈ parameter set, as shown in formula (7). In formula (7), d is the dimension of the sample set S′, P i is the number of ∈ neighborhood objects of the i-th object, and m*s is the total number of objects in the dataset S′.

[0055]

[0056] (3) Based on the sample set S′, the candidate neighborhood distance ∈ parameter and min_samples parameter corresponding to p = 1 are input into the DBSCAN algorithm to divide and obtain the corresponding number of clusters J1.

[0057] (4) Increase the p value with a step size of 1. When the p value is taken, the corresponding candidate neighborhood distance ∈ parameter and min_samples parameter are input into the DBSCAN algorithm to divide the corresponding number of clusters J. i .

[0058] (5) Repeat step (4) until the number of clusters after three divisions is the same for the first time, i.e., J i =J i-1 =J i-2 , at this time record J i is the optimal number of clusters J opt .

[0059] (6) Repeat step (4) and continue to increase the p value with a step size of 1 until the number of clusters obtained by DBSCAN partitioning is not equal to the optimal number of clusters J opt . Take the p value at this time as the optimal p value p opt , then the candidate neighborhood distance ∈ parameter and min_samples parameter corresponding to the p value are the optimal parameters to obtain the optimal neighborhood radius ∈ and the threshold value min_samples of the number of data objects in the optimal neighborhood.

[0060] See also Figure 5, Figure 5 A flowchart of adaptively obtaining the optimal neighborhood radius ∈ and the threshold value min_samples of the number of data objects in the optimal neighborhood in the DBSCAN algorithm provided by the embodiment of the present invention is shown in FIG. Figure 5 In , the data set is a sample set.

[0061] S130: Input the optimal neighborhood radius and the data object number threshold in the optimal neighborhood into a clustering algorithm for clustering to obtain a centroid set.

[0062] In the embodiment of the present invention, the optimal neighborhood radius and the data object number threshold in the optimal neighborhood are input into the DBSCAN algorithm for clustering to obtain a centroid set C. It should be noted that in addition to the centroid set C, a noise set N is also obtained.

[0063] S140, performing risk diagnosis on the data in the centroid set to extract sample data to obtain an initial screening data set.

[0064] In an embodiment of the present invention, the centroid risk value r (r = |u1| + |u2|) of each centroid in the centroid set is calculated, and the centroid risk value is corresponded to the cluster where the centroid is located, and the centroid risk value r is sorted from large to small; the sample data of the first k clusters in the cluster where the sorted centroid risk value is greater than the set risk diagnosis threshold is selected to obtain the initial screening data set.

[0065] S150: Cluster the remaining data sets in the initial data set according to the initial screening data set and the optimal neighborhood radius to obtain a target screening data set.

[0066] In an embodiment of the present invention, the remaining data set in the initial data set D is obtained, and the clustering distance between the data in the remaining data set and the centroid corresponding to the sample data of k clusters in the initial screening data set is calculated; if the clustering distance is greater than the optimal neighborhood radius, indicating that clustering is not required, the data in the remaining data set corresponding to the clustering distance is filtered; if the distance is not greater than the optimal neighborhood radius, indicating that clustering is required, the data in the remaining data corresponding to the clustering distance is allocated to the cluster closest to the centroid in the initial screening data set, thereby obtaining the target screening data set R.

[0067] S160: Calculate the risk value of each sample point in the target screening data set, and sort the risk values ​​to obtain a potential risk user sequence.

[0068] In the embodiment of the present invention, the distance from each sample point in the target screening data set R to the centroid point μ of its corresponding cluster is calculated. i The Manhattan distance s (i), as shown in formula (8)

[0069] s (i) =u1 (i) -μ1 (i) |+|u2 (i) -μ2 (i) | (8)

[0070] Understandably, the s corresponding to the sample point (i) The smaller the value, the greater the risk. (i) By sorting the values ​​in ascending order, we can obtain the final risk potential ranking of users, that is, the sequence of potential risk users.

[0071] See also Figure 6 , Figure 6 The architecture diagram of the risk user screening for the Internet of Things platform provided by the embodiment of the present invention is Figure 6 In the process, read the overall data set (initial data set) - > introduce the kernel function to map it to the high-dimensional space - > data dimension reduction - > first sampling - > perform the first-stage DBSCAN clustering on the sampled data - > risk diagnosis - > second sampling - > second-stage clustering of the remaining sample data - > calculate the risk value - > sort to obtain the risk customer sequence (potential risk user sequence).

[0072] It should be noted that, in this embodiment, by introducing the Gaussian kernel function, the data is nonlinearly mapped to a high-dimensional space to obtain the kernel matrix K, so that the original nonlinearly distributed data can be decomposed into eigenvalues, data dimensionality can be reduced, and the computational cost can be reduced. At the same time, the problem that DBSCAN clustering is not suitable for processing high-dimensional data can also be solved; multi-stage clustering is adopted, the first stage is random sampling and combined with DBSCAN clustering; the second stage is to complete the data secondary sampling through risk diagnosis, and the secondary clustering is completed based on the sample centroid. While ensuring the accuracy of the data, the amount of calculation is reduced in stages, thereby achieving efficient clustering for large-scale data and improving the accuracy and efficiency of risk user screening. It should also be noted that in this embodiment, DBSCAN clustering is completed by adaptively, without setting parameters in advance, and the optimal neighborhood radius and the threshold value of the number of data objects in the optimal neighborhood can be adaptively obtained based on the data sample situation, and the optimal parameters are adaptively applied to the DBSCAN algorithm to cluster the data samples based on density.

[0073] Figure 7 FIG. 2 is a schematic block diagram of a risk user screening device 200 for an Internet of Things platform provided by an embodiment of the present invention. Figure 7As shown, corresponding to the above risk user screening method for the Internet of Things platform, the present invention also provides a risk user screening device 200 for the Internet of Things platform. The risk user screening device 200 for the Internet of Things platform includes a unit for executing the above risk user screening method for the Internet of Things platform. Figure 7 The risk user screening device 200 for the Internet of Things platform includes an acquisition processing unit 201, a calculation dimension reduction unit 202, a first extraction unit 203, a first clustering unit 204, a second extraction unit 205, a second clustering unit 206 and a calculation sorting unit 207.

[0074] The acquisition processing unit 201 is used to acquire the user's initial data set, and process the initial data set to obtain a clustering kernel matrix; the calculation dimension reduction unit 202 is used to calculate the eigenvalues ​​and eigenvectors of the clustering kernel matrix, and reduce the dimensionality of the eigenvector according to the eigenvalues ​​to obtain a data sample set; the first extraction unit 203 is used to extract M samples from the data sample set to obtain a sampled sample set, and adaptively calculate the optimal neighborhood radius and the threshold value of the number of data objects in the optimal neighborhood according to the sampled sample set through a clustering algorithm, wherein M≥1; the first clustering unit 204 is used to The optimal neighborhood radius and the data object number threshold in the optimal neighborhood are input into the clustering algorithm for clustering to obtain a centroid set; the second extraction unit 205 is used to perform risk diagnosis on the data in the centroid set to extract sample data to obtain an initial screening data set; the second clustering unit 206 is used to cluster the remaining data sets in the initial data set according to the initial screening data set and the optimal neighborhood radius to obtain a target screening data set; the calculation and sorting unit 207 is used to calculate the risk value of each sample point in the target screening data set, and sort the risk values ​​to obtain a potential risk user sequence.

[0075] In some embodiments, such as the present embodiment, the acquisition processing unit 201 includes a cleaning unit, a centering unit, a mapping unit, and a linear transformation unit.

[0076] Among them, the cleaning unit is used to clean the initial data set to obtain a cleaned data set; the centralization unit is used to centralize the non-binary dimensions in the cleaned data set to obtain a central data set; the mapping unit is used to map the central data set to a high-dimensional space through a Gaussian kernel function to obtain a kernel matrix; the linear transformation unit is used to perform a linear transformation on the kernel matrix to obtain the clustered kernel matrix.

[0077] In some embodiments, such as the present embodiment, the computing dimension reduction unit 202 includes an arrangement unit and a dimension reduction unit.

[0078] Among them, the arrangement unit is used to arrange the eigenvectors into an initial matrix by rows according to the size of the eigenvalues; the dimension reduction unit is used to select the first preset number of rows in the initial matrix to form the data sample set after dimensionality reduction.

[0079] In some embodiments, such as the present embodiment, the first extraction unit 203 includes a first calculation unit, a second calculation unit, an adaptive unit, and a third calculation unit.

[0080] The first calculation unit is used to calculate the distance Dis between each data point in the sample set and its pth nearest neighbor data point. p , where 1≤p≤M; the second calculation unit is used to calculate Dis of all data points in the sample set. p The average value of the nearest neighbor distance with different p values ​​is obtained, and a candidate neighborhood distance parameter set is formed based on the average nearest neighbor distance with different p values; the adaptive unit is used to adaptively determine the threshold value of the number of data objects in the neighborhood according to the candidate neighborhood distance parameter set; the third calculation unit is used to calculate the optimal neighborhood radius and the threshold value of the number of data objects in the optimal neighborhood through the clustering algorithm according to the candidate neighborhood distance parameter set and the threshold value of the number of data objects in the neighborhood.

[0081] In some embodiments, such as the present embodiment, the third computing unit includes an input partitioning unit, an as unit, and a return execution unit.

[0082] The input partitioning unit is used to input the candidate neighborhood distance corresponding to p=i and the data object number threshold obtained from the candidate neighborhood distance parameter set and the data object number threshold in the neighborhood into the DBSCAN algorithm to partition the cluster number J corresponding to p=i. i , where 1≤i≤M; the unit is used for the first occurrence of J i =J i-1 =J i-2 , then J i As the optimal number of clusters J opt The return execution unit is used to add 1 to i, and return to execute the candidate neighborhood distance corresponding to p = i and the data object number threshold obtained from the candidate neighborhood distance parameter set and the neighborhood data object number threshold to the DBSCAN algorithm to divide the cluster number J corresponding to p = i i Steps until J i Not equal to J opt So far, the p value at this time is taken as the optimal p value p opt , p optThe corresponding candidate neighborhood distance and the data object number threshold are used as the optimal neighborhood radius and the data object number threshold in the optimal neighborhood.

[0083] In some embodiments, such as this embodiment, the second extraction unit 205 includes a fourth calculation unit and a selection unit.

[0084] Among them, the fourth calculation unit is used to calculate the centroid risk value of each centroid in the centroid set, and correspond the centroid risk value to the cluster where the centroid is located, and sort the centroid risk values ​​from large to small; the selection unit is used to select the sample data of the first k clusters in the cluster where the sorted centroid risk value is greater than the set risk diagnosis threshold to obtain the initial screening data set.

[0085] In some embodiments, such as this embodiment, the second clustering unit 206 includes a fifth calculation unit, a filtering unit, and an allocation unit.

[0086] Among them, the fifth calculation unit is used to obtain the remaining data set in the initial data set, and calculate the clustering distance between the data in the remaining data set and the centroid corresponding to the sample data of k clusters in the initial screening data set; the filtering unit is used to filter the data in the remaining data set corresponding to the clustering distance if the clustering distance is greater than the optimal neighborhood radius; the allocation unit is used to allocate the data in the remaining data corresponding to the clustering distance to the cluster closest to the centroid in the initial screening data set to obtain the target screening data set if the distance is not greater than the optimal neighborhood radius.

[0087] It should be noted that technical personnel in the relevant field can clearly understand that the specific implementation process of the above-mentioned risk user screening device 200 and each unit for the Internet of Things platform can refer to the corresponding description in the aforementioned method embodiment, and for the convenience and conciseness of the description, it will not be repeated here.

[0088] The above-mentioned risk user screening device for the Internet of Things platform can be implemented in the form of a computer program. The computer program can be used in Figure 8 Runs on the computer device shown.

[0089] See also Figure 8 , Figure 8 900 is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device 900 is a device with an Internet of Things platform.

[0090] See also Figure 8The computer device 900 includes a processor 902 , a memory and an interface 907 connected via a system bus 901 , wherein the memory may include a storage medium 903 and an internal memory 904 .

[0091] The storage medium 903 can store an operating system 9031 and a computer program 9032. When the computer program 9032 is executed, the processor 902 can execute the above-mentioned risk user screening method for the Internet of Things platform.

[0092] The processor 902 is used to provide computing and control capabilities to support the operation of the entire computer device 900 .

[0093] The internal memory 904 provides an environment for the operation of the computer program 9032 in the storage medium 903. When the computer program 9032 is executed by the processor 902, the processor 902 can execute a risk user screening method for the Internet of Things platform.

[0094] The interface 905 is used to communicate with other devices. Those skilled in the art will appreciate that Figure 8 The structure shown in the figure is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device 900 to which the solution of the present application is applied. The specific computer device 900 may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0095] The processor 902 is used to run a computer program 9032 stored in the memory to implement the process steps of the embodiment of the above method.

[0096] It should be understood that in the embodiment of the present application, the processor 902 may be a central processing unit (CPU), and the processor 902 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0097] It can be understood by those skilled in the art that all or part of the processes in the method for implementing the above embodiment can be completed by instructing related hardware through a computer program. The computer program can be stored in a storage medium, which is a computer-readable storage medium. The computer program is executed by at least one processor in the wireless communication system to implement the process steps of the embodiment of the above method.

[0098] Therefore, the present invention also provides a storage medium. The storage medium may be a computer-readable storage medium. The storage medium stores a computer program. When the computer program is executed by a processor, the processor executes any embodiment of the above-mentioned risk user screening method for an Internet of Things platform.

[0099] The storage medium may be a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk, or an optical disk, etc., which are computer-readable storage media that can store program codes.

[0100] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, wireless communication software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0101] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of each unit is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0102] The steps in the method of the embodiment of the present invention can be adjusted in order, combined and deleted according to actual needs. The units in the device of the embodiment of the present invention can be combined, divided and deleted according to actual needs. In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0103] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The wireless communication software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal wireless communication, terminal, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention.

[0104] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0105] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

[0106] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for screening risky users on an Internet of Things platform, characterized in that: include: Acquire an initial data set of a user, and process the initial data set to obtain a clustering kernel matrix; Calculating the eigenvalues ​​and eigenvectors of the aggregation kernel matrix, and reducing the dimensionality of the eigenvectors according to the eigenvalues ​​to obtain a data sample set; Extracting M samples from the data sample set to obtain a sample set, and adaptively calculating an optimal neighborhood radius and a threshold value of the number of data objects in the optimal neighborhood according to the sample set through a clustering algorithm, wherein M≥1; Inputting the optimal neighborhood radius and the data object number threshold in the optimal neighborhood into a clustering algorithm for clustering to obtain a centroid set; Performing risk diagnosis on the data in the centroid set to extract sample data to obtain an initial screening data set; Clustering the remaining data sets in the initial data set according to the initial screening data set and the optimal neighborhood radius to obtain a target screening data set; Calculating the risk value of each sample point in the target screening data set, and sorting the risk values ​​to obtain a potential risk user sequence; The step of clustering the remaining data sets in the initial data set according to the initial screening data set and the optimal neighborhood radius to obtain a target screening data set includes: Obtaining the remaining data sets in the initial data set, and calculating the clustering distance between the data in the remaining data sets and the centroids corresponding to the sample data of the k clusters in the initial screening data set; If the cluster distance is greater than the optimal neighborhood radius, filtering the data in the remaining data set corresponding to the cluster distance; If the cluster distance is not greater than the optimal neighborhood radius, the data in the remaining data corresponding to the cluster distance is allocated to the cluster closest to the centroid in the initial screening data set to obtain the target screening data set.

2. The method for screening risky users for the Internet of Things platform according to claim 1 is characterized in that: The processing of the initial data set to obtain a clustered kernel matrix includes: Cleaning the initial data set to obtain a cleaned data set; Centralizing the non-binary dimensions in the cleaned data set to obtain a central data set; Mapping the central data set to a high-dimensional space by a Gaussian kernel function to obtain a kernel matrix; The kernel matrix is ​​linearly transformed to obtain the aggregation kernel matrix.

3. The risk user screening method for the Internet of Things platform according to claim 1 is characterized in that: The step of reducing the dimension of the feature vector according to the feature value to obtain a data sample set includes: Arranging the eigenvectors into an initial matrix by rows according to the magnitude of the eigenvalues; The first preset number of rows in the initial matrix are selected to form the data sample set after dimensionality reduction.

4. The method for screening risky users for the Internet of Things platform according to claim 1 is characterized in that: The step of adaptively calculating the optimal neighborhood radius and the threshold value of the number of data objects in the optimal neighborhood by using a clustering algorithm according to the sample set includes: Calculate the distance Dis between each data point in the sample set and its pth nearest neighbor data point p , where 1≤p≤M; Calculate the Dis of all data points in the sample set p The average value of the nearest neighbor distance with different p values ​​is obtained, and the candidate neighborhood distance parameter set is formed based on the average nearest neighbor distance with different p values; Adaptively determine a threshold value of the number of data objects in a neighborhood according to the candidate neighborhood distance parameter set; The optimal neighborhood radius and the threshold value of the number of data objects in the optimal neighborhood are calculated by the clustering algorithm according to the candidate neighborhood distance parameter set and the threshold value of the number of data objects in the neighborhood.

5. The method for screening risky users for the Internet of Things platform according to claim 4 is characterized in that: The step of calculating the optimal neighborhood radius and the data object number threshold in the optimal neighborhood by the clustering algorithm according to the candidate neighborhood distance parameter set and the data object number threshold in the neighborhood includes: The candidate neighborhood distance corresponding to p=i and the data object number threshold obtained from the candidate neighborhood distance parameter set and the data object number threshold in the neighborhood are input into the DBSCAN algorithm to divide the cluster number J corresponding to p=i i , where 1≤i≤M; If the first occurrence of J i =J i-1 =J i-2 , then J i As the optimal number of clusters J opt ; Add 1 to i, and return to the execution of the candidate neighborhood distance parameter set and the threshold value of the number of data objects in the neighborhood corresponding to p = i, and input the candidate neighborhood distance and the threshold value of the number of data objects into the DBSCAN algorithm to divide the number of clusters J corresponding to p = i i Steps until J i Not equal to J opt So far, the p value at this time is taken as the optimal p value p opt , p opt The corresponding candidate neighborhood distance and the data object number threshold are used as the optimal neighborhood radius and the data object number threshold in the optimal neighborhood.

6. The method for screening risky users for the Internet of Things platform according to claim 1 is characterized in that: The step of performing risk diagnosis on the data in the centroid set to extract sample data to obtain an initial screening data set includes: Calculate the centroid risk value of each centroid in the centroid set, correspond the centroid risk value to the cluster where the centroid belongs, and sort the centroid risk values ​​from large to small; The sample data of the first k clusters in the cluster where the centroid risk value after sorting is greater than the set risk diagnosis threshold are selected to obtain the initial screening data set.

7. A risk user screening device for an Internet of Things platform, characterized in that: include: An acquisition processing unit, used for acquiring an initial data set of a user, and processing the initial data set to obtain a clustering kernel matrix; A dimension reduction calculation unit is used to calculate the eigenvalues ​​and eigenvectors of the aggregation kernel matrix, and reduce the dimension of the eigenvector according to the eigenvalues ​​to obtain a data sample set; A first extraction unit is used to extract M samples from the data sample set to obtain a sample set, and adaptively calculate an optimal neighborhood radius and a threshold value of the number of data objects in the optimal neighborhood according to the sample set through a clustering algorithm, wherein M≥1; A first clustering unit, used for inputting the optimal neighborhood radius and the data object number threshold in the optimal neighborhood into a clustering algorithm for clustering to obtain a centroid set; A second extraction unit is used to perform risk diagnosis on the data in the centroid set to extract sample data to obtain an initial screening data set; a second clustering unit, configured to cluster the remaining data sets in the initial data set according to the initial screening data set and the optimal neighborhood radius to obtain a target screening data set; A calculation and sorting unit, used to calculate the risk value of each sample point in the target screening data set, and sort the risk values ​​to obtain a potential risk user sequence; Wherein, the second clustering unit comprises: a fifth calculation unit, configured to obtain a remaining data set in the initial data set, and calculate a cluster distance between the data in the remaining data set and the centroids corresponding to the sample data of the k clusters in the initial screening data set; A filtering unit, configured to filter the data in the remaining data set corresponding to the clustering distance if the clustering distance is greater than the optimal neighborhood radius; An allocating unit is used to allocate the data in the remaining data corresponding to the clustering distance to the cluster closest to the centroid in the initial screening data set to obtain the target screening data set if the clustering distance is not greater than the optimal neighborhood radius.

8. A computer device, characterized in that: The computer device is built with an Internet of Things platform, and the computer device includes a memory and a processor. The memory stores a computer program, and the processor implements the method according to any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 can be implemented.