Distributed personnel relationship mining evaluation method and device

Abstract

The present invention discloses a distributed personnel relationship mining and evaluation method and device, the method comprising: obtaining a personnel relationship training data set; using the personnel relationship training data set to train a personnel relationship mining model; obtaining personnel relationship detection data; preprocessing the personnel relationship detection data to obtain preprocessed personnel relationship detection data; using the personnel relationship mining model to process the preprocessed personnel relationship detection data to obtain personnel category mining result information; evaluating the personnel category mining result information to obtain a personnel relationship evaluation result value; the personnel relationship evaluation result value is used to characterize the threat level of personnel relationships. The present invention solves the problems existing in network personnel tracking technology, such as unclear personnel relationship analysis purpose, unclear required data fields, inaccurate mining and analysis algorithms, and inability to accurately evaluate personnel threats.

Description

Technical Field

[0001] The present invention relates to the field of information security technology, and in particular to a distributed personnel relationship mining and evaluation method and device. Background Art

[0002] At present, in the field of information security, mining the identity information and association relationships of network personnel can effectively support applications such as key strike target recommendations, key target analysis breakthroughs, and important intelligence interception and acquisition. The existing network personnel tracking technology has problems such as unclear personnel relationship analysis purposes, unclear required data fields, inaccurate mining and analysis algorithms, and inability to accurately assess personnel threats, which leads to problems such as the analysis results being difficult to support practical applications. Summary of the invention

[0003] In view of the problems existing in the existing network personnel tracking technology, such as unclear personnel relationship analysis purpose, unclear required data fields, inaccurate mining and analysis algorithms, and inability to accurately assess personnel threats, which make it difficult for the analysis results to support practical applications, the present invention discloses a distributed personnel relationship mining and evaluation method and device. The method defines personnel in detail, supports the extraction of key elements of personnel, solves the problem of insufficient personnel tracking data foundation, performs data analysis from four directions: real identity category information, virtual identity category information, trust relationship category information, and activity trajectory category information, and realizes relationship mining and evaluation of personnel on the Internet.

[0004] The present invention discloses a distributed personnel relationship mining and evaluation method, comprising:

[0005] S1, obtaining a personnel relationship training data set; the personnel relationship training data set includes personnel relationship training data and corresponding personnel category label data; the personnel category label data includes real identity category label information, virtual identity category label information, trust relationship category label information, and activity trajectory category label information;

[0006] S2, using the personnel relationship training data set to train a personnel relationship mining model;

[0007] S3, obtaining personnel relationship detection data;

[0008] S4, preprocessing the personnel relationship detection data to obtain preprocessed personnel relationship detection data;

[0009] S5, using the personnel relationship mining model to process the pre-processed personnel relationship detection data to obtain personnel category mining result information; the personnel category mining result information includes real identity category information, virtual identity category information, trust relationship category information, and activity trajectory category information;

[0010] S6, evaluating and processing the personnel category mining result information to obtain a personnel relationship evaluation result value; the personnel relationship evaluation result value is used to represent the threat level of the personnel relationship.

[0011] The personnel relationship mining model includes a feature extraction network and a classification network;

[0012] The feature extraction network is connected to the classification network and includes a first processing module, a second processing module, a third processing module and a fourth processing module, which are used to perform feature extraction processing on the input data to obtain feature data;

[0013] The first processing module includes an input module, a first convolution module, a first activation module, a first residual module, and a first normalization module; the first port of the first processing module is the input end of the input module; the second port of the first processing module is the output end of the first normalization module; the first port of the first processing module is used as the input end of the feature extraction network to receive input data; the second port of the first processing module is connected to the first port of the second processing module; in the first processing module, the input module, the first convolution module, the first activation module, the first residual module, and the first normalization module are connected in sequence;

[0014] The second processing module includes a second convolution module, a second activation module, and a second normalization module; the first port of the second processing module is the input end of the second convolution module; the second port of the second processing module is the output end of the second normalization module; the second port of the second processing module is connected to the first port of the third processing module; in the second processing module, the second convolution module, the second activation module, and the second normalization module are connected in sequence;

[0015] The third processing module includes a third convolution module, a third activation module, and a third residual module; the first port of the third processing module is the input end of the third convolution module; the second port of the third processing module is the output end of the third residual module; the second port of the third processing module is connected to the first port of the fourth processing module; in the third processing module, the third convolution module, the third activation module, and the third residual module are connected in sequence;

[0016] The fourth processing module includes a fourth convolution module, a fourth activation module, and a fourth residual module; the first port of the fourth processing module is the input end of the fourth convolution module; the second port of the fourth processing module is the output end of the fourth residual module; the second port of the fourth processing module is used as the output end of the feature extraction network to output feature data; in the fourth processing module, the fourth convolution module, the fourth activation module, and the fourth residual module are connected in sequence;

[0017] The classification network is used to perform relationship classification and recognition processing on the feature data to obtain personnel category mining result information or personnel category prediction result information, including a first relationship feature extraction module, a first encoding module, a first decoding module, a first feature reconstruction module, a second relationship feature extraction module, a second encoding module, a second decoding module, a second feature reconstruction module and a full connection module; the first encoding module includes a first personnel category encoding module and a first personnel trajectory encoding module; the second encoding module includes a second personnel category encoding module and a second personnel trajectory encoding module;

[0018] The input end of the first relationship feature extraction module and the input end of the second relationship feature extraction module are both connected to the second port of the fourth processing module;

[0019] The output end of the first relationship feature extraction module is connected to the input end of the first encoding module; the output ends of the first personnel category encoding module and the second personnel trajectory encoding module are connected to the input end of the first decoding module; the output ends of the second personnel category encoding module and the first personnel trajectory encoding module are connected to the input end of the second decoding module; the output end of the first decoding module is connected to the input end of the first feature reconstruction module; the output end of the second decoding module is connected to the input end of the second feature reconstruction module; the output end of the first feature reconstruction module and the output end of the second feature reconstruction module are both connected to the input end of the full connection module; the output end of the full connection module is used to output the personnel category mining result information or the personnel category prediction result information obtained by the personnel relationship mining model.

[0020] The feature extraction network of the personnel relationship mining model realizes the effective extraction of useful information of personnel relationship information and the effective suppression of irrelevant information by setting a multi-level convolution processing module; the classification network ensures the accuracy of the relationship classification results by adopting multi-level relationship feature extraction and encoding personnel categories and personnel trajectories respectively.

[0021] The training process of the personnel relationship mining model using the personnel relationship training data set includes:

[0022] S21, initialize the number of training iterations;

[0023] S22, taking each personnel relationship training data in the personnel relationship training data set as input data, inputting into the personnel relationship mining model, and obtaining personnel category prediction result information;

[0024] S23, performing difference calculation processing on the personnel category label data corresponding to the input data and the personnel category prediction result information to obtain a difference value;

[0025] S24, judging whether the difference value satisfies a convergence condition, and obtaining a first judgment result;

[0026] When the first judgment result is no, judging whether the training iteration number value is equal to the training number threshold, and obtaining a second judgment result;

[0027] When the second judgment result is no, determining that the model training state does not meet the training termination condition;

[0028] When the second judgment result is yes, determining that the model training state satisfies the training termination condition;

[0029] When the first judgment result is yes, determining that the model training state satisfies a training termination condition;

[0030] When the model training state does not meet the training termination condition, the classification network parameters are updated using the parameter update model, the training iteration number value is increased by 1, and S22 is executed;

[0031] When the model training state satisfies the training termination condition, the training process of the personnel relationship mining model is completed to obtain a trained personnel relationship mining model.

[0032] The parameter updating model is:

[0033]

[0034] θ←θ+v;

[0035] In the formula, x (i) is the i-th data in the personnel relationship training data set, is the loss function for the j-th category of personnel category label data, v is the parameter update value, θ is the parameter of the classification network, η is the initial parameter learning rate, α is the momentum angle parameter, 0≤α≤1, It means to find the partial derivative of variable θ, f(x (i) ; θ) represents the personnel category label data calculated by the personnel relationship mining model for the i-th data of the personnel relationship training data set, and f(·) is the calculation function corresponding to the personnel relationship mining model.

[0036] The preprocessing of the personnel relationship detection data to obtain the preprocessed personnel relationship detection data includes:

[0037] S41, performing data cleaning processing on the personnel relationship detection data to obtain cleaned data;

[0038] S42, performing data range limiting processing on the cleaned data to obtain limited data;

[0039] S43, normalizing the limited data to obtain pre-processed personnel relationship detection data.

[0040] The evaluating and processing the personnel category mining result information to obtain the personnel relationship evaluation result includes:

[0041] S61, obtaining personnel category mining result information within N time periods;

[0042] S62, quantifying the personnel category mining result information within the N time periods to obtain a personnel category mining result matrix A;

[0043] The row vectors in the personnel category mining result matrix A are vectors obtained by quantizing the personnel category mining result information within a time period;

[0044] S63, performing feature vector calculation processing on the personnel category mining result matrix A to obtain a personnel category feature vector V;

[0045] S64, using the personnel category feature vector, performing weighted sum processing on the personnel category mining result matrix to obtain a personnel relationship evaluation result value.

[0046] The process of performing feature vector calculation on the personnel category mining result matrix A to obtain the personnel category feature vector V includes:

[0047] S631, perform QR decomposition on the personnel category mining result matrix A to obtain

[0048] A=U s V s T ,

[0049] Among them, U s 、V s denote the column orthogonal matrix and upper triangular matrix of matrix A respectively;

[0050] S632, using U s All diagonal elements of are used to construct the personnel category feature vector V.

[0051] According to a second aspect of an embodiment of the present invention, a distributed personnel relationship mining and evaluation device is disclosed, the device comprising:

[0052] A memory storing executable program code;

[0053] a processor coupled to the memory;

[0054] The processor calls the executable program code stored in the memory to execute the distributed personnel relationship mining and evaluation method.

[0055] According to a third aspect of an embodiment of the present invention, a computer storable medium is disclosed, wherein the computer storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the distributed personnel relationship mining and evaluation method.

[0056] According to a fourth aspect of the embodiments of the present invention, an information data processing terminal is disclosed, wherein the information data processing terminal is used to implement the distributed personnel relationship mining and evaluation method.

[0057] The beneficial effects of the present invention are:

[0058] In order to address the problems of unclear purpose of personnel relationship analysis, unclear required data fields, inaccurate mining and analysis algorithms, inability to assess personnel threats, etc., which make it difficult for analysis results to support practical applications, the personnel tracking technology based on distributed data mining accurately defines the personnel information fields, and during the analysis process, it splits the data and tasks into various computing nodes and distributes them to each computing node. Data mining is performed in parallel by each computing node, which greatly reduces the time consumption of data mining processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a flow chart for implementing the method of the present invention. DETAILED DESCRIPTION

[0060] In order to better understand the content of the present invention, an embodiment is given here.

[0061] In a first aspect of an embodiment of the present invention, a distributed personnel relationship mining and evaluation method is disclosed. Figure 1 As shown, including:

[0062] S1, obtaining a personnel relationship training data set; the personnel relationship training data set includes personnel relationship training data and corresponding personnel category label data; the personnel category label data includes real identity category label information, virtual identity category label information, trust relationship category label information, and activity trajectory category label information;

[0063] S2, using the personnel relationship training data set to train a personnel relationship mining model;

[0064] S3, obtaining personnel relationship detection data;

[0065] S4, preprocessing the personnel relationship detection data to obtain preprocessed personnel relationship detection data;

[0066] S5, using a personnel relationship mining model to process the pre-processed personnel relationship detection data to obtain personnel category mining result information; the personnel category mining result information includes real identity category information, virtual identity category information, trust relationship category information, and activity trajectory category information;

[0067] S6, evaluating the personnel category mining result information to obtain a personnel relationship evaluation result value. The personnel relationship evaluation result value is used to represent the threat level of the personnel relationship.

[0068] The personnel relationship training data set can be constructed using known Internet security news, government public information, and information security knowledge graphs; specifically, texts, pictures, audios, etc. in known Internet security news, government public information, and information security knowledge graphs can be used as personnel relationship training data, and the personnel information in the above information can be used as corresponding personnel category label data; the personnel information includes real identity category information, virtual identity category information, trust relationship category information, and activity trajectory category information; the real identity category information can take values ​​such as network security management personnel, network security technicians, general personnel, etc.; the virtual identity category information can take values ​​such as threat personnel, detection personnel, tourists, etc.; the trust relationship category information can take values ​​such as trust, general trust, distrust, and danger; the activity trajectory category information can take values ​​such as danger, general danger, general safety, and safety.

[0069] The obtaining of personnel relationship detection data includes: obtaining personnel relationship detection data involving personnel relationships from information security professional websites, government databases, and information security knowledge graphs;

[0070] The personnel relationship mining model includes a feature extraction network and a classification network;

[0071] The feature extraction network is connected to the classification network and includes a first processing module, a second processing module, a third processing module and a fourth processing module, which are used to perform feature extraction processing on the input data to obtain feature data;

[0072] The first processing module includes an input module, a first convolution module, a first activation module, a first residual module, and a first normalization module; the first port of the first processing module is the input end of the input module; the second port of the first processing module is the output end of the first normalization module; the first port of the first processing module is used as the input end of the feature extraction network to receive input data; the second port of the first processing module is connected to the first port of the second processing module; in the first processing module, the input module, the first convolution module, the first activation module, the first residual module, and the first normalization module are connected in sequence;

[0073] The second processing module includes a second convolution module, a second activation module, and a second normalization module; the first port of the second processing module is the input end of the second convolution module; the second port of the second processing module is the output end of the second normalization module; the second port of the second processing module is connected to the first port of the third processing module; in the second processing module, the second convolution module, the second activation module, and the second normalization module are connected in sequence;

[0074] The third processing module includes a third convolution module, a third activation module, and a third residual module; the first port of the third processing module is the input end of the third convolution module; the second port of the third processing module is the output end of the third residual module; the second port of the third processing module is connected to the first port of the fourth processing module; in the third processing module, the third convolution module, the third activation module, and the third residual module are connected in sequence;

[0075] The fourth processing module includes a fourth convolution module, a fourth activation module, and a fourth residual module; the first port of the fourth processing module is the input end of the fourth convolution module; the second port of the fourth processing module is the output end of the fourth residual module; the second port of the fourth processing module is used as the output end of the feature extraction network to output feature data; in the fourth processing module, the fourth convolution module, the fourth activation module, and the fourth residual module are connected in sequence;

[0076] The classification network is used to perform relationship classification and recognition processing on the feature data to obtain personnel category mining result information or personnel category prediction result information, including a first relationship feature extraction module, a first encoding module, a first decoding module, a first feature reconstruction module, a second relationship feature extraction module, a second encoding module, a second decoding module, a second feature reconstruction module and a full connection module; the first encoding module includes a first personnel category encoding module and a first personnel trajectory encoding module; the second encoding module includes a second personnel category encoding module and a second personnel trajectory encoding module;

[0077] The input end of the first relationship feature extraction module and the input end of the second relationship feature extraction module are both connected to the second port of the fourth processing module to receive the feature data;

[0078] The output end of the first relationship feature extraction module is connected to the input end of the first encoding module; the output ends of the first personnel category encoding module and the second personnel trajectory encoding module are connected to the input end of the first decoding module; the output ends of the second personnel category encoding module and the first personnel trajectory encoding module are connected to the input end of the second decoding module; the output end of the first decoding module is connected to the input end of the first feature reconstruction module; the output end of the second decoding module is connected to the input end of the second feature reconstruction module; the output end of the first feature reconstruction module and the output end of the second feature reconstruction module are both connected to the input end of the full connection module; the output end of the full connection module is used to output the personnel category mining result information or the personnel category prediction result information obtained by the personnel relationship mining model;

[0079] The activation module is used to implement an activation function; the activation function may be a sigmoid function;

[0080] The residual module can be implemented using a Resnet network;

[0081] The normalization module can be implemented by using the Quantile Normalization function in phthon.

[0082] The convolution module can be implemented by using a convolutional network conv;

[0083] The relationship feature extraction module can be implemented by using a recurrent neural network, a Transformer or a feature pyramid;

[0084] The first or second encoding module can be implemented by using an autoencoding neural network or an autoencoder or a DAE network;

[0085] The first or second decoding module may be implemented using a self-decoding neural network or an automatic decoder;

[0086] The first or second feature reconstruction module can be implemented using a graph convolutional neural network.

[0087] The fully connected module can be implemented by using a fully connected layer.

[0088] The personnel category encoding module can be implemented by using the LSTM autoencoder in Tensflow;

[0089] The personnel trajectory encoding module can be implemented by using a graph convolutional network (GCN), a graph attention network (GAT), etc.

[0090] The training process of the personnel relationship mining model using the personnel relationship training data set includes:

[0091] S21, initialize the number of training iterations;

[0092] S22, taking each personnel relationship training data in the personnel relationship training data set as input data, inputting into the personnel relationship mining model, and obtaining personnel category prediction result information;

[0093] S23, performing difference calculation processing on the personnel category label data corresponding to the input data and the personnel category prediction result information to obtain a difference value;

[0094] S24, judging whether the difference value satisfies a convergence condition, and obtaining a first judgment result;

[0095] When the first judgment result is no, judging whether the training iteration number value is equal to the training number threshold, and obtaining a second judgment result;

[0096] When the second judgment result is no, determining that the model training state does not meet the training termination condition;

[0097] When the second judgment result is yes, determining that the model training state satisfies the training termination condition;

[0098] When the first judgment result is yes, determining that the model training state satisfies a training termination condition;

[0099] When the model training state does not meet the training termination condition, the classification network parameters are updated using the parameter update model, the training iteration number value is increased by 1, and S22 is executed;

[0100] When the model training state satisfies the training termination condition, the training process of the personnel relationship mining model is completed to obtain a trained personnel relationship mining model.

[0101] The convergence condition means that when the difference value is less than a set threshold, the convergence condition is met; when the difference value is not less than the set threshold, the convergence condition is not met.

[0102] The parameter updating model is:

[0103]

[0104] θ←θ+v;

[0105] In the formula, x (i)is the i-th data in the personnel relationship training data set, is the loss function for the j-th category of personnel category label data, v is the parameter update value, θ is the parameter of the classification network, η is the initial parameter learning rate, α is the momentum angle parameter, 0≤α≤1, It means to find the partial derivative of variable θ, f(x (i) ; θ) represents the personnel category label data calculated by the personnel relationship mining model for the i-th data of the personnel relationship training data set, f(·) is the calculation function corresponding to the personnel relationship mining model; exp represents the power operation of the constant e; η and α are preset values.

[0106] in, Indicates the use of Update the parameter v; θ←θ+v, which means updating θ using θ+v.

[0107] The preprocessing of the personnel relationship detection data to obtain the preprocessed personnel relationship detection data includes:

[0108] S41, performing data cleaning processing on the personnel relationship detection data to obtain cleaned data;

[0109] S42, performing data range limiting processing on the cleaned data to obtain limited data;

[0110] S43, normalizing the limited data to obtain pre-processed personnel relationship detection data.

[0111] The evaluating and processing the personnel category mining result information to obtain the personnel relationship evaluation result includes:

[0112] S61, obtaining personnel category mining result information within N time periods;

[0113] S62, quantifying the personnel category mining result information within the N time periods to obtain a personnel category mining result matrix A;

[0114] The row vectors in the personnel category mining result matrix A are vectors obtained by quantizing the personnel category mining result information within a time period;

[0115] The quantification processing is to quantify and score the category information of each category according to its value; for example, the value of the real identity category information can be network security management personnel, network security technicians, general personnel, etc., and the corresponding quantitative scores are 4, 3, 2 and 1 respectively; the value of the virtual identity category information can be threat personnel, detection personnel, tourists, etc., and the corresponding quantitative scores are 4, 3, 2 and 1 respectively; the value of the trust relationship category information can be trust, general trust, distrust and danger, and the corresponding quantitative scores are 4, 3, 2 and 1 respectively; the value of the activity trajectory category information can be danger, general danger, general safety and safety, and the corresponding quantitative scores are 4, 3, 2 and 1 respectively.

[0116] S63, performing feature vector calculation processing on the personnel category mining result matrix A to obtain a personnel category feature vector V;

[0117] S64, using the personnel category feature vector, performing weighted sum processing on the personnel category mining result matrix to obtain a personnel relationship evaluation result value;

[0118] The personnel category feature vector is used to perform weighted sum processing on the personnel category mining result matrix to obtain the personnel relationship evaluation result value, and its calculation expression is:

[0119] q = sum(VA);

[0120] Among them, VA represents the result vector obtained by multiplying the personnel category feature vector V and the personnel category mining result matrix A, and sum represents the sum operation of all elements in the vector.

[0121] The process of performing feature vector calculation on the personnel category mining result matrix A to obtain the personnel category feature vector V includes:

[0122] S631, perform QR decomposition on the personnel category mining result matrix A to obtain

[0123] A=U s V s T ,

[0124] Among them, U s 、V s denote the column orthogonal matrix and upper triangular matrix of matrix A respectively;

[0125] S632, using U s All diagonal elements of are used to construct the personnel category feature vector V.

[0126] The process of performing feature vector calculation on the personnel category mining result matrix A to obtain the personnel category feature vector V includes:

[0127] Treat each row vector of matrix A as a point in high-dimensional space;

[0128] Perform kernel mapping on matrix A to obtain the corresponding kernel matrix;

[0129] Performing eigendecomposition on the kernel matrix to obtain an eigenvector;

[0130] Determine the feature vector as a personnel category feature vector V;

[0131] The kernel mapping process may adopt regeneration kernel mapping or Mercer kernel mapping.

[0132] According to a second aspect of an embodiment of the present invention, a distributed personnel relationship mining and evaluation device is disclosed, the device comprising:

[0133] A memory storing executable program code; a processor coupled to the memory; the processor calling the executable program code stored in the memory to execute the distributed personnel relationship mining and evaluation method.

[0134] According to a third aspect of an embodiment of the present invention, a computer storable medium is disclosed, wherein the computer storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the distributed personnel relationship mining and evaluation method.

[0135] According to a fourth aspect of the embodiments of the present invention, an information data processing terminal is disclosed, wherein the information data processing terminal is used to implement the distributed personnel relationship mining and evaluation method.

[0136] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A distributed personnel relationship mining and evaluation method, characterized in that: include: S1, obtain the personnel relationship training data set; The personnel relationship training data set includes personnel relationship training data and corresponding personnel category label data; The personnel category label data includes real identity category label information, virtual identity category label information, trust relationship category label information, and activity trajectory category label information; S2, using the personnel relationship training data set to train a personnel relationship mining model; S3, obtaining personnel relationship detection data; S4, preprocessing the personnel relationship detection data to obtain preprocessed personnel relationship detection data; S5, using the personnel relationship mining model to process the pre-processed personnel relationship detection data to obtain personnel category mining result information; The personnel category mining result information includes real identity category information, virtual identity category information, trust relationship category information, and activity track category information; S6, evaluating the personnel category mining result information to obtain a personnel relationship evaluation result value; The personnel relationship assessment result value is used to represent the threat level of the personnel relationship; The training process of the personnel relationship mining model using the personnel relationship training data set includes: S21, initialize the number of training iterations; S22, taking each personnel relationship training data in the personnel relationship training data set as input data, inputting into the personnel relationship mining model, and obtaining personnel category prediction result information; S23, performing difference calculation processing on the personnel category label data corresponding to the input data and the personnel category prediction result information to obtain a difference value; S24, judging whether the difference value satisfies a convergence condition, and obtaining a first judgment result; When the first judgment result is no, judging whether the training iteration number value is equal to the training number threshold, and obtaining a second judgment result; When the second judgment result is no, determining that the model training state does not meet the training termination condition; When the second judgment result is yes, determining that the model training state satisfies the training termination condition; When the first judgment result is yes, determining that the model training state satisfies a training termination condition; When the model training state does not meet the training termination condition, the classification network parameters are updated using the parameter update model, the training iteration number value is increased by 1, and S22 is executed; When the model training state satisfies the training termination condition, the training process of the personnel relationship mining model is completed to obtain a trained personnel relationship mining model; The parameter updating model is: θ←θ+v; In the formula, x (i) is the i-th data in the personnel relationship training data set, is the loss function for the j-th category of personnel category label data, v is the parameter update value, θ is the parameter of the classification network, η is the initial parameter learning rate, α is the momentum angle parameter, 0≤α≤1, It means to find the partial derivative of variable θ, f(x (i) ; θ) represents the personnel category label data calculated by the personnel relationship mining model for the i-th data of the personnel relationship training data set, and f(.) is the calculation function corresponding to the personnel relationship mining model.

2. The distributed personnel relationship mining and evaluation method according to claim 1, characterized in that: The personnel relationship mining model includes a feature extraction network and a classification network; The feature extraction network is connected to the classification network and includes a first processing module, a second processing module, a third processing module and a fourth processing module, which are used to perform feature extraction processing on the input data to obtain feature data; The first processing module includes an input module, a first convolution module, a first activation module, a first residual module, and a first normalization module; the first port of the first processing module is the input end of the input module; the second port of the first processing module is the output end of the first normalization module; the first port of the first processing module is used as the input end of the feature extraction network to receive input data; the second port of the first processing module is connected to the first port of the second processing module; in the first processing module, the input module, the first convolution module, the first activation module, the first residual module, and the first normalization module are connected in sequence; The second processing module includes a second convolution module, a second activation module, and a second normalization module; the first port of the second processing module is the input end of the second convolution module; the second port of the second processing module is the output end of the second normalization module; the second port of the second processing module is connected to the first port of the third processing module; in the second processing module, the second convolution module, the second activation module, and the second normalization module are connected in sequence; The third processing module includes a third convolution module, a third activation module, and a third residual module; the first port of the third processing module is the input end of the third convolution module; the second port of the third processing module is the output end of the third residual module; the second port of the third processing module is connected to the first port of the fourth processing module; in the third processing module, the third convolution module, the third activation module, and the third residual module are connected in sequence; The fourth processing module includes a fourth convolution module, a fourth activation module, and a fourth residual module; the first port of the fourth processing module is the input end of the fourth convolution module; the second port of the fourth processing module is the output end of the fourth residual module; the second port of the fourth processing module is used as the output end of the feature extraction network to output feature data; In the fourth processing module, the fourth convolution module, the fourth activation module, and the fourth residual module are connected in sequence; The classification network is used to perform relationship classification and recognition processing on the feature data to obtain personnel category mining result information or personnel category prediction result information, including a first relationship feature extraction module, a first encoding module, a first decoding module, a first feature reconstruction module, a second relationship feature extraction module, a second encoding module, a second decoding module, a second feature reconstruction module and a full connection module; the first encoding module includes a first personnel category encoding module and a first personnel trajectory encoding module; the second encoding module includes a second personnel category encoding module and a second personnel trajectory encoding module; The input end of the first relationship feature extraction module and the input end of the second relationship feature extraction module are both connected to the second port of the fourth processing module; The output end of the first relationship feature extraction module is connected to the input end of the first encoding module; the output ends of the first personnel category encoding module and the second personnel trajectory encoding module are connected to the input end of the first decoding module; the output ends of the second personnel category encoding module and the first personnel trajectory encoding module are connected to the input end of the second decoding module; the output end of the first decoding module is connected to the input end of the first feature reconstruction module; the output end of the second decoding module is connected to the input end of the second feature reconstruction module; the output end of the first feature reconstruction module and the output end of the second feature reconstruction module are both connected to the input end of the full connection module; the output end of the full connection module is used to output the personnel category mining result information or the personnel category prediction result information obtained by the personnel relationship mining model.

3. The distributed personnel relationship mining and evaluation method according to claim 1, characterized in that: The preprocessing of the personnel relationship detection data to obtain the preprocessed personnel relationship detection data includes: S41, performing data cleaning processing on the personnel relationship detection data to obtain cleaned data; S42, performing data range limiting processing on the cleaned data to obtain limited data; S43, normalizing the limited data to obtain pre-processed personnel relationship detection data.

4. The distributed personnel relationship mining and evaluation method according to claim 1, characterized in that: The evaluating and processing the personnel category mining result information to obtain the personnel relationship evaluation result includes: S61, obtaining personnel category mining result information within N time periods; S62, quantifying the personnel category mining result information within the N time periods to obtain a personnel category mining result matrix A; The row vectors in the personnel category mining result matrix A are vectors obtained by quantizing the personnel category mining result information within a time period; S63, performing feature vector calculation processing on the personnel category mining result matrix A to obtain a personnel category feature vector V; S64, using the personnel category feature vector, performing weighted sum processing on the personnel category mining result matrix to obtain a personnel relationship evaluation result value.

5. The distributed personnel relationship mining and evaluation method according to claim 4, characterized in that: The process of performing feature vector calculation on the personnel category mining result matrix A to obtain the personnel category feature vector V includes: S631, perform QR decomposition on the personnel category mining result matrix A to obtain Among them, U s 、V s denote the column orthogonal matrix and upper triangular matrix of matrix A respectively; S632, using U s All diagonal elements of are used to construct the personnel category feature vector V.

6. A distributed personnel relationship mining and evaluation device, characterized in that: The device comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the distributed personnel relationship mining and evaluation method according to any one of claims 1 to 5.

7. A computer storable medium, characterized in that: The computer storable medium stores computer instructions, and when the computer instructions are called, they are used to execute the distributed personnel relationship mining and evaluation method according to any one of claims 1 to 5.

8. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the distributed personnel relationship mining and evaluation method as described in any one of claims 1 to 5.

Images