CN115271980A - Risk value prediction method and device, computer equipment and storage medium - Google Patents
Risk value prediction method and device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN115271980A CN115271980A CN202210933030.2A CN202210933030A CN115271980A CN 115271980 A CN115271980 A CN 115271980A CN 202210933030 A CN202210933030 A CN 202210933030A CN 115271980 A CN115271980 A CN 115271980A
- Authority
- CN
- China
- Prior art keywords
- feature vector
- risk value
- social
- target
- matrix
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 239000013598 vector Substances 0.000 claims abstract description 193
- 239000011159 matrix material Substances 0.000 claims abstract description 127
- 238000003062 neural network model Methods 0.000 claims abstract description 54
- 230000006870 function Effects 0.000 claims description 58
- 238000012549 training Methods 0.000 claims description 15
- 238000004422 calculation algorithm Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 13
- 238000000605 extraction Methods 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 4
- 238000013473 artificial intelligence Methods 0.000 abstract description 3
- 238000005516 engineering process Methods 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 11
- 238000004364 calculation method Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 238000012706 support-vector machine Methods 0.000 description 5
- 238000003064 k means clustering Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/08—Insurance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Technology Law (AREA)
- Development Economics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Primary Health Care (AREA)
- Tourism & Hospitality (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to the field of artificial intelligence, and can generate the feature vector according to the user attribute and the social relation of a user, so that the efficiency and the accuracy of risk value prediction are improved. A risk value prediction method, apparatus, device and medium are provided, the method comprising: obtaining target social data, and constructing a topological graph of the target social data to obtain a social network topological graph; extracting characteristics of the social network topological graph to obtain a characteristic matrix and an adjacency matrix corresponding to the social network topological graph; inputting the characteristic matrix and the adjacency matrix into a neural network model of the graph to generate characteristic vectors, and obtaining target characteristic vectors of the social network topological graph; and inputting the target characteristic vectors into a risk value prediction model to predict the risk values, and obtaining a risk value prediction result of each target user. In addition, the application also relates to a block chain technology, and the target social data is stored in the block chain.
Description
Technical Field
The present application relates to the field of artificial intelligence, and in particular, to a method and an apparatus for predicting a risk value, a computer device, and a storage medium.
Background
With the improvement of living standard of people, the purchasing intention of people to insurance business is gradually enhanced. Before recommending the service to the user, the risk value of the user needs to be predicted, and then different services or different offers can be given based on the risk value of the user. The existing risk value prediction methods are generally two, wherein the first method is to score the risk value of a user according to subjective experience through personal data of the user; the second is to predict the risk value of the user based on the classification algorithm of machine learning. The first risk value prediction method has a problem of low efficiency. For the second risk value prediction mode, the risk scoring is only carried out according to the text information of the personal data of the user, the behavior characteristics of the user cannot be effectively constructed, and the accuracy of predicting the risk value is reduced.
Therefore, how to improve the efficiency and accuracy of predicting the risk value becomes an urgent problem to be solved.
Disclosure of Invention
The application provides a risk value prediction method, a risk value prediction device, computer equipment and a storage medium, wherein feature vectors are generated by inputting a feature matrix corresponding to a social network topological graph and an adjacency matrix into a graph neural network model, so that the feature vectors can be generated according to user attributes and social relations of users, and the efficiency and accuracy of risk value prediction are improved.
In a first aspect, the present application provides a method for predicting a risk value, the method comprising:
obtaining target social data, constructing a topological graph of the target social data, and obtaining a social network topological graph, wherein the target social data comprises social data of at least one target user;
extracting features of the social network topological graph to obtain a feature matrix and an adjacent matrix corresponding to the social network topological graph;
generating a feature vector of the feature matrix and the adjacency matrix input graph neural network model to obtain a target feature vector of the social network topological graph;
and inputting the target characteristic vectors into a risk value prediction model for risk value prediction to obtain a risk value prediction result of each target user.
In a second aspect, the present application further provides a risk value prediction apparatus, including:
the topological graph construction module is used for obtaining target social data, carrying out topological graph construction on the target social data and obtaining a social network topological graph, wherein the target social data comprises social data of at least one target user;
the characteristic extraction module is used for extracting characteristics of the social network topological graph to obtain a characteristic matrix and an adjacent matrix corresponding to the social network topological graph;
the characteristic vector generation module is used for generating characteristic vectors of the characteristic matrix and the adjacency matrix input graph neural network model to obtain target characteristic vectors of the social network topological graph;
and the risk value prediction module is used for inputting the target characteristic vectors into a risk value prediction model to perform risk value prediction, and obtaining a risk value prediction result of each target user.
In a third aspect, the present application also provides a computer device comprising a memory and a processor;
the memory for storing a computer program;
the processor is configured to execute the computer program and to implement the risk value prediction method as described above when executing the computer program.
In a fourth aspect, the present application also provides a computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to implement the risk value prediction method as described above.
The application discloses a risk value prediction method, a risk value prediction device, computer equipment and a storage medium, wherein the user attribute and the social relation of a target user can be represented according to a social network topological graph by constructing the topological graph of target social data; by extracting the characteristics of the social network topological graph, a characteristic matrix and an adjacency matrix corresponding to the social network topological graph can be obtained; the feature matrix and the adjacency matrix are input into the neural network model of the graph to generate the feature vector, so that the target feature vector of the social network topological graph is obtained, the feature vector can be generated according to the user attribute and the social relation of the user, the problem that the user attribute and the social relation of the user cannot be considered simultaneously when the risk value of the user is predicted in the prior art is solved, and the efficiency and the accuracy of the subsequent risk value prediction are improved; the target characteristic vectors are input into the risk value prediction model to predict the risk values, so that the risk value prediction result of each target user is obtained, and the efficiency and accuracy of risk value prediction are improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a schematic flow chart of a risk value prediction method provided in an embodiment of the present application;
FIG. 2 is a schematic flow chart diagram illustrating sub-steps of a training graph neural network model provided in an embodiment of the present application;
fig. 3 is a schematic block diagram of a risk value prediction apparatus provided in an embodiment of the present application;
fig. 4 is a schematic block diagram of a structure of a computer device according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application without making any creative effort belong to the protection scope of the present application.
The flowcharts shown in the figures are illustrative only and do not necessarily include all of the contents and operations/steps, nor do they necessarily have to be performed in the order described. For example, some operations/steps may be decomposed, combined or partially combined, so that the actual execution sequence may be changed according to the actual situation.
It is to be understood that the terminology used in the description of the present application herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used in the specification of the present application and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
It should also be understood that the term "and/or" as used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.
The embodiment of the application provides a risk value prediction method, a risk value prediction device, computer equipment and a storage medium. The risk value prediction method can be applied to a server or a terminal, and the feature vector is generated by inputting the feature matrix corresponding to the social network topological graph and the adjacency matrix into the graph neural network model, so that the feature vector can be generated according to the user attribute and the social relation of the user, and the efficiency and the accuracy of risk value prediction are improved.
The server may be an independent server, or may be a cloud server that provides basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a Network service, cloud communication, a middleware service, a domain name service, a security service, a Content Delivery Network (CDN), a big data and artificial intelligence platform, and the like. The terminal can be an electronic device such as a smart phone, a tablet computer, a notebook computer, a desktop computer and the like.
Some embodiments of the present application will be described in detail below with reference to the accompanying drawings. The embodiments described below and the features of the embodiments can be combined with each other without conflict.
As shown in fig. 1, the risk value prediction method includes steps S10 to S40.
S10, obtaining target social data, constructing a topological graph of the target social data, and obtaining a social network topological graph, wherein the target social data comprises social data of at least one target user.
It should be noted that the risk value prediction method provided in the embodiment of the present application is applied to predicting a risk value according to social data and recommending a service to a user according to the risk value, and feature vectors are generated by inputting a feature matrix corresponding to a social network topological graph and an adjacency matrix into a graph neural network model, so that feature vectors can be generated according to user attributes and social relationships of the user, the efficiency and accuracy of risk value prediction are improved, and further, the efficiency and accuracy of service recommendation can be improved.
Illustratively, target social data to be subjected to risk value prediction can be acquired according to uploading operation of a user; and target social data to be subjected to risk value prediction can be acquired according to the selected operation of the user.
Illustratively, the target social data includes social data of at least one target user over different time periods. For example, the target social data may include social data for target user a within 7 days; as another example, the target social data may include social data for target user B within 30 days. The social data may include user attributes such as name, age, gender, occupation, hobbies, and the like of the user, and may also include social relationships between the user and other users, and the like. Social relationships may include, but are not limited to, friend relationships, alumni relationships, family relationships, stranger relationships, and the like.
For example, the target social data may be stored in a local disk or a local database. In this embodiment, to further ensure the privacy and security of the target social data, the target social data may be stored in a node of a block chain.
Wherein the social network topology may include at least two nodes and edges between the two nodes. It should be noted that each node represents a user, a value corresponding to the node is a user attribute, and an edge between two nodes represents a social relationship between two users.
Illustratively, constructing a topology map of the target social data to obtain a social network topology map may include: acquiring a current user in target social data and social relations between the current user and other users; according to the user attributes of the current user and the current user, establishing attribute values corresponding to the current node and the current node in the social network topological graph corresponding to the target social data; and according to the social relationship between the current user and other users, edges between the current node and other nodes in the social network topological graph corresponding to the target social data are constructed.
For example, the social network topology may be represented as G = (V, E, X), where V = { V = 1 ,v 2 ,...,v n Expressing a node sequence, wherein each node represents a user, and n expresses the number of nodes; e = { E = { E) i,j },e i,j Representing a node v i And node v j Presence of a connecting edge, representing v i And v j There is a social relationship between the two represented users. The characteristic matrix X belongs to R n×f Each row in (a) represents a user attribute of a different user, and f represents the number of columns of user attributes of the user.
By constructing the topological graph of the target social data, the user attribute and the social relation of the target user can be characterized according to the social network topological graph.
And S20, extracting characteristics of the social network topological graph to obtain a characteristic matrix and an adjacent matrix corresponding to the social network topological graph.
Illustratively, the feature matrix may be represented by X; the adjacency matrix may be denoted a. The feature matrix X in the social network topological graph can be directly extracted, and then the adjacency matrix is determined according to the social relationship in the social network topological graph. Therein, for example, when node v in the social network topology graph i And node v j When there is a social relationship, the value corresponding to the ith row and the jth column in the adjacency matrix a is 1, otherwise, it is 0.
It is understood that in the embodiments of the present application, the social relationships of the users are characterized by the adjacency matrix.
By extracting the characteristics of the social network topological graph, a characteristic matrix and an adjacency matrix corresponding to the social network topological graph can be obtained.
And S30, generating a feature vector by inputting the feature matrix and the adjacency matrix into a neural network model of the graph, and obtaining a target feature vector of the social network topological graph.
In the embodiment of the application, after the feature matrix and the adjacency matrix corresponding to the social network topological graph are obtained, the social network topological graph can be input into the graph neural network model to generate the feature vector, and the target feature vector of the social network topological graph is obtained.
The feature matrix and the adjacency matrix are input into the neural network model of the graph to generate the feature vector, the target feature vector of the social network topological graph is obtained, the feature vector can be generated according to the user attribute and the social relation of the user, the problem that the user attribute and the social relation of the user cannot be considered simultaneously when the risk value of the user is predicted in the prior art is solved, and the efficiency and the accuracy of subsequent risk value prediction are improved.
It should be noted that the neural network model is a model trained in advance. For example, the sample social network topological graph can be input into the graph neural network model for iterative training based on a mutual information maximization strategy of a clustering sub-network, so that the trained graph neural network model is obtained.
By way of example, the Graph neural network model may include, but is not limited to, graph Convolution Networks (GCNs), graph Attention Networks (Graph Attention Networks), graph Auto-encoders (Graph Auto encoders), graph generation Networks (Graph generated Networks), and Graph spatio-temporal Networks (Graph Spatial-temporal Networks), among others.
In the embodiment of the present application, a graph neural network model is taken as an example of a graph convolution network, so as to explain how to perform model training in detail. The graph convolution network has low time calculation complexity and strong expandability, and can quickly acquire dynamic characteristics in the social network topological graph. By training the graph neural network model, the trained graph neural network model can better represent and extract information such as user attributes, social relations and the like in the social network topological graph, the characteristic vector can be generated according to the user attributes and the social relations of the users, the efficiency and the accuracy of risk value prediction are improved, and the efficiency and the accuracy of service recommendation can be improved.
In the embodiment of the present application, a training process of the neural network model will be described in detail.
Referring to fig. 2, fig. 2 is a schematic flowchart illustrating sub-steps of training a neural network model according to an embodiment of the present application, which may specifically include the following steps S301 to S307.
Step S301, obtaining sample social data, and constructing a sample social network topological graph according to the sample social data, wherein the sample social data comprises social data of at least one sample user.
For example, sample social data may be obtained, and a social network topology may be constructed based on the sample social data. The sample social data refers to social data of sample users. The social data may include user attributes such as name, age, gender, occupation, hobbies and the like of the user, and may further include social relationships between the user and other users. Social relationships may include, but are not limited to, friendships, alumni relationships, family relationships, stranger relationships, and so forth. The sample social data may be stored in a local disk or local database. In an embodiment of the present application, to further ensure the privacy and security of the sample social data, the sample social data may be stored in a node of a block chain.
For an exemplary process of constructing a social network topology map according to the sample social data, reference may be made to the process of constructing a topology map for the target social data in the foregoing embodiment, and specific processes are not described herein again.
Step S302, inputting the feature matrix and the adjacency matrix corresponding to the sample social network topological graph into the graph neural network model for feature vector generation, and obtaining a first feature vector.
For example, feature vectors can be generated by using a feature matrix corresponding to the sample social network topological graph and the adjacency matrix input graph neural network model, so as to obtain a first feature vector. The specific process of generating the feature vector is not limited herein.
Illustratively, the first feature vector may be represented as H, and the first feature vector H may be obtained by the following equation:
wherein W represents a weight matrix;
representing an adjacency matrix plus an identity matrix;
is a degree matrix whose diagonal elements are
σ is a linear rectification function (ReLu); h = { H = 1 ,h 2 ,…,h n And expressing feature vectors corresponding to the nodes in the social network topological graph.
Feature vectors are generated by inputting the feature matrix corresponding to the sample social network topological graph and the adjacency matrix into the graph neural network model, and first feature vectors corresponding to the sample social network topological graph can be obtained.
And S303, constructing a contrast characteristic matrix corresponding to the characteristic matrix, and inputting the contrast characteristic matrix and the adjacency matrix into the graph neural network model for characteristic extraction to obtain a second characteristic vector.
In the embodiment of the application, the feature matrix and the comparison feature matrix can be compared and learned by the graph neural network model by constructing the comparison feature matrix corresponding to the feature matrix, so that the accuracy of generating the feature vector by the trained graph neural network model is improved.
In some embodiments, constructing a contrast feature matrix corresponding to the feature matrix may include: and adjusting the positions of the elements in each row in the feature matrix, and determining the adjusted feature matrix as a contrast feature matrix.
For example, when the positions of the rows of elements in the feature matrix are adjusted, the positions of the rows of elements may be rearranged to obtain a contrast feature matrix, where the contrast feature matrix may be represented as
For example, when the first row element in the feature matrix X is [1.01,2.09,3.87,0.287 ]]Then, after the position adjustment, the first row element may become [3.87,2.09,0.287,1.01, ]]。
For example, after a contrast feature matrix corresponding to the feature matrix is constructed, the contrast feature matrix and the adjacency matrix may be input to the neural network model of the graph for feature extraction, so as to obtain a second feature vector. Wherein the generation of the second feature vector may be represented as:
in the formula (I), the compound is shown in the specification,
step S304, generating at least one clustering sub-network corresponding to the social network topological graph, and determining a global feature vector corresponding to each clustering sub-network.
It should be noted that, if the mutual information maximization is directly performed on the entire social network topological graph, the representation of the feature vector learned by the model is too coarse, and the accuracy of predicting the risk value is further reduced. In the embodiment of the application, in order to improve the fine granularity of the feature vector generated by the trained graph neural network model, the similar nodes in the social network topological graph can be clustered to obtain the corresponding clustering sub-networks, and then mutual information maximization can be performed on the more refined clustering sub-networks, so that the model can learn to express the feature vector in a more fine-grained manner.
It is understood that mutual information refers to a measure of interdependency between variables.
In some embodiments, generating at least one clustering subnetwork corresponding to the social network topology graph may include: and based on a clustering algorithm, clustering the characteristic matrix to obtain at least one clustering subnetwork.
It should be noted that the cluster analysis model may include, but is not limited to, a K-means clustering algorithm, a system clustering algorithm, a DBSCAN algorithm, and the like.
For example, a feature matrix X in the sample social network topological graph may be clustered based on a K-means clustering algorithm to obtain at least one clustering sub-network. For example, the clustering process can be calculated by:
{g 1 ,..,g m }＝Kmeans(X)
wherein, { g 1 ,..,g m Denotes that m clustering subnetworks are obtained. The specific calculation process of the clustering process is not limited herein.
For example, for a sample social network topology of 10 nodes, if two clustering subnetworks are obtained, the corresponding clustering subnetworks may be represented as: g 1 ＝{1,5,9,7,3}；g 2 ＝{2,4,6,8,10}。
Clustering processing is carried out on the characteristic matrix based on a clustering algorithm, so that the similar nodes in the social network topological graph can be clustered, and at least one clustering sub-network corresponding to the social network topological graph is obtained.
It should be noted that after determining the clustering sub-networks corresponding to the sample social network topology, the global feature vector corresponding to each clustering sub-network needs to be determined according to the first feature vector corresponding to the sample social network topology.
In some embodiments, determining the global feature vector corresponding to each clustering subnetwork may include: extracting the feature vectors of the first feature vectors to obtain the feature vectors corresponding to the nodes in each clustering subnetwork; and adding the feature vectors corresponding to all the nodes in each clustering sub-network to obtain an average value, so as to obtain a global feature vector corresponding to each clustering sub-network.
For example, the global feature vector corresponding to the clustering sub-network can be represented as s r (r is more than or equal to 1 and less than or equal to m); wherein the global feature vector s r Can be calculated by the following formula:
wherein r represents the r-th clustering subnetwork;
representing a feature vector corresponding to an ith clustering subnetwork node in an r clustering subnetwork; p denotes the number of nodes in the clustering subnetwork.
For example, through the above formula, feature vectors corresponding to all nodes in each clustering sub-network may be added to obtain an average value, and then the obtained feature vector is input into a linear rectification function (ReLu) to be modified, so as to obtain a global feature vector corresponding to each clustering sub-network. For example, for clustering subnetwork g 1 = {1,5,9,7,3} and g 2 = 2,4,6,8,10, the corresponding global feature vector is s 1 And s 2 。
Step S305, generating at least one positive sample pair and at least one negative sample pair corresponding to the sample social network topology map according to the first feature vector, the second feature vector, and the global feature vector corresponding to each clustering sub-network.
In some embodiments, generating at least one positive sample pair and at least one negative sample pair corresponding to the sample social network topology from the first feature vector, the second feature vector, and the global feature vector corresponding to each clustering subnetwork may include: extracting the feature vectors of the first feature vectors to obtain first class feature vectors of each node in each clustering sub-network; extracting the second characteristic vector to obtain a second class of characteristic vectors of each node in each clustering subnetwork; combining the global feature vector corresponding to each clustering sub-network with the corresponding first class feature vector to obtain at least one positive sample pair; and combining the global feature vector corresponding to each clustering sub-network with the corresponding second-class feature vector to obtain at least one negative sample pair.
Illustratively, for clustering subnetwork g 1 = {1,5,9,7,3} and g 2 = 2,4,6,8,10, for the first eigenvector H = { H = 1 ,h 2 ,…,h n Extracting the feature vector to obtain a clustering subnetwork g 1 The first type of feature vector corresponding to each node in (1) is { h } 1 ,h 5 ,h 9 ,h 7 ,h 3 }, and clustering sub-network g 2 The first type of feature vector corresponding to each node in (1) is { h } 2 ,h 4 ,h 6 ,h 8 ,h 10 }. For the second feature vector
Extracting feature vector to obtain clustering sub-network g 1 The second type of feature vector corresponding to each node in (1) is
And clustering subnetwork g 2 The second type of feature vector corresponding to each node in (1) is
Illustratively, the global feature vector corresponding to each clustering subnetwork and the corresponding first class feature vector are combined to obtain at least one positive sample pair. Wherein a positive sample pair can be represented as
For example, for clustering sub-network g 1 Corresponding global feature vector s 1 And clustering subnetwork g 1 Corresponding first class eigenvector { h } 1 ,h 5 ,h 9 ,h 7 ,h 3 Combine to get positive sample pair
As another example, for clustering subnetwork g 2 Corresponding global feature vector s 2 And clustering subnetwork g 2 Corresponding first class eigenvector { h } 2 ,h 4 ,h 6 ,h 8 ,h 10 Combine to get positive sample pair
Illustratively, the global feature vector corresponding to each clustering subnetwork and the corresponding second-class feature vector are combined to obtain at least one negative sample pair. Wherein a negative sample pair can be represented as
For example, for clustering subnetwork g 1 Corresponding global feature vector s 1 And clustering subnetwork g 1 Corresponding feature vectors of the second type
Are combined to obtain a negative sample pair
As another example, for clustering subnetwork g 2 Corresponding global feature vector s 2 And clustering subnetwork g 2 Corresponding feature vectors of the second type
Are combined to obtain a negative sample pair
It should be noted that by combining the global feature vector corresponding to each clustering subnetwork with the corresponding first class of feature vectors and combining the global feature vector corresponding to each clustering subnetwork with the corresponding second class of feature vectors, at least one positive sample pair and negative sample pair are obtained, so that the model can be compared and learned by the positive sample pair and the negative sample pair, and the capability of the model for characterizing the feature vectors is enhanced.
And S306, calculating loss function values of all the positive sample pairs and all the negative sample pairs to obtain the loss function values corresponding to the sample social network topological graph.
In some embodiments, performing the loss function value calculation on all the positive sample pairs and all the negative sample pairs to obtain the loss function values corresponding to the sample social network topology may include: scoring each positive sample pair according to a preset discriminator to obtain a first scoring value of each positive sample pair; scoring each negative sample pair according to the discriminator to obtain a second scoring value of each negative sample pair; and calculating a loss function value according to all the first score values and all the second score values based on a preset loss function to obtain the loss function value corresponding to the sample social network topological graph.
It should be noted that, in the embodiment of the present application, the loss function value corresponding to the sample social network topology may be calculated through the loss function and the identifier. The discriminator may be a Bilinear Scoring Function (Bilinear Scoring Function) for discriminating and Scoring the positive sample pairs and all the negative sample pairs. The discriminator may be denoted as D.
Illustratively, each positive sample pair may be scored according to discriminator D, resulting in a first score value for each positive sample pair; and scoring each negative sample pair according to the discriminator D to obtain a second score value of each negative sample pair.
Exemplary, the firstA score value can be expressed as
The second score value may be expressed as
Wherein the content of the first and second substances,
where σ is sigmoid function, M r Is a parameter matrix. By the above formula, can
The score between is converted to a positive number.
It should be noted that, in the embodiment of the present application, when the positive sample pair and the negative sample pair are scored through the discriminator, the positive sample pair may be given a higher score as much as possible, and the negative sample pair may be given a lower score, so that the difference between the positive sample pair and the negative sample pair may be maximized.
Exemplary loss functions may include, but are not limited to, 0-1 loss functions, absolute value loss functions, logarithmic loss functions, quadratic loss functions, exponential loss functions, and cross-entropy loss functions, among others. In the embodiment of the present application, a cross entropy loss function may be used to perform the calculation of the loss function value.
Exemplary, the cross entropy loss function is calculated as follows:
for example, based on the cross entropy loss function, a loss function value calculation may be performed according to all the first score values and all the second score values, so as to obtain a loss function value corresponding to the sample social network topological graph. The specific calculation process is not limited herein.
By calculating the loss function value according to the first score value of the positive sample pair and the second score value of the negative sample pair, mutual information maximization can be achieved, namely the difference between the positive sample pair and the negative sample pair is maximized, comparison learning of the model is achieved, and accuracy of generating the feature vector of the model is improved.
Step S307, if the loss function value is larger than a preset first loss threshold value, adjusting parameters of the graph neural network model, performing next round of training until the obtained loss function value meets a preset condition, and ending the training to obtain the trained graph neural network model.
For example, the preset condition may be that the obtained loss function value is less than or equal to a first loss threshold, the difference between two adjacent loss function values is less than or equal to a second loss threshold, and the number of iterations is greater than or equal to a preset number of iterations. The first loss threshold and the second loss threshold may be set according to actual conditions, specific values are not limited herein, and the first loss threshold is not equal to the second loss threshold. The preset iteration number may be set according to actual conditions, and the specific numerical value is not limited herein.
For example, if the loss function value is greater than the first loss threshold, the parameters of the graph neural network model are adjusted, the next round of training is performed until the obtained loss function value is less than or equal to the first loss threshold, and the training is finished to obtain the trained graph neural network model. Wherein parameters of the neural network model may be adjusted according to a gradient descent algorithm. In addition, the clustering coefficient K in the K-means clustering algorithm can be adjusted according to the gradient descent algorithm; the larger the value of the clustering coefficient k is, the higher the fine granularity degree of the node is, but the overfitting phenomenon of the model can occur due to the fact that the clustering coefficient k is too large, so that the clustering coefficient k needs to be adjusted to improve the accuracy of the model for generating the feature vector.
By adjusting the parameters of the graph neural network model when the loss function value is greater than a preset first loss threshold value, the model can learn the user attribute and social relation of the user, meanwhile, the model can express the feature vectors in a finer granularity mode, and then the accuracy of subsequent classification according to the feature vectors output by the model can be improved.
And S40, inputting the target characteristic vectors into a risk value prediction model to predict risk values, and obtaining a risk value prediction result of each target user.
In the embodiment of the present application, the risk value prediction model may be a Support Vector Machine (SVM) model, or may be other classification models, which is not limited herein.
Illustratively, the target feature vectors are input into a trained support vector machine model for risk value prediction, and a risk value prediction result of each target user is obtained. Wherein the risk value prediction result comprises a risk value. For example, user 1 corresponds to a risk value of 6, user 2 corresponds to a risk value of 8, and so on.
It should be noted that, in the embodiment of the present application, social data of a user with a known risk value may be used in advance to output the trained neural network model of the graph for feature vector generation, and the obtained feature vector is used as sample data to perform iterative training on the support vector machine model, so as to obtain the trained support vector machine model.
The target characteristic vector output by the graph neural network model is input into the risk value prediction model to predict the risk value, so that the efficiency and the accuracy of risk value prediction can be improved.
In some embodiments, after inputting the target feature vector into the risk value prediction model to perform risk value prediction and obtaining a risk value prediction result of each target user, the risk value prediction method provided in the embodiment of the present application may further include: determining candidate services corresponding to each target user according to the risk value corresponding to each target user based on the corresponding relation between the preset risk value and the services; and recommending the corresponding candidate service to each target user.
For example, the corresponding relationship between the risk value and the business may be preset. For example, a risk value of 1 corresponds to business a, a risk value of 1 corresponds to business B, and so on. The service may include service types such as accident risk, serious disease risk, medical risk, life risk and the like, and may further include value levels corresponding to the services.
Exemplarily, the candidate service corresponding to each target user may be determined according to the risk value corresponding to each target user; and recommending corresponding candidate services to each target user. For example, for the user 1, the risk value is 6, and if it is determined that the service corresponding to the risk value 6 is an accident risk, the accident risk may be recommended to the user 1. For example, if the risk value of the user 2 is 8 and the business corresponding to the risk value 8 is determined to be a serious risk and the value rank is one rank, the one rank serious risk can be recommended to the user 2.
By recommending the service to each target user based on the risk value prediction result of each target user, the purpose of recommending the appropriate service to the target user can be realized, and the probability of purchasing the service by the target user is improved.
According to the risk value prediction method provided by the embodiment, the user attribute and the social relation of the target user can be represented according to the social network topological graph by constructing the topological graph of the target social data; by extracting the characteristics of the social network topological graph, a characteristic matrix and an adjacency matrix corresponding to the social network topological graph can be obtained; the feature matrix and the adjacency matrix are input into the neural network model of the graph to generate the feature vector, so that the target feature vector of the social network topological graph is obtained, the feature vector can be generated according to the user attribute and the social relation of the user, the problem that the user attribute and the social relation of the user cannot be considered simultaneously when the risk value of the user is predicted in the prior art is solved, and the efficiency and the accuracy of the subsequent risk value prediction are improved; clustering processing is carried out on the characteristic matrix based on a clustering algorithm, so that the similar nodes in the social network topological graph can be clustered, and at least one clustering sub-network corresponding to the social network topological graph is obtained; the global feature vector corresponding to each clustering sub-network is combined with the corresponding first class of feature vectors, and the global feature vector corresponding to each clustering sub-network is combined with the corresponding second class of feature vectors to obtain at least one positive sample pair and one negative sample pair, so that the model can be contrastingly learned by the positive sample pair and the negative sample pair, and the capability of the model for representing the feature vectors is enhanced; by calculating the loss function value according to the cross entropy function, the mutual information maximization can be realized, namely the maximization of the difference between the positive sample pair and the negative sample pair is realized, the comparison learning of the model is further realized, and the accuracy of the model for generating the feature vector is improved; when the loss function value is larger than a preset first loss threshold value, parameters of the graph neural network model are adjusted, so that the model can learn the user attributes and the social relations of the users, meanwhile, the model can express the feature vectors in a finer granularity mode, and the accuracy of subsequent classification according to the feature vectors output by the model can be improved; the target characteristic vector output by the graph neural network model is input into the risk value prediction model to predict the risk value, so that the efficiency and the accuracy of risk value prediction are improved; by recommending the service to each target user based on the risk value prediction result of each target user, the purpose of recommending the appropriate service to the target user can be realized, and the probability of purchasing the service by the target user is improved.
Referring to fig. 3, fig. 3 is a schematic block diagram of a risk value prediction apparatus 1000 for performing the risk value prediction method according to the embodiment of the present disclosure. The risk value prediction device may be configured in a server or a terminal.
As shown in fig. 3, the risk value prediction apparatus 1000 includes: a topological graph building module 1001, a feature extraction module 1002, a feature vector generation module 1003 and a risk value prediction module 1004.
The topological graph constructing module 1001 is configured to obtain target social data, construct a topological graph for the target social data, and obtain a social network topological graph, where the target social data includes social data of at least one target user.
The feature extraction module 1002 is configured to perform feature extraction on the social network topological graph to obtain a feature matrix and an adjacency matrix corresponding to the social network topological graph.
A feature vector generation module 1003, configured to perform feature vector generation on the feature matrix and the adjacency matrix input graph neural network model, to obtain a target feature vector of the social network topological graph.
And a risk value prediction module 1004, configured to input the target feature vector into a risk value prediction model to perform risk value prediction, so as to obtain a risk value prediction result of each target user.
It should be noted that, for convenience and simplicity of description, it may be clearly understood by those skilled in the art that the specific working processes of the apparatus and each module described above may refer to the corresponding processes in the foregoing method embodiments, and details are not described herein again.
The apparatus described above may be implemented in the form of a computer program which is executable on a computer device as shown in fig. 4.
Referring to fig. 4, fig. 4 is a schematic block diagram of a computer device according to an embodiment of the present disclosure.
Referring to fig. 4, the computer device includes a processor and a memory connected by a system bus, wherein the memory may include a storage medium and an internal memory. The storage medium may be a nonvolatile storage medium or a volatile storage medium.
The processor is used to provide computing and control capabilities to support the operation of the entire computer device.
The internal memory provides an environment for the execution of a computer program on a storage medium, which when executed by a processor causes the processor to perform any of the risk value prediction methods.
It should be understood that the Processor may be a Central Processing Unit (CPU), and the Processor may be other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, etc. Wherein a general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
Wherein, in one embodiment, the processor is configured to execute a computer program stored in the memory to implement the steps of:
obtaining target social data, constructing a topological graph of the target social data, and obtaining a social network topological graph, wherein the target social data comprises social data of at least one target user; performing feature extraction on the social network topological graph to obtain a feature matrix and an adjacency matrix corresponding to the social network topological graph; generating a feature vector of the feature matrix and the adjacency matrix input graph neural network model to obtain a target feature vector of the social network topological graph; and inputting the target characteristic vectors into a risk value prediction model for risk value prediction to obtain a risk value prediction result of each target user.
In one embodiment, before implementing feature vector generation of the feature matrix and the adjacency matrix input graph neural network model to obtain a target feature vector of the social network topology, the processor is further configured to implement:
obtaining sample social data, and constructing a sample social network topological graph according to the sample social data, wherein the sample social data comprises social data of at least one sample user; inputting a feature matrix and an adjacency matrix corresponding to the sample social network topological graph into the graph neural network model for feature vector generation to obtain a first feature vector; constructing a contrast characteristic matrix corresponding to the characteristic matrix, and inputting the contrast characteristic matrix and the adjacency matrix into the graph neural network model for characteristic extraction to obtain a second characteristic vector; generating at least one clustering sub-network corresponding to the social network topological graph, and determining a global feature vector corresponding to each clustering sub-network; generating at least one positive sample pair and at least one negative sample pair corresponding to the sample social network topology graph according to the first feature vector, the second feature vector and the global feature vector corresponding to each clustering subnetwork; calculating loss function values of all the positive sample pairs and all the negative sample pairs to obtain the loss function values corresponding to the sample social network topological graph; and if the loss function value is larger than a preset first loss threshold value, adjusting parameters of the graph neural network model, carrying out next round of training until the obtained loss function value meets a preset condition, and ending the training to obtain the trained graph neural network model.
In one embodiment, when implementing to construct the contrast feature matrix corresponding to the feature matrix, the processor is configured to implement:
and adjusting the positions of the elements in each row in the feature matrix, and determining the adjusted feature matrix as the contrast feature matrix.
In one embodiment, the processor, when implementing generating at least one clustering subnetwork corresponding to the social network topology, is configured to implement:
and based on a clustering algorithm, clustering the characteristic matrix to obtain at least one clustering subnetwork.
In one embodiment, the processor, when being configured to determine the global feature vector corresponding to each of the clustering subnetworks, is configured to:
extracting a feature vector from the first feature vector to obtain a feature vector corresponding to each node in each clustering subnetwork; and adding the feature vectors corresponding to all the nodes in each clustering sub-network to obtain an average value, so as to obtain a global feature vector corresponding to each clustering sub-network.
In one embodiment, the processor, in implementing generating at least one positive sample pair and at least one negative sample pair corresponding to the sample social network topology map based on the first feature vector, the second feature vector, and the global feature vector corresponding to each of the clustering sub-networks, is configured to implement:
extracting the first feature vector to obtain a first class of feature vectors of each node in each clustering sub-network; extracting the second feature vector to obtain a second class of feature vectors of each node in each clustering sub-network; combining the global feature vector corresponding to each clustering sub-network with the corresponding first-class feature vector to obtain at least one positive sample pair; and combining the global feature vector corresponding to each clustering sub-network with the corresponding second-class feature vector to obtain at least one negative sample pair.
In one embodiment, the processor, when performing the calculation of the loss function value for all the positive sample pairs and all the negative sample pairs to obtain the loss function value corresponding to the sample social network topology, is configured to:
scoring each positive sample pair according to a preset discriminator to obtain a first scoring value of each positive sample pair; scoring each negative sample pair according to the discriminator to obtain a second score value of each negative sample pair; and calculating a loss function value according to all the first score values and all the second score values based on a preset loss function to obtain the loss function value corresponding to the sample social network topological graph.
In one embodiment, the risk value prediction result comprises a risk value; after the processor inputs the target feature vectors into a risk value prediction model for risk value prediction and obtains a risk value prediction result of each target user, the processor is further configured to:
determining candidate services corresponding to each target user according to the risk value corresponding to each target user based on the corresponding relation between the preset risk value and the service; and recommending corresponding candidate services to each target user.
The embodiment of the present application further provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, where the computer program includes program instructions, and the processor executes the program instructions to implement any one of the risk value prediction methods provided in the embodiments of the present application.
For example, the program is loaded by a processor and may perform the following steps:
obtaining target social data, constructing a topological graph of the target social data, and obtaining a social network topological graph, wherein the target social data comprises social data of at least one target user; extracting features of the social network topological graph to obtain a feature matrix and an adjacent matrix corresponding to the social network topological graph; generating a feature vector of the feature matrix and the adjacency matrix input graph neural network model to obtain a target feature vector of the social network topological graph; and inputting the target characteristic vectors into a risk value prediction model for risk value prediction to obtain a risk value prediction result of each target user.
The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiment, for example, a hard disk or a memory of the computer device. The computer readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital Card (SD Card), a Flash memory Card (Flash Card), and the like provided on the computer device.
Further, the computer-readable storage medium may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function, and the like; the storage data area may store data created according to the use of the blockchain node, and the like.
The block chain referred by the application is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, a consensus mechanism, an encryption algorithm and the like. A block chain (Blockchain), which is essentially a decentralized database, is a string of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, which is used for verifying the validity (anti-counterfeiting) of the information and generating a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like.
The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily think of various equivalent modifications or substitutions within the technical scope of the present application, and these modifications or substitutions should be covered within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. A method for risk value prediction, comprising:
obtaining target social data, constructing a topological graph of the target social data, and obtaining a social network topological graph, wherein the target social data comprises social data of at least one target user;
performing feature extraction on the social network topological graph to obtain a feature matrix and an adjacency matrix corresponding to the social network topological graph;
generating a feature vector of the feature matrix and the adjacency matrix input graph neural network model to obtain a target feature vector of the social network topological graph;
and inputting the target characteristic vectors into a risk value prediction model for risk value prediction to obtain a risk value prediction result of each target user.
2. The method of predicting a risk value according to claim 1, wherein before generating the feature vector of the feature matrix and the adjacency matrix input graph neural network model to obtain the target feature vector of the social network topology, the method further comprises:
obtaining sample social data, and constructing a sample social network topological graph according to the sample social data, wherein the sample social data comprises social data of at least one sample user;
inputting a feature matrix and an adjacent matrix corresponding to the sample social network topological graph into the graph neural network model for feature vector generation to obtain a first feature vector;
constructing a contrast characteristic matrix corresponding to the characteristic matrix, and inputting the contrast characteristic matrix and the adjacency matrix into the graph neural network model for characteristic extraction to obtain a second characteristic vector;
generating at least one clustering sub-network corresponding to the social network topological graph, and determining a global feature vector corresponding to each clustering sub-network;
generating at least one positive sample pair and at least one negative sample pair corresponding to the sample social network topology graph according to the first feature vector, the second feature vector and the global feature vector corresponding to each clustering subnetwork;
calculating loss function values of all the positive sample pairs and all the negative sample pairs to obtain the loss function values corresponding to the sample social network topological graph;
and if the loss function value is larger than a preset first loss threshold value, adjusting parameters of the graph neural network model, carrying out next round of training until the obtained loss function value meets a preset condition, and ending the training to obtain the trained graph neural network model.
3. The method for predicting the risk value according to claim 2, wherein the constructing of the comparison feature matrix corresponding to the feature matrix comprises:
and adjusting the positions of the elements in each row in the feature matrix, and determining the adjusted feature matrix as the contrast feature matrix.
4. The method of claim 2, wherein the generating at least one clustering sub-network corresponding to the social network topology comprises:
based on a clustering algorithm, clustering the characteristic matrix to obtain at least one clustering subnetwork;
the determining the global feature vector corresponding to each clustering subnetwork comprises:
extracting a feature vector from the first feature vector to obtain a feature vector corresponding to each node in each clustering subnetwork;
and adding the feature vectors corresponding to all the nodes in each clustering sub-network to obtain an average value, so as to obtain a global feature vector corresponding to each clustering sub-network.
5. The method of predicting risk values of claim 2, wherein generating at least one positive sample pair and at least one negative sample pair corresponding to the sample social network topology based on the first feature vector, the second feature vector, and the global feature vector corresponding to each of the clustering subnetworks comprises:
extracting the first feature vector to obtain a first class of feature vectors of each node in each clustering sub-network;
extracting the second feature vector to obtain a second class of feature vectors of each node in each clustering subnetwork;
combining the global feature vector corresponding to each clustering sub-network with the corresponding first-class feature vector to obtain at least one positive sample pair;
and combining the global feature vector corresponding to each clustering sub-network with the corresponding second-class feature vector to obtain at least one negative sample pair.
6. The method of predicting risk values according to claim 2, wherein the calculating the loss function values for all the positive sample pairs and all the negative sample pairs to obtain the loss function values corresponding to the sample social network topology includes:
scoring each positive sample pair according to a preset discriminator to obtain a first score value of each positive sample pair;
scoring each negative sample pair according to the discriminator to obtain a second score value of each negative sample pair;
and calculating a loss function value according to all the first score values and all the second score values based on a preset loss function to obtain the loss function value corresponding to the sample social network topological graph.
7. The method of predicting risk values of claim 1, wherein the risk value prediction result comprises a risk value; after the target feature vector is input into a risk value prediction model for risk value prediction and a risk value prediction result of each target user is obtained, the method further includes:
determining candidate services corresponding to each target user according to the risk value corresponding to each target user based on the corresponding relation between the preset risk value and the service;
and recommending the corresponding candidate service to each target user.
8. A risk value prediction apparatus, comprising:
the topological graph construction module is used for obtaining target social data, carrying out topological graph construction on the target social data and obtaining a social network topological graph, wherein the target social data comprises social data of at least one target user;
the characteristic extraction module is used for extracting characteristics of the social network topological graph to obtain a characteristic matrix and an adjacency matrix corresponding to the social network topological graph;
the characteristic vector generation module is used for generating characteristic vectors of the characteristic matrix and the adjacency matrix input graph neural network model to obtain target characteristic vectors of the social network topological graph;
and the risk value prediction module is used for inputting the target characteristic vectors into a risk value prediction model to perform risk value prediction, and obtaining a risk value prediction result of each target user.
9. A computer device, wherein the computer device comprises a memory and a processor;
the memory for storing a computer program;
the processor for executing the computer program and when executing the computer program implementing a risk value prediction method as claimed in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that a computer program is stored which, when being executed by a processor, causes the processor to carry out a risk value prediction method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210933030.2A CN115271980A (en) | 2022-08-04 | 2022-08-04 | Risk value prediction method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210933030.2A CN115271980A (en) | 2022-08-04 | 2022-08-04 | Risk value prediction method and device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115271980A true CN115271980A (en) | 2022-11-01 |
Family
ID=83748252
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210933030.2A Pending CN115271980A (en) | 2022-08-04 | 2022-08-04 | Risk value prediction method and device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115271980A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117061252A (en) * | 2023-10-12 | 2023-11-14 | 杭州智顺科技有限公司 | Data security detection method, device, equipment and storage medium |
| CN117113148A (en) * | 2023-08-30 | 2023-11-24 | 上海智租物联科技有限公司 | Risk identification method, device and storage medium based on time sequence diagram neural network |
-
2022
- 2022-08-04 CN CN202210933030.2A patent/CN115271980A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117113148A (en) * | 2023-08-30 | 2023-11-24 | 上海智租物联科技有限公司 | Risk identification method, device and storage medium based on time sequence diagram neural network |
| CN117113148B (en) * | 2023-08-30 | 2024-05-17 | 上海智租物联科技有限公司 | Risk identification method, device and storage medium based on time sequence diagram neural network |
| CN117061252A (en) * | 2023-10-12 | 2023-11-14 | 杭州智顺科技有限公司 | Data security detection method, device, equipment and storage medium |
| CN117061252B (en) * | 2023-10-12 | 2024-03-12 | 杭州智顺科技有限公司 | Data security detection method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tama et al. | Tree-based classifier ensembles for early detection method of diabetes: an exploratory study | |
| CN108491511B (en) | Data mining method and device based on graph data and model training method and device | |
| CN115271980A (en) | Risk value prediction method and device, computer equipment and storage medium | |
| CN111966904B (en) | Information recommendation method and related device based on multi-user portrait model | |
| US20100185935A1 (en) | Systems and methods for community detection | |
| WO2019102984A1 (en) | Learning device and learning method, identification device and identification method, program, and recording medium | |
| CN113822315A (en) | Attribute graph processing method and device, electronic equipment and readable storage medium | |
| CN113011529B (en) | Training method, training device, training equipment and training equipment for text classification model and readable storage medium | |
| CN112163929B (en) | Service recommendation method, device, computer equipment and storage medium | |
| Shehab et al. | Semi-supervised policy recommendation for online social networks | |
| CN112348079B (en) | Data dimension reduction processing method and device, computer equipment and storage medium | |
| CN113656699B (en) | User feature vector determining method, related equipment and medium | |
| CN116471281A (en) | Decentralised service combination method considering node selfiness | |
| CN113762648B (en) | Method, device, equipment and medium for predicting male Wei Heitian goose event | |
| CN108038496B (en) | Marriage and love object matching data processing method and device based on big data and deep learning, computer equipment and storage medium | |
| CN116304518A (en) | Heterogeneous graph convolution neural network model construction method and system for information recommendation | |
| Wang et al. | Enhanced by mobility? Effect of users’ mobility on information diffusion in coupled online social networks | |
| Farina et al. | Asynchronous distributed learning from constraints | |
| CN114511024A (en) | Node classification method, apparatus, device, medium, and computer program product | |
| CN115099875A (en) | Data classification method based on decision tree model and related equipment | |
| CN114898184A (en) | Model training method, data processing method and device and electronic equipment | |
| CN110163761B (en) | Suspicious item member identification method and device based on image processing | |
| CN114676755A (en) | Non-supervision domain self-adaptive classification method based on graph convolution network | |
| Jahanbakhsh et al. | Human contact prediction using contact graph inference | |
| Azeez Joodi et al. | A New Proposed Hybrid Learning Approach with Features for Extraction of Image Classification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |