WO2024007599A1

WO2024007599A1 - Heterogeneous graph neural network-based method and apparatus for determining target service

Info

Publication number: WO2024007599A1
Application number: PCT/CN2023/077879
Authority: WO
Inventors: 宋孟楠; 王磊
Original assignee: 上海淇玥信息技术有限公司
Priority date: 2022-07-04
Filing date: 2023-02-23
Publication date: 2024-01-11
Also published as: CN115310762A

Abstract

The present invention relates to the technical field of computer information processing, and provides a heterogeneous graph neural network-based method and apparatus for determining a target service. The method comprises: obtaining social information of a target to be evaluated; constructing a social relationship heterogeneous graph according to the social information, and determining an attention weight corresponding to the target; and inputting the attention weight into a preset security assessment model to calculate a degree of credibility of the target. According to the present invention, a social relationship heterogeneous graph is constructed according to social information of a target to be evaluated, an attention weight corresponding to the target is calculated, an attention weight vector is inputted into a preset security assessment model to calculate a degree of credibility of the target, and a service policy is determined according to the degree of credibility. In the present solution, by using the degree of credibility obtained by means of the social relationship of the target, a response policy is promptly made for a service provider before providing a service, thereby preventing loss caused by mismatching the provided service, and guaranteeing the data security of the service.

Description

Target service determination method and device based on heterogeneous graph neural network

Technical field

The invention belongs to the technical field of computer information processing, and more specifically relates to a target service determination method, device, electronic equipment and computer-readable medium based on heterogeneous graph neural network.

Background technique

For service providers, a comprehensive and in-depth analysis of service users can help provide better services to service users. However, in many cases, the information submitted by service users is insufficient, especially for service users who have just registered. The service agency can only obtain simple user information. In this case, it has become an important trend to integrate service user data scattered in various places and institutions.

In the field of data services, data is insecure. Therefore, organizations usually use a large amount of historical sample data to train the scoring model and generate a scoring model through training to judge the trust of the target to be evaluated. However, when the information collected by the organization is relatively small, Over time, it will be very difficult to determine the services to be provided, and the prediction results will be inaccurate. If the prediction results are not accurate, it will also be difficult to formulate a matching service strategy.

Contents of the invention

(1) Technical problems to be solved

The present invention aims to solve the technical problem of how to accurately predict the trust degree of a target to be evaluated when there is less information.

(2) Technical solutions

In order to solve the above technical problems, one aspect of the present invention proposes a target service determination method based on heterogeneous graph neural network, including:

Obtain social information of the target to be evaluated;

Construct a heterogeneous social relationship graph with different types of social relationships based on the social information, and determine the attention weight corresponding to the target to be evaluated;

Input the attention weight into a preset security evaluation model to calculate the trust degree of the target to be evaluated;

Based on the comparison result between the trust degree and the trust degree trend policy, the service policy of the target to be evaluated is determined and pushed.

According to a preferred embodiment of the present invention, constructing a heterogeneous social relationship graph with different types of social relationships based on the social information and determining the attention weight corresponding to the target to be evaluated further includes:

Use the targets that have a social relationship with the target to be evaluated as neighbor nodes, and use the social information between the targets as edges to construct a heterogeneous graph of social relationships;

Calculate the node attention weights between different neighbor nodes and the target to be evaluated based on the heterogeneous social relationship graph;

Calculate the edge attention weight between the target to be evaluated and each neighbor node according to the node attention weight.

According to a preferred embodiment of the present invention, inputting the attention weight into a preset security evaluation model to calculate the trust degree of the target to be evaluated further includes:

Perform a weighted sum of the node attention weights to obtain a node attention weight vector;

A weighted sum of the edge attention weights is used to obtain an edge attention weight vector;

The node attention weight vector and edge attention weight vector of the target to be evaluated are respectively input into the security evaluation model to obtain the trust degree of the target to be evaluated.

According to a preferred embodiment of the present invention, determining the service strategy of the target to be evaluated based on the comparison result of the trust trend strategy and pushing it further includes:

Set different trust threshold intervals;

Set corresponding service policies for different trust threshold intervals;

A corresponding service policy is provided according to the trust threshold interval in which the trust level of the target to be evaluated is located.

According to a preferred embodiment of the present invention, obtaining the social information of the target to be evaluated further includes:

Obtain the social relationship information of the target to be evaluated, including call information, address book information and geographical location information.

According to a preferred embodiment of the present invention, before constructing a social relationship heterogeneous graph with different types of social relationships based on the social information and determining the attention weight corresponding to the target to be evaluated, the method further includes:

Use social information of historical users as training samples;

Construct a heterogeneous social relationship graph based on the training samples, and calculate the attention weight vector of the historical users;

The attention weight vector is input into the convolutional neural network model for training to obtain the security assessment model.

According to a preferred embodiment of the present invention, inputting the attention weight vector into a convolutional neural network model for training to obtain the security assessment model further includes:

Input the attention weight vector of historical users into the neural network model, and compare the output safety score with the actual safety score of the historical user;

The model parameters are adjusted through the loss function to minimize the error between the security score output by the model and the actual security score of the historical users, and finally the security assessment model is obtained.

According to a preferred embodiment of the present invention, the method further includes:

The training samples are regularly updated, and the security assessment model is updated based on the updated training samples.

A second aspect of the present invention proposes a target service determination device based on a heterogeneous graph neural network, including:

Information acquisition module, used to obtain social information of the target to be evaluated;

An attention weight calculation module, configured to construct a heterogeneous social relationship graph with different types of social relationships based on the social information, and determine the attention weight corresponding to the target to be evaluated;

An evaluation module, configured to input the attention weight into a preset security evaluation model to calculate the trust degree of the target to be evaluated;

A service determination module, configured to determine and push the service policy of the target to be evaluated based on the comparison result between the trust degree and the trust degree trend policy.

According to a preferred embodiment of the present invention, the attention weight calculation module further includes:

A social relationship heterogeneous graph creation unit is used to construct a social relationship heterogeneous graph using targets that have a social relationship with the target to be evaluated as neighbor nodes and social information between targets as edges;

The node attention weight calculation unit is used to calculate the node attention weight between different neighbor nodes and the target to be evaluated based on the social relationship heterogeneous graph;

An edge attention weight calculation unit is used to calculate edge attention weights between the target to be evaluated and each neighbor node according to the node attention weight.

According to a preferred embodiment of the present invention, the evaluation module further includes:

A node attention weight vector calculation unit is used to perform a weighted sum of the node attention weights to obtain a node attention weight vector;

An edge attention weight vector calculation unit is used to perform a weighted sum of the edge attention weights to obtain an edge attention weight vector;

An evaluation unit is used to respectively input the node attention weight vector and the edge attention weight vector of the target to be evaluated into the security evaluation model to obtain the trust degree of the target to be evaluated.

A third aspect of the present invention proposes an electronic device, including a processor and a memory. The memory is used to store a computer executable program. When the computer program is executed by the processor, the processor executes the method. .

A fourth aspect of the present invention also provides a computer-readable medium that stores a computer-executable program. When the computer-executable program is executed, the method is implemented.

(3) Beneficial effects

This invention obtains the social information of the target to be evaluated, builds a heterogeneous social relationship graph with different types of social relationships based on the social information, determines the attention weight corresponding to the target to be evaluated, and inputs the attention weight vector into the preset The security assessment model calculates the trust degree of the target to be evaluated. In this solution, the trust degree obtained through the social relationship of the target to be evaluated allows the service provider to make timely response strategies before providing services to avoid mismatches in the provided services. loss caused and ensure the data security of the service.

Description of the drawings

Figure 1 is a system block diagram of a target service determination method and device based on a heterogeneous graph neural network according to an embodiment of the present invention;

Figure 2 is a schematic flow chart of a method for determining target services based on heterogeneous graph neural networks according to one embodiment of the present invention;

Figure 3 is a schematic diagram of a target service determination device based on a heterogeneous graph neural network according to an embodiment of the present invention.

Detailed ways

During the introduction of specific embodiments, the detailed description of the structure, performance, effects or other features is to enable those skilled in the art to fully understand the embodiments. However, this does not exclude those skilled in the art from being able to implement the present invention under specific circumstances with technical solutions that do not contain the above-mentioned structures, performances, effects or other features.

The same reference numerals in the various drawings represent the same or similar elements, components or parts, and thus repeated descriptions of the same or similar elements, components or parts may be omitted below. It should also be understood that, although the attributive terms first, second, third, etc. indicating numbering may be used herein to describe various devices, elements, components or portions, these devices, elements, components or portions should not be limited by these attributions. . In other words, these attributives are only used to distinguish one from the other. For example, the first device may also be called a second device, without departing from the essential technical solution of the present invention. Furthermore, the terms "and/or" and "and/or" are intended to include all combinations of any one or more of the listed items.

The present invention proposes a target service determination method based on a heterogeneous graph neural network. By constructing a social relationship heterogeneous graph with the social information of the target to be identified, the corresponding attention weight is calculated from the node dimension and the view dimension, and finally the attention weight is The vector inputs the trained classification model to calculate the trust level of the user.

In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

Figure 1 is a system block diagram of a target service determination method and device based on a heterogeneous graph neural network according to an exemplary embodiment.

As shown in Figure 1, the system architecture 100 may include one or more of user terminals 101, 102, 103, a network 104 and a server 105. Network 104 is a medium used to provide communication links between user terminals 101, 102, 103 and server 105. Network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.

It should be understood that the number of user terminals, networks and servers in Figure 1 is only illustrative. Depending on implementation needs, there can be any number of user terminals, networks, and servers. For example, the server 105 may be a server cluster composed of multiple servers.

Users can use user terminals 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages, etc. The user terminals 101, 102, and 103 may be various electronic devices with display screens, including but not limited to smart phones, tablet computers, portable computers, desktop computers, and so on.

In some embodiments, the device security level identification method provided by the embodiment of the present invention is generally executed by the server 105. Correspondingly, the device security level identification device is generally provided in the server 105. In other embodiments, some terminals may have functions similar to those of the server to perform this method. Therefore, the device security level identification method provided by the embodiment of the present invention is not limited to being executed on the server side.

Figure 2 is a schematic flowchart of a method for determining target services based on heterogeneous graph neural networks according to one embodiment of the present invention.

As shown in Figure 2, this method includes:

S201. Obtain social information of the target to be evaluated.

In this embodiment, the target to be evaluated may be an individual user or an enterprise user. When the target to be evaluated is an individual user, the social information may include basic information authorized by the user. For example, the basic information may be call information, telephone communication Directory information, social software address book, user location information, transaction information, interactive information, etc.; social information can also include behavioral information. For example, behavioral information can be transaction information between users, communication information between users, etc.; When the target to be evaluated is an enterprise user, the social information can be equity penetration information between enterprises, news information between enterprises, social information of managers within the enterprise, etc.; the specific content of the social information can be determined according to the actual application scenario. , there is no restriction here.

In a specific application, the node to be identified can be a user, and the social information is the interaction information between the user and other users. The interaction information can include: interactions on the Internet, appearing together in the same geographical location at the same time, sharing through the same hardware device, etc. Determine whether there is an association between users by whether there is interaction, and then generate the associated user according to the method in this application, and then based on the behavior of the associated user (for example, whether there are violations, whether there are bad records, etc.) To determine whether the associated user is a suspicious person who will affect the security of user information.

In this embodiment of the present application, the node to be identified may also be a terminal device or a server, where the social information may be data interaction information, data transmission information, device status information, etc. in the terminal device or server.

In a specific application, the node to be identified can be a terminal device, and the social information can be other terminals that have had data transmission with the terminal device, or other terminals in the same network segment as the terminal device, or It is another terminal under the same company name as the terminal device. The associated terminal of the terminal device can be determined by whether data has been transmitted between the terminal devices and whether there is a physical connection relationship between the terminal devices.

In this embodiment, social information can be attention information or communication information between terminal devices or social accounts used by users; or transaction information between different targets; or, information stored in the device obtained with user authorization. Communication records and contact information; or communication information between users collected from video data; or information about the positions of employees within the company or the department information to which the employees belong, etc.

In this solution, the target to be evaluated is used as an example. With the user's authorization, the social information of the target to be evaluated is first obtained. The social information can include transaction information, call information, interactions on the Internet, and people in the same geographical location. The locations appear and go together at the same time, the same hardware device has been shared, etc. The above social situation can be data disclosed by users using communication devices on the Internet service platform. In this embodiment, the social association can be a temporary association, such as a certain Users who have had a phone call relationship with the user for one day, or are geographically close; it can also be a permanent association, such as family association, users in the phone address book, work unit association, etc. There are no specific restrictions here. The protection of user privacy can be achieved by deleting or anonymizing information in user information that can identify the user's identity. Anonymization can process data through encryption.

In this solution, since there may be social connections between users who live close to each other, it is possible to obtain historical users who are geographically close to the target to be evaluated, for example, within a kilometer of the user's location, or who live in the same community. historical users.

S202. Construct a social relationship heterogeneous graph with different types of social relationships based on the social information, and determine the attention weight corresponding to the target to be evaluated.

For the security assessment model, in the embodiment of the present invention, it is obtained through training with a large number of historical sample data. In this embodiment, after constructing a social relationship heterogeneous graph with different types of social relationships based on the social information, it is determined with Before the attention weight corresponding to the target to be evaluated, user information of a large number of historical users is first obtained, and the social information of each historical user is obtained as a training sample, a social relationship heterogeneous graph is constructed based on the training sample, and the historical users are calculated The attention weight vector, since the security situation of historical users is known, use the known safety situation as a label, input the attention weight vector obtained from these training samples into the model for training, and the output security score is consistent with the historical user Compare the actual safety scores, and adjust the model parameters through the loss function to minimize the error between the safety scores output by the model and the actual safety scores of historical users, and finally obtain a safety assessment model. The models used in the embodiments of the present invention include but are not limited to convolutional neural networks. network model.

In some embodiments, after obtaining the social information of the target to be evaluated, a social relationship heterogeneous graph is constructed based on the social information, in which the mobile phone number of the target to be evaluated is used as the user to be evaluated, and the users who have a social relationship with the target to be evaluated are respectively Each user serves as a neighbor node, and the corresponding social relationships between users are used as edges to construct a heterogeneous social relationship graph. Social relationships can be divided into different types, such as call relationships, address book relationships, geographical location relationships, etc.

This invention calculates the attention weight of the target to be evaluated through two aspects: node dimensions and view dimensions.

1) Node dimension: Use the following formula to calculate the attention weight of the node dimension:

k is included in

where u represents the user to be evaluated, i represents the neighbor node, Represents the feature vector of neighbor node i under the current social relationship heterogeneous graph, Represents the attention vector of neighbor node i that needs to be learned under the current social relationship heterogeneous graph, Represents the set of neighbor nodes of user u to be evaluated under the current social relationship heterogeneous graph, Represents the feature vector of neighbor node k under the current social relationship heterogeneous graph, Represents the attention vector of neighbor node k that needs to be learned under the current social relationship heterogeneous graph.

The attention vector represented is a vector composed of the social relationship between the user to be evaluated and the neighbor node i; further, the vector can also be formed according to the score corresponding to the social relationship. For example, there may be many differences between the user to be evaluated and the neighbor node In this solution, corresponding scores can be given to different social relationships, or a scoring function can be set to score each different social relationship, and the scores corresponding to various social relationships can be composed of attention vector.

The calculated attention weight α _ui is normalized.

After obtaining the attention coefficient of the user to be evaluated, perform a weighted summation with the surrounding neighbor node features to obtain the node attention weight vector of the user to be evaluated, expressed by the following formula:

k is included in

Among them, α _uk represents the attention weight between the user to be evaluated and neighbor node k, and e _uk represents the feature vector of neighbor node k of the user to be evaluated.

In the node dimension, users compare the influence between neighbor nodes of the same user social relationship and the user to be evaluated, such as comparing neighbor nodes in the user's call information, comparing neighbor nodes in the user's address book information, and comparing neighbor nodes near the user's geographical location. Compare.

2) View dimensions:

Calculate edge attention weights that are different from the social relationship of the user to be evaluated based on the node attention weights. Input the node attention weight vector into the fully connected layer of the neural network

Among them, ReLU is the linear rectification function, is the node attention weight vector, wi is the characteristic of neighbor node i; b is the bias term, and the initial value is a preset constant; the function of the front connection layer is rectified through a linear rectification function to obtain the node attention weight vector.

Then calculate the edge attention weight coefficient: The calculation method is the same as calculating the node attention coefficient. Finally, a weighted sum is performed, and each view is spliced to obtain the final edge attention weight vector: k is included in

in, is the edge attention weight of the edge between the target to be evaluated and neighbor node i, is the i-th element in the node attention weight vector, is the feature vector of the edge between the target to be evaluated and neighbor node i, is the k-th element in the node attention weight vector, is the feature vector of the edge between the target to be evaluated and neighbor node k, Represents the edge attention weight of the edge between the user to be evaluated and neighbor node k, is the k-th element in the node attention weight vector, and concat represents the merging function.

View dimension users compare the influence of nodes between different users' social relationships, for example, compare the influence of users' call information nodes and users' address book nodes on the target to be evaluated, or compare the influence of users' address book information nodes and nodes near the user's geographical location on the target to be evaluated. Evaluate the impact of your goals, etc.

In this embodiment, after calculating the edge attention weights between each neighbor node and the user to be evaluated, the edge attention weights between neighbor nodes with the same type of social relationship and the user to be evaluated can be summarized to obtain The average edge attention weight of nodes of this type of social relationship is used as the edge attention weight of each neighbor node of this type of social relationship and the user to be evaluated to reduce the impact of excessive attention on a single target on calculations , through this step, the average degree of attention of the users to be evaluated to the goals of different types of social relationships can be obtained. From a macro perspective, the attention of the users to be evaluated to the goals of different social relationships can be obtained, and the overall impact of excessive attention to a single target can be reduced. , improve prediction accuracy. Among them, the summary method can adopt mean calculation, harmonic average, weighted average, etc. Furthermore, the variance or volatility of the edge attention weights between nodes with the same type of social relationship and the user to be evaluated can be determined. Based on the variance or volatility, the attention of the user to be evaluated to the node with this type of social relationship can be effectively determined. Power, that is, to determine the impact of nodes with different types of social relationships on the users to be evaluated.

S203. Enter the attention weight into a preset security assessment model to calculate the trust degree of the target to be assessed.

The node attention weight vector and edge attention weight vector of the target to be evaluated, which are obtained by the weighted sum of the attention weight coefficients, are respectively input into the security assessment model trained in the above embodiment, and the values of the targets to be evaluated between the same social relationships are obtained. Trust, as well as the trust of the target to be evaluated between different social relationships, and finally the user's trust is obtained through comprehensive evaluation using different strategies.

Preferably, in order to ensure the accuracy of the results output by the text classification model, it is necessary to regularly update the training samples, select training samples that are relatively recent, count the trust of the samples, and train the security assessment model based on the updated samples, and evaluate the security assessment model. parameters are updated.

This invention obtains the social information of the target to be evaluated, builds a heterogeneous social relationship graph with different types of social relationships based on the social information, determines the attention weight corresponding to the target to be evaluated, and inputs the attention weight vector into the preset The security assessment model calculates the trust of the target to be evaluated, which solves the problem that user trust cannot be predicted through user characteristics or the prediction results are inaccurate when there is little user data. Users should make timely response strategies before performing services to avoid losses and improve service security.

S204. Based on the comparison result between the trust degree and the trust degree trend policy, determine the service policy of the target to be evaluated and push it.

For example, the trust trend strategy can be a preset trust threshold, and the statistical analysis value for the target is extracted from the user data of a large number of historical users to generate multiple preset trust intervals. Develop different service strategies for different trust intervals; the trust trend strategy can also set trust thresholds based on the trend of changes in trust over time. For example, you can set multiple different baseline trust thresholds and a unit trust degree threshold. When the change trend of social relationships over time is that the trust degree of the target to be identified gradually decreases, a change curve is generated based on the trust degree change trend, and the change curve is derived to obtain the reduction rate. Multiple different baseline trust degree thresholds are Subtract the reduction rate multiplied by the unit trust threshold to obtain the trust threshold corresponding to the current trust change trend, and determine the final service strategy based on the multiple trust thresholds obtained. In the same way, when the trend of trust over time is that trust gradually increases, the base trust threshold plus the growth rate multiplied by the unit trust threshold are used to determine the final service strategy based on the multiple trust thresholds obtained. Ultimately, different trust thresholds can be set under different trust trends to improve the accuracy of pushed services.

Those skilled in the art will understand that all or part of the steps to implement the above-described embodiments are implemented as programs (computer programs) executed by computer data processing equipment. When the computer program is executed, the above method provided by the present invention can be implemented. Moreover, the computer program can be stored in a computer-readable storage medium. The storage medium can be a readable storage medium such as a magnetic disk, an optical disk, a ROM, or a RAM, or it can be a storage array composed of multiple storage media, such as a magnetic disk or a storage medium. Tape storage array. The storage medium is not limited to centralized storage, it can also be distributed storage, such as cloud storage based on cloud computing.

The following describes a device embodiment of the present invention, which device can be used to perform the method embodiment of the present invention. Details described in the device embodiments of the present invention should be regarded as supplements to the above method embodiments; details not disclosed in the device embodiments of the present invention can be implemented with reference to the above method embodiments.

Figure 3 is a schematic diagram of a target service determination device based on a heterogeneous graph neural network according to one embodiment of the present invention. As shown in Figure 3, the device 300 includes:

Information acquisition module 301, used to acquire social information of the target to be evaluated;

The attention weight calculation module 302 is used to construct a social relationship heterogeneous graph with different types of social relationships based on the social information, and determine the attention weight corresponding to the target to be evaluated;

Evaluation module 303, configured to input the attention weight into a preset security evaluation model to calculate the trust degree of the target to be evaluated;

The service determination module 304 is configured to determine and push the service policy of the target to be evaluated based on the comparison result between the trust degree and the trust degree trend policy.

Among them, the information acquisition module 301 further includes:

Social relationship acquisition unit: used to obtain social relationship information of the target to be evaluated, including call information, address book information and geographical location information.

According to a preferred embodiment of the present invention, the device further includes a model training module for using the social information of historical users as training samples; constructing a social relationship heterogeneous graph based on the training samples, and calculating the attention weight of the historical users Vector; input the attention weight vector into the convolutional neural network model for training to obtain the security assessment model.

According to a preferred embodiment of the present invention, the model training module is further used to: input the attention weight vector of historical users into the neural network model, compare the output safety score with the actual safety score of the historical user; adjust the model parameters through a loss function The error between the security score output by the model and the actual security score of the historical user is minimized, and the security assessment model is finally obtained.

According to a preferred embodiment of the present invention, the attention weight calculation module 302 further includes: a social relationship heterogeneous graph creation unit, configured to use targets that have social relationships with the target to be evaluated as neighbor nodes, and use social information between targets as Edges, constructing a heterogeneous graph of social relationships;

According to the preferred embodiment of the present invention, the evaluation module 303 further includes: a node attention weight vector calculation unit, used to weight and sum the node attention weights to obtain a node attention weight vector;

According to the preferred embodiment of the present invention, the service determination module 304 further includes: a trust threshold interval setting unit, used to set different trust threshold intervals; a service policy formulation unit, used to set corresponding trust threshold intervals for different trust levels. Service policy; a service determination unit, configured to provide a corresponding service policy according to the trust threshold interval in which the trust of the target to be evaluated is located.

According to a preferred embodiment of the present invention, the device further includes an update module for regularly updating the training samples and updating the security assessment model according to the updated training samples.

The information acquisition module 301 acquires different types of social information of the user. The attention weight calculation module 302 constructs a social relationship heterogeneous graph with different types of social relationships based on the social information, determines the attention weight corresponding to the target to be evaluated, and evaluates The module 303 calculates the attention weight vector according to the calculated attention weight, and inputs the preset security assessment model to calculate the trust degree of the target to be evaluated. The service determination module 304 calculates the trust degree based on the ratio of the evaluated trust degree and the trust degree trend policy. Based on the results, the service strategy of the target to be evaluated is determined and pushed. It solves the problem that user trust cannot be predicted through user characteristics or the prediction results are inaccurate when there is little user data. Based on the predicted user trust, response strategies can be made in a timely manner before providing services to users to avoid losses. Ensure the data security of the service.

An electronic device according to an embodiment of the present invention. The electronic device includes a processor and a memory. The memory is used to store a computer executable program. When the computer program is executed by the processor, the processor executes a program based on an A target service determination method for graphing neural networks.

Electronic devices take the form of general-purpose computing devices. There can be one processor or multiple processors working together. The present invention also does not exclude distributed processing, that is, the processors can be dispersed in different physical devices. The electronic device of the present invention is not limited to a single entity, and can also be the sum of multiple physical devices.

The memory stores computer-executable programs, typically machine-readable code. The computer-readable program can be executed by the processor, so that the electronic device can perform the method of the present invention, or at least part of the steps in the method.

The memory includes volatile memory, such as a random access memory unit (RAM) and/or a cache memory unit, and may also be a non-volatile memory, such as a read-only memory unit (ROM).

Optionally, in this embodiment, the electronic device further includes an I/O interface, which is used for data exchange between the electronic device and external devices. The I/O interface may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or using any of a variety of bus structures. local bus.

It should be understood that the electronic device of the present invention may also include elements or components not shown in the above examples. For example, some electronic devices also include display units such as display screens, and some electronic devices also include human-computer interaction components, such as buttons and keyboards. As long as the electronic device can execute the computer-readable program in the memory to implement the method of the present invention or at least part of the steps of the method, it can be considered as an electronic device covered by the present invention.

A computer-readable recording medium according to an embodiment of the present invention. The computer-readable recording medium stores a computer-executable program. When the computer-executable program is executed, the above-mentioned target service determination method based on heterogeneous graph neural network of the present invention is implemented. The computer-readable storage medium may include a data signal propagated in baseband or as part of a carrier wave carrying readable program code therein. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A readable storage medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transport the program for use by or in connection with an instruction execution system, apparatus, or device. Program code contained on a readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above.

The above-mentioned computer-readable medium carries one or more programs. When the above-mentioned one or more programs are executed by a device, the computer-readable medium realizes the following functions: obtain the social information of the target to be evaluated; and according to the social information Construct a heterogeneous social relationship graph with different types of social relationships, and determine the attention weight corresponding to the target to be evaluated; input the attention weight into a preset security assessment model to calculate the trust degree of the target to be evaluated.

Through the above description of the embodiments, those skilled in the art can easily understand that the present invention can be implemented by hardware capable of executing specific computer programs, such as the system of the present invention, and the electronic processing unit, server, client, etc. included in the system. Mobile phones, control units, processors, etc. The invention may also be implemented by computer software executing the method of the invention. However, it should be noted that the computer software for executing the method of the present invention is not limited to being executed by one or a specific hardware entity. It can also be implemented by unspecified hardware in a distributed manner, such as computer program execution. Some of the method steps can be performed on the mobile client, and other parts can be performed on smart tables, smart recognition pens, etc. For computer software, the software product can be stored in a computer-readable storage medium (can be a CD-ROM, USB flash drive, mobile hard disk, etc.), or can be distributed on the network, as long as it can enable the electronic device to execute according to this method of invention.

In summary, the present invention can be implemented in hardware, or in a software module running on one or more processors, or in a combination thereof. Those skilled in the art should understand that general data processing devices such as microprocessors or digital signal processors (DSP) may be used in practice to implement some or all functions of some or all components according to embodiments of the present invention. The invention may also be implemented as an apparatus or apparatus program (eg, computer program and computer program product) for performing part or all of the methods described herein. Such a program implementing the present invention may be stored on a computer-readable medium, or may be in the form of one or more signals. Such signals may be downloaded from an Internet website, or provided on a carrier signal, or in any other form.

The above-mentioned specific embodiments further describe the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the present invention is not inherently related to any specific computer, virtual device or electronic device, and various general-purpose devices are also The invention can be implemented. The above are only specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of the present invention. within the scope of protection.

Claims

A target service determination method based on heterogeneous graph neural network, which is characterized by including:

Obtain social information of the target to be evaluated;

Construct a social relationship heterogeneous graph with different types of social relationships based on the social information, and determine the attention weight corresponding to the target to be evaluated through the social relationships of nodes in the social relationship heterogeneous graph;

Input the attention weight into a preset security evaluation model to calculate the trust degree of the target to be evaluated;

Based on the comparison result between the trust degree and the trust degree trend policy, the service policy of the target to be evaluated is determined and pushed.
The target service determination method based on heterogeneous graph neural network according to claim 1, characterized in that, constructing a social relationship heterogeneous graph with different types of social relationships according to the social information, determining the target service that is related to the target to be evaluated. The corresponding attention weight further includes:

Use the targets that have a social relationship with the target to be evaluated as neighbor nodes, and use the social information between the targets as edges to construct a heterogeneous graph of social relationships;

Calculate the node attention weights between different neighbor nodes and the target to be evaluated based on the heterogeneous social relationship graph;

Calculate the edge attention weight between the target to be evaluated and each neighbor node according to the node attention weight.
The target service determination method based on heterogeneous graph neural network according to claim 2, characterized in that the said attention weight is input into a preset security assessment model to calculate the trust degree of the target to be assessed, further comprising: :

Perform a weighted sum of the node attention weights to obtain a node attention weight vector;

A weighted sum of the edge attention weights is used to obtain an edge attention weight vector;

The node attention weight vector and edge attention weight vector of the target to be evaluated are respectively input into the security evaluation model to obtain the trust degree of the target to be evaluated.
The target service determination method based on heterogeneous graph neural network according to claim 1, characterized in that the comparison result of the trust trend strategy, determining the service strategy of the target to be evaluated and pushing it further includes:

Set different trust threshold intervals;

Set corresponding service policies for different trust threshold intervals;

A corresponding service policy is provided according to the trust threshold interval in which the trust level of the target to be evaluated is located.
The target service determination method based on heterogeneous graph neural network according to claim 1, characterized in that, after constructing a social relationship heterogeneous graph with different types of social relationships according to the social information, it is determined that the target service corresponding to the target to be evaluated is Before the attention weight, the method also includes:

Use social information of historical users as training samples;

Construct a heterogeneous social relationship graph based on the training samples, and calculate the attention weight vector of the historical users;

The attention weight vector is input into the convolutional neural network model for training to obtain the security assessment model.
The target service determination method based on heterogeneous graph neural network according to claim 5, characterized in that the attention weight vector is input into a convolutional neural network model for training to obtain the security assessment model, further comprising: :

Input the attention weight vector of historical users into the neural network model, and compare the output safety score with the actual safety score of the historical user;

The model parameters are adjusted through the loss function to minimize the error between the security score output by the model and the actual security score of the historical users, and finally the security assessment model is obtained.
The target service determination method based on heterogeneous graph neural network according to claim 6, characterized in that the method further includes:

The training samples are regularly updated, and the security assessment model is updated based on the updated training samples.
A target service determination device based on heterogeneous graph neural network, which is characterized by including:

Information acquisition module, used to obtain social information of the target to be evaluated;

An attention weight calculation module, configured to construct a heterogeneous social relationship graph with different types of social relationships based on the social information, and determine the attention weight corresponding to the target to be evaluated;

An evaluation module, configured to input the attention weight into a preset security evaluation model to calculate the trust degree of the target to be evaluated;

A service determination module, configured to determine and push the service policy of the target to be evaluated based on the comparison result between the trust degree and the trust degree trend policy.
The target service determination device based on heterogeneous graph neural network according to claim 8, characterized in that the attention weight calculation module further includes:

A social relationship heterogeneous graph creation unit is used to construct a social relationship heterogeneous graph using targets that have a social relationship with the target to be evaluated as neighbor nodes and social information between targets as edges;

The node attention weight calculation unit is used to calculate the node attention weight between different neighbor nodes and the target to be evaluated based on the social relationship heterogeneous graph;

An edge attention weight calculation unit is used to calculate edge attention weights between the target to be evaluated and each neighbor node according to the node attention weight.
The target service determination device based on heterogeneous graph neural network according to claim 9, characterized in that the evaluation module further includes:

A node attention weight vector calculation unit is used to perform a weighted sum of the node attention weights to obtain a node attention weight vector;

An edge attention weight vector calculation unit is used to perform a weighted sum of the edge attention weights to obtain an edge attention weight vector;

An evaluation unit is used to respectively input the node attention weight vector and the edge attention weight vector of the target to be evaluated into the security evaluation model to obtain the trust degree of the target to be evaluated.
The target service determination method based on heterogeneous graph neural network according to claim 8, characterized in that the comparison result of the trust trend strategy, determining the service strategy of the target to be evaluated and pushing it further includes:

Set different trust threshold intervals;

Set corresponding service policies for different trust threshold intervals;

A corresponding service policy is provided according to the trust threshold interval in which the trust level of the target to be evaluated is located.
An electronic device including a processor and a memory, the memory being used to store computer executable programs, characterized by:

When the computer program is executed by the processor, the processor performs the method of any one of claims 1-7.
A computer-readable medium storing a computer-executable program, characterized in that when the computer-executable program is executed, the method according to any one of claims 1-7 is implemented.