Trust feature complementation and trust relationship prediction method and system for trust information loss

Abstract

The invention provides a trust information missing trust feature complementation and trust relationship prediction method and system, and the method comprises the steps: constructing a social relation graph and a trust relation graph, and obtaining social attribute features and trust attribute features respectively; compressing the social attribute features and the trust attribute features, and constructing a feature converter to convert the compressed social attribute features; constructing a user trust relationship prediction model, and outputting a prediction label by taking the compression characteristics of the user as input; taking the converted and compressed social attribute features as input, and outputting a prediction tag; and taking a real label as supervision information, combining reconstruction loss and information bottleneck loss, training the model, and outputting a trust relationship. According to the method, the social behavior graph and the trust relation graph are constructed, and the social attribute characteristics and the trust attribute characteristics of the user are learned respectively. According to the multi-view learning mode, the feature information of the user can be more comprehensively captured, and the accuracy of trust relationship prediction is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of social relationship prediction, and in particular to a method and system for completing trust features and predicting trust relationships when trust information is missing. Background Art

[0002] Trust assessment is the process of evaluating trust by quantifying the attributes that affect trust. It has many applications in sensor networks, social networks, mobile crowdsourcing, blockchain and other fields. Traditional trust assessment schemes are based on heuristic trust theory. Usually, based on user attributes and scores between users, the above information is integrated based on a predefined trust model to obtain a comprehensive trust score of the user or a trust relationship between users. The effectiveness of such methods depends greatly on the predefined model, has poor universality, and cannot be flexibly applied to different scenarios. In view of the limitations of traditional trust assessment algorithms, some scholars have applied machine learning algorithms to trust assessment, and automatically processed trust-related data through machine learning models, extracted user trust features and predicted trust relationships between users. Traditional trust assessment algorithms based on machine learning are usually based on matrix decomposition theory. By representing the trust relationship between users as a matrix and decomposing the matrix, the node trust features and the trusted features are obtained, and then the matrix multiplication is performed to obtain the complete trust relationship matrix. For example, Fazeli et al. explored the effectiveness of different measurement methods in trust relationship evaluation based on matrix decomposition and proposed a matrix decomposition trust evaluation algorithm based on implicit trust score. By incorporating implicit trust score into matrix decomposition instead of explicit trust relationship, the model was experimentally confirmed to achieve similar performance without explicit trust score. Similar work has also been applied in. The matrix decomposition-based method has shown good effectiveness, but this method is based on a shallow model and cannot mine the deep complex features of nodes, making it difficult to effectively discover the complex nonlinear mapping relationship between user trust features and trust relationships.

[0003] In response to the above problems, some scholars have introduced deep learning models for the extraction of trust features and the prediction of trust relationships. By modeling the trust relationship as a graph and processing it with a graph convolutional neural network, this type of model has achieved remarkable results. For example: Guardian uses a graph convolutional neural network to process and fuse the two trust features of the user separately, and builds a trust relationship prediction scheme based on a graph convolutional neural network; on this basis, KGTrust further integrates other social networks besides the trust relationship network to expand the trust relationship graph and achieve a certain performance improvement; GATTrust integrates trust features from different fields based on the attention mechanism to improve trust prediction performance; some work further considers the dynamic nature of trust and builds a dynamic prediction mechanism for trust relationships based on a dynamic graph neural network, such as TrustGuard. The above work has solved the problem of trust relationship prediction very well.

[0004] However, the above work has the following problems: 1. Existing trust relationship prediction methods based on graph neural networks usually require obtaining sufficient trust relationship interactions among nodes. However, in practical applications, it is not realistic to ensure that all nodes have sufficient trust interactions, which makes it difficult for existing trust relationship prediction algorithms based on graph neural networks to play a role. 2. Although some algorithms (such as KGTrust) introduce external knowledge to supplement the trust relationship, these methods still require the predicted nodes to have trust interactions, and it is also difficult to overcome the problem of lack of trust relationships. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for completing trust features and predicting trust relationships when trust information is missing, so as to solve the above problems.

[0006] To achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a trust feature completion and trust relationship prediction method for missing trust information, comprising: Based on the social behaviors of users and the trust relationships between users, a social relationship graph and a trust relationship graph are constructed respectively, and based on the constructed social relationship graph and the trust relationship graph, social attribute features and trust attribute features are obtained respectively; Compress the social attribute features and trust attribute features, and build a feature converter to convert the compressed social attribute features; Build a user trust relationship prediction model, take the user's compressed features as input, and output the predicted label; then take the converted compressed social attribute features as input, and output the predicted label; Using the true label as supervision information, the model is trained by combining reconstruction loss and information bottleneck loss to output the trust relationship.

[0007] Furthermore, the step of constructing a social relationship graph and a trust relationship graph based on the social behaviors of users and the trust relationships between users respectively includes: By preprocessing the user's behavior information on social networks, a user social behavior network diagram is constructed for different types of user social behaviors. ; According to the trust relationship between users, a user trust relationship network diagram is constructed. , and builds initial trust features for users based on their trust relationships , specifically: Standardize the trust characteristics between users to form discrete trust levels, and model the trust relationship between users as a directed weighted graph; By user For example, according to the trust relationship between the user and other users and the trust relationship between other users and the user, the out-degree trust features are initialized respectively. and in-degree trust characteristics , thus forming a complete trust relationship network diagram .

[0008] Furthermore, the social attribute characteristics and trust attribute characteristics are obtained based on constructing the social relationship graph and the trust relationship graph, respectively, including: exist A graph neural network model based on meta-path is constructed for feature extraction. The features of different neighbor nodes and the inter-semantic features extracted under different meta-paths are fused based on the intra-semantic attention mechanism and the inter-semantic attention mechanism, and finally a social attribute feature set of the user node is formed. , specifically: according to The relationship type definition meta path collection in ; According to the defined meta path, Divide into Figure ; Using graph convolutional neural network and multi-head attention mechanism Process and extract user features ; For different pictures Extracted user features , establish an attention mechanism to integrate it into the user's social features ; Constructing the initial features of users based on the trust relationship of nodes and trust relationship network , using the attention mechanism to fuse the node’s neighbor information and form the trust feature of the user node , specifically: Build The graph convolutional neural network of the layer is based on the user's initial in-degree trust feature The initial feature , assuming that The output of the layer is ,if Existence right Trust Edge , then it is believed that yes The in-degree neighbors of , first calculate the attention score as follows:

[0009] Based on the score, the fused features are calculated:

[0010] go through After the layer is trained, its in-degree features are output ; Build The graph convolutional neural network of the layer is based on the user's initial in-degree trust feature The initial feature , assuming that The output of the layer is ,if Existence right Trust Edge , then it is believed that yes The out-degree neighbors of , first calculate the attention score as follows:

[0011] Based on the score, the fused features are calculated:

[0012] go through After the layer is trained, its in-degree features are output ; Will and After concatenation, input the fully connected network and activation layer, output Trust characteristics .

[0013] Furthermore, the compressing of the social attribute features and the trust attribute features includes: Given and , respectively construct the encoder and right and Processing to obtain and , construct information bottleneck loss based on information bottleneck theory as follows: .

[0014] Furthermore, given , further construct the conversion encoder right Processing for learning arrive The mapping relationship based on the feature conversion loss is as follows: .

[0015] Furthermore, the user trust relationship prediction model is constructed, taking the compressed features of the user as input and outputting the prediction label; and then taking the converted compressed social attribute features as input and outputting the prediction label, including: Based on the multi-layer perceptron MLP, a user trust relationship prediction model is constructed. and Compression characteristics and is the input and output is the predicted label ; At the same time and is the input and output is the predicted label .

[0016] Furthermore, the real label is used as the supervision information, and the reconstruction loss and the information bottleneck loss are combined to train the model and output the trust relationship, including: The feature extraction module, feature compression module and feature conversion module are used to train the data set and Process the graph data and finally obtain the predicted relationship , combined with a real trust relationship Constructing cross entropy loss ; Combination , and , construct the overall loss as follows:

[0017] Based on the training data and Conduct model training and guide model convergence.

[0018] In a second aspect, the present invention provides a trust feature completion and trust relationship prediction system for missing trust information, comprising: A graph construction module, used to construct a social relationship graph and a trust relationship graph based on the social behavior of users and the trust relationship between users, and obtain social attribute features and trust attribute features based on the constructed social relationship graph and the trust relationship graph; A compression conversion module is used to compress social attribute features and trust attribute features, and build a feature converter to convert the compressed social attribute features; The prediction label output module is used to build a user trust relationship prediction model, which takes the user's compressed features as input and outputs the prediction label; it also takes the converted compressed social attribute features as input and outputs the prediction label; The training output module is used to train the model using the true label as supervision information, combined with the reconstruction loss and information bottleneck loss, and output the trust relationship.

[0019] In a third aspect, the present invention provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the method for completing the trust features and predicting the trust relationship when the trust information is missing are implemented.

[0020] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method for completing trust features and predicting trust relationships when trust information is missing.

[0021] Compared with the prior art, the present invention has the following technical effects: The present invention introduces a multi-view node social relationship graph as a feature supplement, and learns a mapper from social attribute features to trust features on a trust relationship graph composed of small-scale nodes with trust interaction relationships. Then, on a large-scale trust relationship missing data set, the learned feature mapper is used to complete the conversion of social attribute features to trust features. In this way, even if a node lacks trust relationship data, its trust features can be effectively learned and trust relationship prediction can be performed.

[0022] The present invention constructs a social behavior graph and a trust relationship graph, and learns the social attribute characteristics and trust attribute characteristics of users respectively. This multi-view learning method can capture the characteristic information of users more comprehensively and improve the accuracy of trust relationship prediction.

[0023] Through the feature compression module and feature conversion module, the technical solution can compress irrelevant information in the original features, extract consistent information between social attribute features and trust attribute features, and further learn the mapping relationship between social attribute features and trust attribute features. These processing steps help reduce noise interference, improve the effectiveness of features, and thus enhance the accuracy of trust relationship prediction.

[0024] The present invention adopts a meta-path-based heterogeneous graph neural network, which can be effectively applied to the modeling and feature learning of different types of social relationships. This enables the model to handle different types of social data and trust relationships more flexibly, enhancing the universality of the model.

[0025] The present invention adopts a modular design concept, including a composition module, a feature learning module, a feature compression and conversion module, a relationship prediction module, etc. This design method allows the model to be flexibly combined and adjusted according to the needs of specific application scenarios, thereby enhancing the flexibility and scalability of the model.

[0026] The present invention constructs information bottleneck loss by introducing information bottleneck theory, and the technical solution can extract useful information as much as possible during feature compression while discarding irrelevant information. This helps to reduce the complexity of the model and improve training efficiency and convergence.

[0027] The present invention constructs an overall loss function by combining information bottleneck loss, feature conversion loss and cross entropy loss, and trains the model. This multi-objective optimization method can more comprehensively consider the performance indicators of the model, promote the convergence of the model and improve its performance.

[0028] In summary, this technical solution effectively solves the problem of missing trust relationship information by introducing multi-view node social relationship graphs as feature supplements, adopting meta-path-based heterogeneous graph neural networks for feature learning, applying information bottleneck theory for feature compression and conversion, and constructing a modular model design. It improves the accuracy of trust relationship prediction, enhances the universality and flexibility of the model, and improves the training efficiency and convergence of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0030] The present invention is further described below in conjunction with the accompanying drawings: Example 1, please refer to Figure 1 The present invention provides a trust feature completion and trust relationship prediction method for missing trust information, including: Based on the social behaviors of users and the trust relationships between users, a social relationship graph and a trust relationship graph are constructed respectively, and based on the constructed social relationship graph and the trust relationship graph, social attribute features and trust attribute features are obtained respectively; Compress the social attribute features and trust attribute features, and build a feature converter to convert the compressed social attribute features; Build a user trust relationship prediction model, take the user's compressed features as input, and output the predicted label; then take the converted compressed social attribute features as input, and output the predicted label; Using the true label as supervision information, the model is trained by combining reconstruction loss and information bottleneck loss to output the trust relationship.

[0031] The trust feature completion and trust relationship prediction method for missing trust information proposed in the present invention realizes the trust relationship prediction when the trust information is incomplete or missing through a series of carefully designed steps.

[0032] By constructing a social relationship graph and a trust relationship graph, the present invention effectively extracts rich feature information from the user's social behavior and the trust relationship between users. This graph structure data representation method can capture the complex relationship and interaction mode between users, providing a solid foundation for subsequent trust relationship prediction.

[0033] Compressing social attribute features and trust attribute features helps reduce data redundancy and noise, and improves the computational efficiency and generalization ability of the model. At the same time, building a feature converter to convert compressed social attribute features can further refine and optimize the features, making them more suitable for predicting trust relationships.

[0034] The present invention constructs a user trust relationship prediction model and adopts a dual-path prediction method. First, the user's compressed features are used as input for prediction; secondly, the converted compressed social attribute features are used as input for further prediction. This dual-path design not only improves the accuracy of the prediction, but also enhances the model's sensitivity and robustness to different feature combinations.

[0035] Using the real label as supervision information and combining reconstruction loss and information bottleneck loss to train the model helps the model to maximize the retention of information related to the trust relationship while maintaining data compression efficiency. The design of this loss function not only ensures the prediction performance of the model, but also avoids the problems of overfitting and underfitting.

[0036] Through the combined effect of the above steps, the present invention can more accurately predict the trust relationship between users. Even when trust information is missing or incomplete, reliable predictions can be made by completing trust features and using social relationship information. This has important application value in the fields of social network analysis, recommendation systems, credit assessment, etc.

[0037] The method of the present invention does not rely on specific data sets or scenarios and can be flexibly applied to different social network and trust relationship prediction tasks. At the same time, with the increase of data volume and continuous training of the model, the prediction performance of the model is expected to be further improved, showing good scalability and adaptability.

[0038] In summary, the trust feature completion and trust relationship prediction method for missing trust information proposed in the present invention has significant technical effects in enhancing data representation capabilities, improving prediction accuracy, optimizing model performance, etc., and provides new ideas and methods for social network analysis and trust relationship prediction.

[0039] Embodiment 2, the invention provides a method for completing trust features and predicting trust relationships when trust information is missing, specifically comprising: Build social relationship graphs based on users' social behaviors and trust relationships between users and trust relationship diagram ; Respectively and Processing to obtain the user's social attribute characteristics and ; Based on the information bottleneck theory and encoder network, a feature compressor is constructed to obtain compressed social attribute features and trust attribute features. and ; Construct feature transformer pairs Perform the transformation and ensure the transformed features based on the reconstruction loss Approach ; The user trust relationship prediction module is built based on the Multi-Layer Perception (MLP) and Compression characteristics and is the input and output is the predicted label ; At the same time and is the input and output is the predicted label ; With real label As supervision information, the reconstruction loss and information bottleneck loss are combined to train the model; The model is applied on the target graph. The target scenario only contains a social relationship graph and lacks a trust relationship graph. The above-mentioned feature learning module, feature compression and conversion module, and relationship prediction module are used for processing to output the trust relationship.

[0040] The composition module mainly includes the following parts: Social behavior graph construction: By preprocessing the user's behavior information on social networks (such as likes, purchases, comments, etc.), a user social behavior network graph is constructed for different types of user social behaviors. ; Trust relationship graph construction: Trust relationship network graph construction module: Build a user trust relationship network graph based on the trust relationship between users , and builds initial trust features for users based on their trust relationships ; The feature learning module mainly includes the following steps: User social attribute feature extraction module based on meta-path: A graph neural network model based on meta-path is constructed for feature extraction. The features of different neighbor nodes and the inter-semantic features extracted under different meta-paths are fused based on the intra-semantic attention mechanism and the inter-semantic attention mechanism, and finally a social attribute feature set of the user node is formed.

[0041] User trust attribute feature extraction module based on graph attention mechanism: construct the user's initial features based on the trust relationship of the node and trust relationship network , using the attention mechanism to fuse the node’s neighbor information and form the trust feature of the user node ; The feature compression and conversion module mainly includes the following steps: Feature compression module: Given and , respectively construct the encoder and right and Processing to obtain and , construct information bottleneck loss based on information bottleneck theory as follows:

[0042] The main purpose of this module is to compress irrelevant information in the original features and extract as much consistent information as possible between social attribute features and trust attribute features for trust relationship prediction; Feature conversion module: Given , further construct the conversion encoder right Processing for learning arrive The mapping relationship based on the feature conversion loss is as follows:

[0043] The model training module mainly includes the following steps: The feature extraction module, feature compression module and feature conversion module are used to train the data set and Process the graph data and finally obtain the predicted relationship , combined with a real trust relationship Constructing cross entropy loss ; Combination , and , construct the overall loss as follows:

[0044] Based on the training data and Conduct model training and guide model convergence.

[0045] The trust relationship graph construction mainly includes the following steps: User trust relationship extraction: standardize the trust characteristics between users to form discrete trust levels, and model the trust relationship between users as a directed weighted graph; User trust feature initialization: For example, according to the trust relationship between the user and other users and the trust relationship between other users and the user, the out-degree trust features are initialized respectively. and in-degree trust characteristics , thus forming a complete trust relationship network diagram ; The meta-path-based user social attribute feature extraction module mainly includes the following steps: Meta path definition: According to The relationship type in (such as like, follow, purchase, etc.) defines the meta-path set , such as “user-product-user” can form a meta-path; Isomorphic graph extraction: According to the defined meta-path, Divide into Figure ; User feature extraction based on graph attention mechanism: Graph convolutional neural network is used to integrate multi-head attention mechanism Process and extract user features ; User social feature fusion based on semantic attention mechanism: Extracted user features , establish an attention mechanism to integrate it into the user's social features ; The user trust attribute feature extraction module based on the graph attention mechanism mainly includes the following steps: In-degree trust feature embedding module based on multi-layer graph convolutional network: Building The graph convolutional neural network of the layer is based on the user's initial in-degree trust feature The initial feature , assuming that The output of the layer is ,if Existence right Trust Edge , then it is believed that yes The in-degree neighbors of , first calculate the attention score as follows:

[0046] Based on the score, the fused features are calculated:

[0047] go through After the layer is trained, its in-degree features are output

[0048] Out-degree trust feature embedding module based on multi-layer graph convolutional network: Construction The graph convolutional neural network of the layer is based on the user's initial in-degree trust feature The initial feature , assuming that The output of the layer is ,if Existence right Trust Edge , then it is believed that yes The out-degree neighbors of , first calculate the attention score as follows:

[0049] Based on the score, the fused features are calculated:

[0050] go through After the layer is trained, its in-degree features are output

[0051] Trust feature fusion and output: and After concatenation, input the fully connected network and activation layer, output Trust characteristics .

[0052] The present invention is aimed at the trust evaluation problem in social network platforms including ubiquitous social networks, mobile crowdsourcing, etc. The platform mainly includes a social network platform and For ease of representation, this patent uses Represents a social platform, which mainly provides online social services, such as instant messaging, information, shopping and other services. It is also responsible for predicting trust relationships based on users' trust relationships and social behaviors to better serve users. The algorithms involved in this patent are deployed in this middle, Maintaining social behavior graphs of users in different social scenarios And the user's trust relationship diagram ; The user set is represented as , user use The social services provided can form trust or distrust relationship information about other users based on personal preferences.

[0053] The present invention learns on a trust relationship graph composed of small-scale nodes with trust interaction relationships, and learns a mapper of social attribute features to trust features on the small-scale data set by introducing a multi-view node social relationship graph as a feature supplement; then, it processes on a data set with a large-scale lack of trust relationships, learns its social attribute features through the rich social relationships of the nodes in the data set, and uses the learned feature mapper to complete the conversion of social attribute features to trust features, thereby ensuring that even if there is a lack of trust relationships, its trust features can still be effectively learned and trust relationship predictions can be performed, effectively making up for the shortcomings of existing work.

[0054] In yet another embodiment of the present invention, a trust feature completion and trust relationship prediction system for missing trust information is provided, which can be used to implement the above-mentioned trust feature completion and trust relationship prediction method for missing trust information. Specifically, the system includes: A graph construction module, used to construct a social relationship graph and a trust relationship graph based on the social behavior of users and the trust relationship between users, and obtain social attribute features and trust attribute features based on the constructed social relationship graph and the trust relationship graph; A compression conversion module is used to compress social attribute features and trust attribute features, and build a feature converter to convert the compressed social attribute features; The prediction label output module is used to build a user trust relationship prediction model, which takes the user's compressed features as input and outputs the prediction label; it also takes the converted compressed social attribute features as input and outputs the prediction label; The training output module is used to train the model using the true label as supervision information, combined with the reconstruction loss and information bottleneck loss, and output the trust relationship.

[0055] The division of modules in the embodiments of the present invention is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present invention may be integrated into one processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0056] In another embodiment of the present invention, a computer device is provided, the computer device including a processor and a memory, the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, which are suitable for implementing one or more instructions, and are specifically suitable for loading and executing one or more instructions in a computer storage medium to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used for the operation of the trust feature completion and trust relationship prediction method for the lack of trust information.

[0057] In another embodiment of the present invention, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in a computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the trust feature completion and trust relationship prediction method for missing trust information in the above embodiment.

[0058] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0059] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0060] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0061] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0062] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A trust feature completion and trust relationship prediction method for missing trust information, characterized in that: include: Based on the social behaviors of users and the trust relationships between users, a social relationship graph and a trust relationship graph are constructed respectively, and based on the constructed social relationship graph and the trust relationship graph, social attribute features and trust attribute features are obtained respectively; Compress the social attribute features and trust attribute features, and build a feature converter to convert the compressed social attribute features; Build a user trust relationship prediction model, take the user's compressed features as input, and output the predicted label; then take the converted compressed social attribute features as input, and output the predicted label; Using the true label as supervision information, the model is trained by combining reconstruction loss and information bottleneck loss to output the trust relationship.

2. The trust feature completion and trust relationship prediction method for missing trust information according to claim 1 is characterized in that: The step of constructing a social relationship graph and a trust relationship graph based on the social behaviors of users and the trust relationships between users respectively includes: By preprocessing the user's behavior information on social networks, a user social behavior network diagram is constructed for different types of user social behaviors. ; According to the trust relationship between users, a user trust relationship network diagram is constructed. , and builds initial trust features for users based on their trust relationships , specifically: Standardize the trust characteristics between users to form discrete trust levels, and model the trust relationship between users as a directed weighted graph; By user For example, according to the trust relationship between the user and other users and the trust relationship between other users and the user, the out-degree trust features are initialized respectively. and in-degree trust characteristics , thus forming a complete trust relationship network diagram .

3. The trust feature completion and trust relationship prediction method for missing trust information according to claim 2 is characterized in that: The step of obtaining social attribute features and trust attribute features based on constructing a social relationship graph and a trust relationship graph respectively includes: exist A graph neural network model based on meta-path is constructed for feature extraction. The features of different neighbor nodes and the inter-semantic features extracted under different meta-paths are fused based on the intra-semantic attention mechanism and the inter-semantic attention mechanism, and finally a social attribute feature set of the user node is formed. , specifically: according to The relationship type definition meta path collection in ; According to the defined meta path, Divide into Figure ; Using graph convolutional neural network and multi-head attention mechanism Process and extract user features ; For different pictures Extracted user features , establish an attention mechanism to integrate it into the user's social features ; Constructing the initial features of users based on the trust relationship of nodes and trust relationship network , using the attention mechanism to fuse the node’s neighbor information and form the trust feature of the user node , specifically: Build The graph convolutional neural network of the layer is based on the user's initial in-degree trust feature The initial feature , assuming that The output of the layer is ,if Existence right Trust Edge , then it is believed that yes The in-degree neighbors of , first calculate the attention score as follows: Based on the score, the fused features are calculated: go through After the layer is trained, its in-degree features are output ; Build The graph convolutional neural network of the layer is based on the user's initial in-degree trust feature The initial feature , assuming that The output of the layer is ,if Existence right Trust Edge , then it is believed that yes The out-degree neighbors of , first calculate the attention score as follows: Based on the score, the fused features are calculated: go through After the layer is trained, its in-degree features are output ; Will and After concatenation, input the fully connected network and activation layer, output Trust characteristics .

4. The trust feature completion and trust relationship prediction method for missing trust information according to claim 1, characterized in that: The compressing of the social attribute features and the trust attribute features includes: Given and , respectively construct the encoder and right and Processing to obtain and , construct information bottleneck loss based on information bottleneck theory as follows: 。 5. The trust feature completion and trust relationship prediction method for missing trust information according to claim 4 is characterized in that: Given , further construct the conversion encoder right Processing for learning arrive The mapping relationship based on the feature conversion loss is as follows: 。 6. The trust feature completion and trust relationship prediction method for missing trust information according to claim 1, characterized in that: The user trust relationship prediction model is constructed, the compressed features of the user are used as input, and the predicted labels are output; and the converted compressed social attribute features are used as input, and the predicted labels are output, including: Based on the multi-layer perceptron MLP, a user trust relationship prediction model is constructed. and Compression characteristics and is the input and output is the predicted label ; At the same time and is the input and output is the predicted label .

7. The trust feature completion and trust relationship prediction method for missing trust information according to claim 1, characterized in that: The real label is used as the supervision information, and the reconstruction loss and the information bottleneck loss are combined to train the model and output the trust relationship, including: The feature extraction module, feature compression module and feature conversion module are used to train the data set and Process the graph data and finally obtain the predicted relationship , combined with a real trust relationship Constructing cross entropy loss ; Combination , and , construct the overall loss as follows: Based on the training data and Conduct model training and guide model convergence.

8. A trust feature completion and trust relationship prediction system for missing trust information, characterized in that: include: A graph construction module, used to construct a social relationship graph and a trust relationship graph based on the social behavior of users and the trust relationship between users, and obtain social attribute features and trust attribute features based on the constructed social relationship graph and the trust relationship graph; A compression conversion module is used to compress social attribute features and trust attribute features, and build a feature converter to convert the compressed social attribute features; The prediction label output module is used to build a user trust relationship prediction model, which takes the user's compressed features as input and outputs the prediction label; it also takes the converted compressed social attribute features as input and outputs the prediction label; The training output module is used to train the model using the true label as supervision information, combined with the reconstruction loss and information bottleneck loss, and output the trust relationship.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for completing trust features and predicting trust relationships when trust information is missing are implemented as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for completing trust features and predicting trust relationships when trust information is missing are implemented as described in any one of claims 1 to 7.

Images