Rumor detection method, system, equipment, product and medium based on dynamic graph

Abstract

The present invention relates to the field of artificial intelligence technology, and provides a rumor detection method, system, device, product, and medium based on dynamic graphs, including obtaining rumor data, obtaining a rumor text sequence from the rumor data, and constructing a rumor propagation graph; dividing the rumor text sequence to establish a snapshot propagation graph of rumor snapshots; calculating a time decay parameter of a rumor node in the snapshot propagation graph, and calculating a rumor cycle parameter of the rumor snapshot through the time decay parameter; fusing the rumor cycle parameter into a GIN encoder to obtain an edge weight GIN encoder, performing multi-layer node embedding and updating on the rumor snapshot in the snapshot propagation graph through the edge weight GIN encoder to obtain a snapshot dynamic propagation graph; inputting the snapshot dynamic propagation graph into a bidirectional long short-term memory network to obtain an encoded graph representation of the rumor snapshot; and inputting the encoded graph representation into a deep neural network to complete rumor detection. The present invention can effectively improve the effect of rumor detection.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a rumor detection method, system, device, product and medium based on dynamic graphs. Background Art

[0002] Social media has become a primary channel for people to obtain news, express opinions, and engage in social interaction. However, it also carries potential risks, particularly the proliferation and rapid spread of rumors, which pose a serious threat to social stability, public trust, and economic order. Therefore, timely and effective detection of rumors on social media is crucial.

[0003] Traditional rumor detection methods typically rely on manually designed features and machine learning classifiers, which are limited by the subjectivity and limitations of feature design. With advances in deep learning technology, researchers have begun to approach rumor detection from multiple perspectives, including text content, user behavior, and propagation structure. In particular, methods based on propagation structure, which utilize graph neural networks to model the information propagation process, and dynamic graph methods that consider temporal information, have achieved significant detection results. However, existing rumor detection models using dynamic graphs typically focus only on coarse-grained temporal information, while ignoring the fine-grained temporal dynamics within and between snapshots.

[0004] The rumor detection method based on fine-grained dynamic graph neural network is an innovative research problem with important research significance and application prospects. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems existing in the related art. To this end, the present invention provides a rumor detection method, system, device, product and medium based on dynamic graphs to achieve rapid and accurate detection of online rumors.

[0006] The present invention provides a rumor detection method based on dynamic graphs, comprising:

[0007] S1: Obtain rumor data, obtain rumor text sequences and rumor source nodes from the rumor data, and construct a rumor propagation graph through the rumor source nodes;

[0008] S2: Divide the rumor text sequence to obtain block rumor sequences, and establish a snapshot propagation graph of rumor snapshots using the block rumor sequences and the rumor propagation graph;

[0009] S3: Calculate the time decay parameter of the rumor node in the snapshot propagation graph, and calculate the rumor period parameter of the rumor node according to the time decay parameter;

[0010] S4: Obtain a GIN encoder, fuse the rumor cycle parameter into the GIN encoder to obtain an edge weight GIN encoder, and perform multi-layer node embedding update on the rumor snapshot through the edge weight GIN encoder to obtain a snapshot dynamic propagation graph;

[0011] S5: Establishing a bidirectional long short-term memory network, inputting the snapshot dynamic propagation graph into the bidirectional long short-term memory network, and obtaining an encoding graph representation of the rumor snapshot;

[0012] S6: Obtain a deep neural network, input the encoded graph representation into the deep neural network, and complete rumor detection through the deep neural network.

[0013] According to the rumor detection method based on dynamic graphs provided by the present invention, step S1 further includes:

[0014] S11: Determine target software, obtain rumor data from the target software, detect the rumor data to obtain potential rumor events, extract rumor texts related to the potential rumor events from the rumor data and sort them to obtain a rumor text sequence;

[0015] S12: The rumor text that is most recent in the rumor text sequence is used as the rumor source node, and the rumor propagation graph is constructed according to the generation time and propagation order of the rumor text.

[0016] According to the rumor detection method based on dynamic graphs provided by the present invention, step S2 further includes:

[0017] S21: extracting the timestamp of the rumor text sequence, dividing the rumor text sequence according to the timestamp to obtain a block rumor sequence including rumor text blocks;

[0018] S22: Divide and textualize the rumor propagation graph through the rumor text blocks in the block rumor sequence to obtain the rumor snapshots, and obtain the snapshot propagation graph based on the rumor snapshots.

[0019] According to the rumor detection method based on dynamic graphs provided by the present invention, step S4 further includes:

[0020] S41: Obtain a GIN encoder, integrate the rumor cycle parameter into the GIN encoder, and obtain an edge weight GIN encoder. The expression of the edge weight GIN encoder is:

[0021]

[0022] in, is the m-th layer hidden feature vector of the v-th rumor node, represents the encoder parameters of the learnable m-th layer hidden feature vector, is the m-1th layer hidden feature vector of the vth rumor node, is the rumor cycle parameter of the vth rumor node, is the m-1th layer hidden feature vector of the uth neighboring node, is the set of neighboring nodes of the vth rumor node, ReLU() represents the activation function, and MLP() represents the multi-layer perceptron function;

[0023] S42: performing node embedding updates in multiple hidden layers of the rumor snapshot using multiple edge weight GIN encoders to obtain snapshot hidden layer outputs;

[0024] S43: Repeat the steps of calculating the time decay parameter of the rumor node until obtaining the snapshot hidden layer output, obtain multiple snapshot hidden layer outputs, perform mean pooling on all the snapshot hidden layer outputs, and obtain the snapshot dynamic propagation graph.

[0025] According to the rumor detection method based on dynamic graphs provided by the present invention, step S5 further includes:

[0026] S51: Acquire a forward long short-term memory network and a backward long short-term memory network, and establish the bidirectional long short-term memory network through the forward long short-term memory network and the backward long short-term memory network;

[0027] S52: Inputting the snapshot dynamic propagation graph into the forward long short-term memory network to obtain a snapshot forward coding graph representation, and inputting the snapshot dynamic propagation graph into the backward long short-term memory network to obtain a snapshot backward coding graph representation;

[0028] S53: Concatenate the forward coding graph representation of the snapshot and the backward coding graph representation of the snapshot to obtain the coding graph representation of the rumor snapshot.

[0029] According to the rumor detection method based on dynamic graphs provided by the present invention, step S6 further includes:

[0030] S61: Obtain a deep neural network, input the encoded graph representation into the fully connected layer and activation function layer of the deep neural network, and obtain a neural network output;

[0031] S62: Weight adjustment and biasing are performed on the neural network output to obtain a rumor snapshot judgment result, and rumor detection is completed through the rumor snapshot judgment result.

[0032] The present invention also provides a rumor detection system based on dynamic graphs, comprising:

[0033] Rumor propagation graph module: used to obtain rumor data, obtain rumor text sequences and rumor source nodes from the rumor data, and construct a rumor propagation graph through the rumor source nodes;

[0034] Snapshot propagation graph module: used to divide the rumor text sequence to obtain block rumor sequences, and establish a snapshot propagation graph of rumor snapshots through the block rumor sequences and the rumor propagation graph;

[0035] Rumor cycle parameter module: used to calculate the time decay parameter of the rumor node in the snapshot propagation graph, and calculate the rumor cycle parameter of the rumor node through the time decay parameter;

[0036] Snapshot dynamic propagation graph module: used to obtain a GIN encoder, fuse the rumor cycle parameter into the GIN encoder to obtain an edge weight GIN encoder, and perform multi-layer node embedding update on the rumor snapshot through the edge weight GIN encoder to obtain a snapshot dynamic propagation graph;

[0037] A coding graph representation module is used to establish a bidirectional long short-term memory network, input the snapshot dynamic propagation graph into the bidirectional long short-term memory network, and obtain a coding graph representation of the rumor snapshot;

[0038] Rumor detection module: used to obtain a deep neural network, input the coding graph representation into the deep neural network, and complete rumor detection through the deep neural network.

[0039] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of any of the above-described dynamic graph-based rumor detection methods are implemented.

[0040] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of any of the above-described dynamic graph-based rumor detection methods are implemented.

[0041] The present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can perform the steps of any of the dynamic graph-based rumor detection methods described above.

[0042] The above one or more technical solutions in the embodiments of the present invention have at least one of the following technical effects:

[0043] The dynamic graph-based rumor detection method, system, device, product, and medium provided by this invention utilizes an edge-weighted GIN encoder to perceive temporal information and employs multi-layer node embedding updates to explore temporal information within and between snapshots, thereby enabling more effective rumor detection. This invention effectively constructs an edge-weighted GIN encoder to capture fine-grained temporal features, establishes a multi-layer node embedding update mechanism, and enhances the model's learning of temporal dynamic representations through a bidirectional long short-term memory network, significantly improving the effectiveness of rumor detection tasks.

[0044] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0046] Figure 1 It is a flow chart of the rumor detection method based on dynamic graph provided by the present invention.

[0047] Figure 2 It is a structural diagram of the rumor detection system based on dynamic graph provided by the present invention.

[0048] Figure 3 It is a structural schematic diagram of the rumor detection device based on dynamic graph provided by the present invention.

[0049] Reference numerals:

[0050] 100. Rumor propagation graph module; 200. Snapshot propagation graph module; 300. Rumor period parameter module; 400. Snapshot dynamic propagation graph module; 500. Coding graph representation module; 600. Rumor detection module; 810. Processor; 820. Communication interface; 830. Memory; 840. Communication bus. DETAILED DESCRIPTION

[0051] To make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below. Obviously, the embodiments described are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The following embodiments are used to illustrate the present invention, but are not used to limit the scope of the present invention.

[0052] In the description of the embodiments of the present invention, it should be noted that the terms “first”, “second” and “third” are used for descriptive purposes only and should not be understood as indicating or implying relative importance.

[0053] In the description of the embodiments of the present invention, it should be noted that, unless otherwise specified or limited, the terms "connected" and "connection" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integral connections; mechanical connections, electrical connections; and direct connections or indirect connections through an intermediary. Those skilled in the art will understand the specific meanings of the above terms in the embodiments of the present invention based on the specific circumstances.

[0054] In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiment of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0055] The following combination Figures 1 to 3 Describing embodiments of the present invention:

[0056] Figure 1 This is a flow chart of the rumor detection method based on dynamic graphs. First, the rumor data is obtained and a rumor propagation graph is constructed. Then, a snapshot propagation graph of the rumor snapshot is established. Subsequently, the rumor cycle parameters are calculated and the edge weight GIN encoder is constructed through the rumor cycle parameters. Then, the rumor snapshot is updated with multi-layer node embedding, a bidirectional long short-term memory network is established, and a coding graph representation is obtained. Finally, the coding graph representation is input into the deep neural network to complete the rumor detection.

[0057] The present invention provides a rumor detection method based on dynamic graphs, comprising:

[0058] S1: Obtain rumor data, obtain rumor text sequences and rumor source nodes from the rumor data, and construct a rumor propagation graph through the rumor source nodes;

[0059] Furthermore, the purpose of this stage is to obtain rumor data from social media, identify rumor source nodes, and then build a rumor propagation graph. Step S1 further includes:

[0060] S11: Determine target software, obtain rumor data from the target software, detect the rumor data to obtain potential rumor events, extract rumor texts related to the potential rumor events from the rumor data and sort them to obtain a rumor text sequence;

[0061] S12: The rumor text that is most recent in the rumor text sequence is used as the rumor source node, and the rumor propagation graph is constructed according to the generation time and propagation order of the rumor text.

[0062] Regarding the above steps, the specific implementation methods in this embodiment are as follows:

[0063] First, at least one social media application is identified as a target application. In this embodiment, Twitter is selected as the target application. Posts and comments within the target application are then extracted to obtain rumor data containing rumors. The rumor data is then monitored and monitored to identify potential rumor events. If a suspected rumor event is detected multiple times within the rumor data, it is identified as a potential rumor event. Rumor text related to the potential rumor event is extracted from the rumor data, including the posts or comments related to the potential rumor event. The text is then sorted chronologically from the beginning to the end to obtain a rumor text sequence.

[0064] Then, the rumor text with the earliest time in the rumor text sequence is taken as the rumor source node, and the rumor source node is taken as the starting point to construct the rumor propagation graph G according to the generation time and propagation order of the rumor text, and , where V represents the rumor node set. Here, each rumor text is regarded as a rumor node; A represents the adjacency matrix, which is used to represent the relationship between rumor nodes. The relationship here includes forwarding and comments between rumor nodes; X is the feature matrix, which includes the text features of the content in the rumor text corresponding to the rumor node.

[0065] S2: Divide the rumor text sequence to obtain block rumor sequences, and establish a snapshot propagation graph of rumor snapshots using the block rumor sequences and the rumor propagation graph;

[0066] Furthermore, the purpose of this stage is to divide the rumor text sequence according to the timestamp, obtain a block rumor sequence including rumor text blocks, and finally obtain a snapshot propagation graph including rumor snapshots. Among them, step S2 further includes:

[0067] S21: extracting the timestamp of the rumor text sequence, dividing the rumor text sequence according to the timestamp to obtain a block rumor sequence including rumor text blocks;

[0068] S22: Divide and textualize the rumor propagation graph through the rumor text blocks in the block rumor sequence to obtain the rumor snapshots, and obtain the snapshot propagation graph based on the rumor snapshots.

[0069] Regarding the above steps, the specific implementation methods in this embodiment are as follows:

[0070] First, extract the timestamp of the rumor text sequence. The timestamp is the time corresponding to each rumor text in the rumor text sequence, and divide the rumor text sequence into S blocks with the same time interval according to the timestamp. In this embodiment, the rumor text sequence needs to be divided into S blocks with the same time interval. The time interval is The calculation method is: ,in, is the timestamp of the last rumor text in the rumor text sequence, is the timestamp of the first rumor text in the rumor text sequence. After division, a block rumor sequence including rumor text blocks is obtained, and each rumor text block includes rumor texts within the same time interval.

[0071] Then, the rumor propagation graph is divided and textualized according to the rumor text blocks, that is, the content corresponding to the rumor text blocks in the rumor propagation graph is converted into natural language to obtain rumor snapshots. The rumor snapshot includes the content of the rumor text corresponding to each rumor text block, the propagation path, etc., and then the snapshot propagation graph of the sth rumor snapshot is obtained. :

[0072]

[0073] in, is the rumor node set of the s-th rumor snapshot, is the adjacency matrix of the s-th rumor snapshot, is the feature matrix of the s-th rumor snapshot.

[0074] S3: Calculate the time decay parameter of the rumor node in the snapshot propagation graph, and calculate the rumor period parameter of the rumor node according to the time decay parameter;

[0075] Furthermore, the purpose of this stage is to introduce the time decay parameter and calculate the rumor period parameter of the rumor snapshot through the time decay parameter to show the characteristics of rumor propagation in time and period. Specifically, since the longer the time interval between the post and the response post during the spread of the rumor, the lower the popularity of the rumor can be considered, so for the rumor node in the snapshot propagation graph, the time decay parameter is introduced. :

[0076]

[0077] Among them, α is the time decay factor set according to experience, t is the timestamp of the vth rumor node in the snapshot propagation graph, It is the timestamp of the parent node of the vth rumor node in the snapshot propagation graph. Here, the parent node means the node from which the rumor comes. For example, when the rumor node forwards, quotes, or mentions the content of a previous rumor node, the previous rumor node can be considered as the parent node of the rumor node.

[0078] Then, based on the time decay parameter, the cosine function is used to obtain the rumor period parameter of the rumor node in the rumor snapshot. :

[0079]

[0080] in, is the first rumor cycle coefficient that can be learned, is the learnable second rumor cycle coefficient.

[0081] S4: Obtain a GIN encoder, fuse the rumor cycle parameter into the GIN encoder to obtain an edge weight GIN encoder, and perform multi-layer node embedding update on the rumor snapshot through the edge weight GIN encoder to obtain a snapshot dynamic propagation graph;

[0082] Furthermore, the purpose of this stage is to combine the rumor cycle coefficient with the GIN encoder (Graph Isomorphism Network) to obtain the edge weight GIN encoder, and use the GIN encoder to perform multi-layer node embedding updates on the rumor snapshot to obtain the snapshot propagation graph. Step S4 further includes:

[0083] S41: Obtain a GIN encoder, integrate the rumor cycle parameter into the GIN encoder, and obtain an edge weight GIN encoder. The expression of the edge weight GIN encoder is:

[0084]

[0085] in, is the m-th layer hidden feature vector of the v-th rumor node, represents the encoder parameters of the learnable m-th layer hidden feature vector, is the m-1th layer hidden feature vector of the vth rumor node, is the rumor cycle parameter of the vth rumor node, is the m-1th layer hidden feature vector of the uth neighboring node, is the set of neighboring nodes of the vth rumor node, ReLU() represents the activation function, and MLP() represents the multi-layer perceptron function;

[0086] S42: performing node embedding updates in multiple hidden layers of the rumor snapshot using multiple edge weight GIN encoders to obtain snapshot hidden layer outputs;

[0087] S43: Repeat the steps of calculating the time decay parameter of the rumor node until obtaining the snapshot hidden layer output, obtain multiple snapshot hidden layer outputs, perform mean pooling on all the snapshot hidden layer outputs, and obtain the snapshot dynamic propagation graph.

[0088] Regarding the above steps, the specific implementation methods in this embodiment are as follows:

[0089] First, we obtain the GIN encoder. The GIN encoder can effectively construct a graph representation, but it does not contain time information and cannot reflect the impact of time on the rumor propagation process. Therefore, we integrate the rumor cycle parameter into the GIN encoder to obtain the edge weight GIN encoder:

[0090]

[0091] in, is the m-th layer hidden feature vector of the v-th rumor node, represents the learnable encoder parameters of the m-th layer hidden feature vector, is the m-1th layer hidden feature vector of the vth rumor node, is the rumor cycle parameter of the vth rumor node, is the m-1th layer hidden feature vector of the uth neighboring node, is the set of neighboring nodes of the vth rumor node, ReLU() indicates activation using the activation function, and MLP() indicates the use of the multi-layer perceptron function. Here, the neighboring nodes are the rumor nodes adjacent to the vth rumor node in the rumor snapshot. Both the neighboring nodes and the rumor nodes are located in the rumor snapshot. In the edge weight GIN encoder, there are multiple layers, each of which can output the hidden feature vector of the layer. In the process of calculating the hidden feature vector of the first layer, and These are text features of rumor nodes and neighboring nodes directly obtained from the feature matrix of the snapshot propagation graph of the rumor snapshot.

[0092] Then, multiple edge-weighted GIN encoders are used to perform embedding updates in multiple hidden layers of the rumor snapshot. In this embodiment, the number of hidden layers is 2:

[0093]

[0094]

[0095]

[0096]

[0097] Among them, EAGIN() means using the edge weight GIN encoder for all rumor nodes in the rumor snapshot. represents the initial first-layer snapshot hidden layer output of the s-th rumor snapshot, Represents the initial input for embedding update of the s-th rumor snapshot. Here the initial input is , represents the first snapshot hidden layer output of the s-th rumor snapshot, β represents the learnable first hidden layer coefficient, () represents the first multi-layer perceptron function, () represents the second multi-layer perceptron function, represents the transformed first-layer snapshot hidden layer output of the s-1th rumor snapshot, γ represents the learnable second hidden layer coefficient, represents the initial second-layer snapshot hidden layer output of the s-th rumor snapshot, represents the second snapshot hidden layer output of the s-th rumor snapshot, Represents the transformed second-layer snapshot hidden layer output of the s-1th rumor snapshot. The parameters of the first multi-layer perceptron function and the second multi-layer perceptron function are different. The solution for transforming the first-layer snapshot hidden layer output and transforming the second-layer snapshot hidden layer output is as follows:

[0098]

[0099]

[0100] Among them, ET Layer() indicates the use of vector transformation layer for processing. represents the initial first-layer snapshot hidden layer output of the s-1th rumor snapshot, Represents the initial second-layer snapshot hidden layer output of the s-1th rumor snapshot. When the rumor snapshot is the first rumor snapshot, the values of the converted first-layer snapshot hidden layer output and the converted second-layer snapshot hidden layer output are 0. In this embodiment, the second-layer snapshot hidden layer output is used as the snapshot hidden layer output.

[0101] Finally, for all rumor snapshots, repeat the steps between calculating the time decay parameter of the rumor node and obtaining the snapshot hidden layer output, obtain multiple snapshot hidden layer outputs, perform mean pooling on the snapshot hidden layer output of the sth rumor snapshot, and obtain the sub-snapshot propagation graph of the sth rumor snapshot. :

[0102]

[0103] Among them, MEAN() means to perform mean pooling on the contents in the brackets, and perform mean pooling on all snapshot hidden layer outputs, and then obtain the snapshot dynamic propagation graph ,and .

[0104] S5: Establishing a bidirectional long short-term memory network, inputting the snapshot dynamic propagation graph into the bidirectional long short-term memory network, and obtaining an encoding graph representation of the rumor snapshot;

[0105] Furthermore, the purpose of this stage is to process the dynamic propagation graph through a bidirectional long short-term memory network to enhance the learning of the dynamic propagation graph on the temporal characteristics. Step S5 further includes:

[0106] S51: Acquire a forward long short-term memory network and a backward long short-term memory network, and establish the bidirectional long short-term memory network through the forward long short-term memory network and the backward long short-term memory network;

[0107] S52: Inputting the snapshot dynamic propagation graph into the forward long short-term memory network to obtain a snapshot forward coding graph representation, and inputting the snapshot dynamic propagation graph into the backward long short-term memory network to obtain a snapshot backward coding graph representation;

[0108] S53: Concatenate the forward coding graph representation of the snapshot and the backward coding graph representation of the snapshot to obtain the coding graph representation of the rumor snapshot.

[0109] Regarding the above steps, the specific implementation methods in this embodiment are as follows:

[0110] First, a forward long short-term memory network and a backward long short-term memory network are obtained, and then the forward long short-term memory network and the backward long short-term memory network are spliced to obtain a bidirectional long short-term memory network.

[0111] Then the snapshot dynamic propagation graph is input into the forward long short-term memory network to capture the forward hidden state of the snapshot dynamic propagation graph in time, and the snapshot forward encoding graph representation is obtained. :

[0112]

[0113] in, () indicates that the content in the brackets is processed through a forward long short-term memory network.

[0114] At the same time, the snapshot dynamic propagation graph is input into the backward long short-term memory network to capture the backward hidden state of the snapshot dynamic propagation graph in time, and the snapshot backward encoding graph representation is obtained. :

[0115]

[0116] in, Indicates that the content in the brackets is processed through the backward long short-term memory network.

[0117] Finally, the snapshot forward encoding graph representation and the snapshot backward encoding graph representation are spliced to obtain the encoding graph representation of the rumor snapshot .

[0118]

[0119] Among them, CONCAT() means concatenating the contents in the brackets.

[0120] S6: Obtain a deep neural network, input the encoded graph representation into the deep neural network, and complete rumor detection through the deep neural network.

[0121] Furthermore, the purpose of this stage is to use a deep neural network to determine whether a potential rumor event is a rumor, obtain a rumor snapshot judgment result, and thus complete rumor detection. Among them, step S6 further includes:

[0122] S61: Obtain a deep neural network, input the encoded graph representation into the fully connected layer and activation function layer of the deep neural network, and obtain a neural network output;

[0123] S62: Weight adjustment and biasing are performed on the neural network output to obtain a rumor snapshot judgment result, and rumor detection is completed through the rumor snapshot judgment result.

[0124] Regarding the above steps, the specific implementation methods in this embodiment are as follows:

[0125] First, a deep neural network is obtained, and then the coding graph representation is input into the fully connected layer and activation function layer of the deep neural network. The deep neural network can judge the probability of a potential rumor event being a rumor through the coding graph representation and obtain the neural network output .

[0126] Then adjust the weight and bias of the neural network output and activate it with the softmax function to get the rumor snapshot judgment result. :

[0127]

[0128] in, is the weight adjustment coefficient determined based on experience, The bias coefficient is determined empirically, and softmax() indicates the use of the softmax function to activate the content within the brackets. When the rumor snapshot judgment result is greater than the rumor threshold set as needed, the potential rumor event is determined to be a rumor, thus completing rumor detection.

[0129] The present invention also verifies the effectiveness of the rumor detection method based on dynamic graphs. Here, we first select BiGCN, Bi-GCN, GACL, RDEA, TrustRD and DynGCN for comparison, and measure the effectiveness of each method through the accuracy, precision, recall rate and F1 value. FGDGNN is the method proposed by the present invention. Table 1 is a comparison table of the rumor judgment experiment results of the present invention and other methods.

[0130] Table 1 Comparison of rumor judgment results between the present invention and other methods

[0131]

[0132] Among them, T indicates that the potential rumor event is not a rumor, F indicates that the potential rumor event is a rumor, RumorEval is a rumor dataset including rumors, and TWITTER is a software that uses Twitter as the target to obtain rumor data and perform detection. It can be seen from Table 1 that the method provided by the present invention has a good effect in detecting rumors.

[0133] Then, an ablation experiment was conducted on the method proposed in the present invention. The results of the ablation experiment are shown in Table 2:

[0134] Table 2 Schematic table of ablation experimental results of the method of the present invention

[0135]

[0136] From the results of the ablation experiment, it can be seen that each part of the method proposed in this invention has made a significant contribution to improving the effect of rumor detection.

[0137] Then, the edge weight GIN encoder used in the method proposed by the present invention was replaced with other encoders for experiments. The experimental results are shown in Table 3:

[0138] Table 3 Schematic table of experimental results of replacing the encoder in the present invention

[0139]

[0140] The experimental results show that the edge-weighted GIN encoder proposed in this experiment can achieve the best effect in the process of rumor detection.

[0141] The following describes a rumor detection device based on a dynamic graph provided by the present invention. The rumor detection device based on a dynamic graph described below and the rumor detection method based on a dynamic graph described above can refer to each other.

[0142] Figure 2 The structural diagram of the rumor detection system based on dynamic graph is shown as follows: Figure 2 As shown, the rumor detection method based on dynamic graph as described above includes:

[0143] Rumor propagation graph module 100: used to obtain rumor data, obtain rumor text sequences and rumor source nodes from the rumor data, and construct a rumor propagation graph through the rumor source nodes;

[0144] Snapshot propagation graph module 200: used to divide the rumor text sequence to obtain block rumor sequences, and establish a snapshot propagation graph of rumor snapshots through the block rumor sequences and the rumor propagation graph;

[0145] Rumor period parameter module 300: used to calculate the time decay parameter of the rumor node in the snapshot propagation graph, and calculate the rumor period parameter of the rumor node according to the time decay parameter;

[0146] Snapshot dynamic propagation graph module 400: used to obtain a GIN encoder, fuse the rumor cycle parameter into the GIN encoder to obtain an edge weight GIN encoder, and perform multi-layer node embedding update on the rumor snapshot through the edge weight GIN encoder to obtain a snapshot dynamic propagation graph;

[0147] The coding graph representation module 500 is used to establish a bidirectional long short-term memory network, input the snapshot dynamic propagation graph into the bidirectional long short-term memory network, and obtain a coding graph representation of the rumor snapshot;

[0148] Rumor detection module 600: used to obtain a deep neural network, input the coding graph representation into the deep neural network, and complete rumor detection through the deep neural network.

[0149] on the other hand, Figure 3 An example of a physical structure diagram of an electronic device is shown below. Figure 3As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other via the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute the rumor detection method based on the dynamic graph, which includes:

[0150] S1: Obtain rumor data, obtain rumor text sequences and rumor source nodes from the rumor data, and construct a rumor propagation graph through the rumor source nodes;

[0151] S2: Divide the rumor text sequence to obtain block rumor sequences, and establish a snapshot propagation graph of rumor snapshots using the block rumor sequences and the rumor propagation graph;

[0152] S3: Calculate the time decay parameter of the rumor node in the snapshot propagation graph, and calculate the rumor period parameter of the rumor node according to the time decay parameter;

[0153] S4: Obtain a GIN encoder, fuse the rumor cycle parameter into the GIN encoder to obtain an edge weight GIN encoder, and perform multi-layer node embedding update on the rumor snapshot through the edge weight GIN encoder to obtain a snapshot dynamic propagation graph;

[0154] S5: Establishing a bidirectional long short-term memory network, inputting the snapshot dynamic propagation graph into the bidirectional long short-term memory network, and obtaining an encoding graph representation of the rumor snapshot;

[0155] S6: Obtain a deep neural network, input the encoded graph representation into the deep neural network, and complete rumor detection through the deep neural network.

[0156] Furthermore, the logic instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0157] On the other hand, the present invention further provides a computer program product, comprising a computer program stored on a non-transitory computer-readable storage medium, wherein the computer program comprises program instructions. When the program instructions are executed by a computer, the computer is capable of performing the rumor detection method based on dynamic graphs provided by the above methods, the method comprising:

[0158] S1: Obtain rumor data, obtain rumor text sequences and rumor source nodes from the rumor data, and construct a rumor propagation graph through the rumor source nodes;

[0159] S2: Divide the rumor text sequence to obtain block rumor sequences, and establish a snapshot propagation graph of rumor snapshots using the block rumor sequences and the rumor propagation graph;

[0160] S3: Calculate the time decay parameter of the rumor node in the snapshot propagation graph, and calculate the rumor period parameter of the rumor node according to the time decay parameter;

[0161] S4: Obtain a GIN encoder, fuse the rumor cycle parameter into the GIN encoder to obtain an edge weight GIN encoder, and perform multi-layer node embedding update on the rumor snapshot through the edge weight GIN encoder to obtain a snapshot dynamic propagation graph;

[0162] S5: Establishing a bidirectional long short-term memory network, inputting the snapshot dynamic propagation graph into the bidirectional long short-term memory network, and obtaining an encoding graph representation of the rumor snapshot;

[0163] S6: Obtain a deep neural network, input the encoded graph representation into the deep neural network, and complete rumor detection through the deep neural network.

[0164] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the rumor detection method based on dynamic graphs provided by the above methods, the method comprising:

[0165] S1: Obtain rumor data, obtain rumor text sequences and rumor source nodes from the rumor data, and construct a rumor propagation graph through the rumor source nodes;

[0166] S2: Divide the rumor text sequence to obtain block rumor sequences, and establish a snapshot propagation graph of rumor snapshots using the block rumor sequences and the rumor propagation graph;

[0167] S3: Calculate the time decay parameter of the rumor node in the snapshot propagation graph, and calculate the rumor period parameter of the rumor node according to the time decay parameter;

[0168] S4: Obtain a GIN encoder, fuse the rumor cycle parameter into the GIN encoder to obtain an edge weight GIN encoder, and perform multi-layer node embedding update on the rumor snapshot through the edge weight GIN encoder to obtain a snapshot dynamic propagation graph;

[0169] S5: Establishing a bidirectional long short-term memory network, inputting the snapshot dynamic propagation graph into the bidirectional long short-term memory network, and obtaining an encoding graph representation of the rumor snapshot;

[0170] S6: Obtain a deep neural network, input the encoded graph representation into the deep neural network, and complete rumor detection through the deep neural network.

[0171] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0172] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0173] 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 aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. The rumor detection method based on dynamic graph is characterized by: include: S1: Obtain rumor data, obtain rumor text sequences and rumor source nodes from the rumor data, and construct a rumor propagation graph through the rumor source nodes; S2: Divide the rumor text sequence to obtain block rumor sequences, and establish a snapshot propagation graph of rumor snapshots using the block rumor sequences and the rumor propagation graph; S3: Calculate the time decay parameter of the rumor node in the snapshot propagation graph, and calculate the rumor period parameter of the rumor node according to the time decay parameter; S4: Obtain a GIN encoder, fuse the rumor cycle parameter into the GIN encoder to obtain an edge weight GIN encoder, and perform multi-layer node embedding update on the rumor snapshot through the edge weight GIN encoder to obtain a snapshot dynamic propagation graph; Step S4 further comprises: S41: Obtain a GIN encoder, integrate the rumor cycle parameter into the GIN encoder, and obtain an edge weight GIN encoder. The expression of the edge weight GIN encoder is: in, is the m-th layer hidden feature vector of the v-th rumor node, represents the learnable encoder parameters of the m-th layer hidden feature vector, is the m-1th layer hidden feature vector of the vth rumor node, is the rumor cycle parameter of the vth rumor node, is the m-1th layer hidden feature vector of the uth neighboring node, is the set of neighboring nodes of the vth rumor node, ReLU() represents the activation function, and MLP() represents the multi-layer perceptron function; S42: performing node embedding updates in multiple hidden layers of the rumor snapshot using multiple edge weight GIN encoders to obtain snapshot hidden layer outputs; S43: Repeat the steps from calculating the time decay parameter of the rumor node to obtaining the snapshot hidden layer output, obtaining multiple snapshot hidden layer outputs, and performing mean pooling on all the snapshot hidden layer outputs to obtain the snapshot dynamic propagation graph; S5: Establishing a bidirectional long short-term memory network, inputting the snapshot dynamic propagation graph into the bidirectional long short-term memory network, and obtaining an encoding graph representation of the rumor snapshot; S6: Obtain a deep neural network, input the encoded graph representation into the deep neural network, and complete rumor detection through the deep neural network.

2. The rumor detection method based on dynamic graph according to claim 1, characterized in that: Step S1 further comprises: S11: Determine target software, obtain rumor data from the target software, detect the rumor data to obtain potential rumor events, extract rumor texts related to the potential rumor events from the rumor data and sort them to obtain a rumor text sequence; S12: The rumor text that is most recent in the rumor text sequence is used as the rumor source node, and the rumor propagation graph is constructed according to the generation time and propagation order of the rumor text.

3. The rumor detection method based on dynamic graph according to claim 1, characterized in that: Step S2 further comprises: S21: extracting the timestamp of the rumor text sequence, dividing the rumor text sequence according to the timestamp to obtain a block rumor sequence including rumor text blocks; S22: Divide and textualize the rumor propagation graph through the rumor text blocks in the block rumor sequence to obtain the rumor snapshots, and obtain the snapshot propagation graph based on the rumor snapshots.

4. The rumor detection method based on dynamic graph according to claim 1, characterized in that: Step S5 further comprises: S51: Acquire a forward long short-term memory network and a backward long short-term memory network, and establish the bidirectional long short-term memory network through the forward long short-term memory network and the backward long short-term memory network; S52: Inputting the snapshot dynamic propagation graph into the forward long short-term memory network to obtain a snapshot forward coding graph representation, and inputting the snapshot dynamic propagation graph into the backward long short-term memory network to obtain a snapshot backward coding graph representation; S53: Concatenate the forward coding graph representation of the snapshot and the backward coding graph representation of the snapshot to obtain the coding graph representation of the rumor snapshot.

5. The rumor detection method based on dynamic graph according to claim 1, characterized in that: Step S6 further comprises: S61: Obtain a deep neural network, input the encoded graph representation into the fully connected layer and activation function layer of the deep neural network, and obtain a neural network output; S62: Weight adjustment and biasing are performed on the neural network output to obtain a rumor snapshot judgment result, and rumor detection is completed through the rumor snapshot judgment result.

6. A rumor detection system based on dynamic graphs, configured to execute the rumor detection method based on dynamic graphs according to any one of claims 1 to 5, characterized in that: include: Rumor propagation graph module: used to obtain rumor data, obtain rumor text sequences and rumor source nodes from the rumor data, and construct a rumor propagation graph through the rumor source nodes; Snapshot propagation graph module: used to divide the rumor text sequence to obtain block rumor sequences, and establish a snapshot propagation graph of rumor snapshots through the block rumor sequences and the rumor propagation graph; Rumor cycle parameter module: used to calculate the time decay parameter of the rumor node in the snapshot propagation graph, and calculate the rumor cycle parameter of the rumor node through the time decay parameter; Snapshot dynamic propagation graph module: used to obtain a GIN encoder, fuse the rumor cycle parameter into the GIN encoder to obtain an edge weight GIN encoder, and perform multi-layer node embedding update on the rumor snapshot through the edge weight GIN encoder to obtain a snapshot dynamic propagation graph; A coding graph representation module is used to establish a bidirectional long short-term memory network, input the snapshot dynamic propagation graph into the bidirectional long short-term memory network, and obtain a coding graph representation of the rumor snapshot; Rumor detection module: used to obtain a deep neural network, input the coding graph representation into the deep neural network, and complete rumor detection through the deep neural network.

7. An electronic 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 rumor detection method based on dynamic graphs as described in any one of claims 1 to 5 are implemented.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the rumor detection method based on dynamic graphs as described in any one of claims 1 to 5 are implemented.

9. A computer program product, comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, characterized in that: When the program instructions are executed by a computer, the computer can perform the steps of the rumor detection method based on dynamic graphs as described in any one of claims 1 to 5.

Images