Rumor tracing method based on deep infection aggregation

Abstract

The invention discloses a rumor tracing method based on infection depth aggregation, which proposes the concepts of infection degree, infection depth and infection aggregation degree, and comprises the following steps: firstly, based on an infection base map network before and after rumor propagation, obtaining a corresponding infection sub-map, and calculating the infection degree of each node in the infection sub-map; traversing nodes of the infection subgraph, and adding the nodes with the infection degree of 1 into a suspicious set of rumor source points; traversing a suspicious set of rumor source points, performing breadth-first search on nodes in the suspicious set, and calculating the infection depth of each node; finally, the aggregation degree of nodes in the suspicious set is calculated, a suspicious source point infection aggregation degree set is added, and in the suspicious source point infection aggregation degree set, the node with the maximum infection aggregation degree is selected as a speculated rumor source point. According to the method, rumor source points can be recognized within several steps, the traceability accuracy and the average error distance are integrated, and the method is excellent and stable in different network topologies.

Description

Technical Field

[0001] The present invention relates to the field of computer network and information security technology, and in particular to a rumor tracing method based on infection depth aggregation. Background Art

[0002] With the rapid development of the Internet, social media, online forums, instant messaging tools and other network platforms have become the main channels for information dissemination. However, these platforms also provide convenient conditions for the rapid spread of rumors. Rumor tracing in complex networks has always been a key topic of concern for scholars.

[0003] The current rumor tracing problem mainly relies on two main methods: one is to construct source estimates based on the infection centrality index through the infection subgraph to maximize the probability of rumor source detection. However, this method has a high demand for a complete propagation subgraph, and it is often difficult to obtain complete propagation paths and infected node information in actual scenarios. The second is to detect the rumor source by setting up observers to collect node infection status and infection time information. The tracing accuracy is limited by the location and number of observers. If the observers are not set properly or the number is insufficient, key information may be missed, thus affecting the accuracy of the tracing results.

[0004] As the scale of the network expands, data acquisition becomes more difficult and computational complexity increases dramatically, which seriously affects the efficiency of rumor tracing. Therefore, how to accurately and efficiently track the source of rumors between networks with different topologies has become an important issue that needs to be solved urgently. Summary of the invention

[0005] The problem to be solved by the present invention is to provide a rumor tracing method based on infection depth aggregation, propose the concepts of node infection degree, infection depth and infection aggregation, quickly identify the source of rumors by selecting suspicious sets, and improve the comprehensive tracing accuracy and average error distance in different network topologies.

[0006] The present invention adopts the following technical solution: a rumor tracing method based on infection depth aggregation, comprising the following steps:

[0007] S1. Based on the infection base graph network before and after the rumor spread, obtain the corresponding infection subgraph and calculate the infection degree of each node in the infection subgraph;

[0008] S2, traverse the nodes of the infected subgraph and add the nodes with an infection degree of 1 to the suspicious set of rumor sources;

[0009] S3, traverse the suspicious set of rumor sources, perform breadth-first search on the nodes in the suspicious set, and calculate the infection depth of each node;

[0010] S4, calculate the aggregation degree of the suspicious concentrated nodes and add them to the suspected source point infection aggregation degree set;

[0011] S5. In the infection concentration of suspected source points, select the node with the largest infection concentration as the presumed rumor source point.

[0012] Specifically, suppose the infection base graph before and after the rumor spread is G(V,E), and the infection subgraph is G i (V i ,E i ), any node Calculate the infection degree Inr(v) of node v, the formula is as follows:

[0013]

[0014] Among them, V, V i They represent the set of nodes in the network before and after infection, E and E i Respectively represent the set of edges in the network before and after infection, k v They represent the degree of node v in the infected subgraph and infected base graph respectively; if Inr(v)=1, add node v to the suspicious set S of the rumor source.

[0015] Furthermore, we traverse the nodes in the suspicious set S, and for any node v∈S, we perform a breadth-first search BFS(G,v) to find the first node u that satisfies Inr(u)<1, and calculate the infection depth starting from node v. The formula is as follows:

[0016] dep i (v)=d(v,u)-1

[0017] Among them, d(v,u) is the shortest path length from node v to node u.

[0018] Furthermore, for any node v∈S in the suspicious set, the infection concentration Cr(v) of node v is calculated as follows:

[0019]

[0020] in, Represents the number of infected nodes with a step length of l from node v; the step length l is set according to the shortest path length from node v.

[0021] Calculate the infection concentration of each node, add the infection concentration set C of the suspected source points, and select the node with the largest infection concentration in C As the source of the speculated rumor.

[0022] The technical solution of the present invention also provides: an electronic device, comprising:

[0023] one or more processors;

[0024] a storage device having one or more programs stored thereon;

[0025] When the one or more programs are executed by the one or more processors, the one or more processors implement any of the above-mentioned rumor tracing methods based on infection depth aggregation.

[0026] The technical solution of the present invention also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in any of the above-mentioned rumor tracing methods based on infection depth aggregation are implemented.

[0027] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0028] 1. The rumor tracing method based on infection depth aggregation is proposed in the present invention. The concepts of infection degree, infection depth and infection aggregation are proposed. By selecting a suspicious set and calculating the infection aggregation of the nodes in the suspicious set, the node with the largest infection aggregation is selected as the rumor source, so as to quickly and effectively locate the rumor source.

[0029] 2. The rumor tracing method of the present invention can identify the source of the rumor within a few steps, and has excellent and stable comprehensive tracing accuracy and average error distance in different network topologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a flow chart of the rumor tracing method of the present invention;

[0031] Figure 2 Schematic diagram of error distances of different rumor tracing methods in four different networks according to embodiments of the present invention;

[0032] Figure 3 A schematic diagram of average error distances of different rumor tracing methods in different network topologies according to an embodiment of the present invention;

[0033] Figure 4 It is a schematic diagram of the evolution process of the average error distance of the rumor tracing method of the present invention with the infection scale under different network topology structures;

[0034] Figure 5 It is a schematic diagram of the evolution process of the average error distance of the rumor tracing method of the present invention as the average degree of the network changes;

[0035] Figure 6 Schematic diagram of the changing trend of the average error distance under different infection rates in different network topologies. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the application is further elaborated in detail below in conjunction with the accompanying drawings. The described embodiments are only a part of the embodiments involved in the present invention. All non-innovative embodiments of other researchers in the field on this embodiment belong to the protection scope of the present invention. At the same time, for the step numbering in the embodiment of the present invention, it is only set for the convenience of explanation, and the order between the steps is not limited in any way. The execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0037] The present invention provides a rumor tracing method based on infection depth aggregation, such as Figure 1 As shown, the following steps are included:

[0038] S1. Based on the infection base graph network before and after the rumor spread, obtain the corresponding infection subgraph and calculate the infection degree of each node in the infection subgraph;

[0039] S2, traverse the nodes of the infected subgraph and add the nodes with an infection degree of 1 to the suspicious set of rumor sources;

[0040] S3, traverse the suspicious set of rumor sources, perform breadth-first search on the nodes in the suspicious set, and calculate the infection depth of each node;

[0041] S4, calculate the aggregation degree of the suspicious concentrated nodes and add them to the suspected source point infection aggregation degree set;

[0042] S5. In the infection concentration of suspected source points, select the node with the largest infection concentration as the presumed rumor source point.

[0043] In one embodiment of the present invention, experiments are conducted based on four different network graphs: WS (Watts-Strogatz network), BA (Barabasi-Albert network), Facebook and LastFM networks, where the Facebook and LastFM networks are from the complex network dataset of Stanford University ( http: / / snap.stanford.edu / data / ). The infection rate is set to 0.8. Based on the SI propagation model, the corresponding infected subgraph is obtained. The parameter settings of each base graph network are shown in Table 1 below:

[0044]

[0045] Furthermore, this embodiment uses the method of the present invention to trace the source of rumors based on the parameter settings of each base map network in the above table.

[0046] First, assume that the base network before and after the rumor spread and the infected subgraph obtained are G(V,E), G i (V i ,E i);

[0047] Traverse the infected subgraph G i For any node v∈G i , record the degree of node v in the infected subgraph and base graph network k v , calculate the infection degree of node v If Inr(v) = 1, that is, node v is fully infected, add v to the suspicious set S;

[0048] Then, traverse the nodes in S, and for the node v∈S, perform a breadth-first search BFS(G,v) to find the first node u that satisfies Inr(u)<1.

[0049] Let the shortest path length from node v to node u be d(v,u), then the infection depth starting from node v is:

[0050] dep i (v)=d(v,u)-1

[0051] Next, traverse the suspicious set S, and for the node v∈S, use It represents the number of infected nodes with a step length of l from node v. The infection concentration of node v is:

[0052]

[0053] Add the suspected source infection clustering set C.

[0054] Finally, select the node with the largest infection concentration in C This is the source of the rumor inferred by the method of this embodiment.

[0055] Furthermore, the error distances of the rumor tracing method based on infection depth aggregation (IDC) in this embodiment and the existing tracing algorithm based on closeness centrality (CC), the tracing algorithm based on betweenness centrality (BC) and the back propagation algorithm (RP) in different basemap networks are compared. Figure 2 as shown.

[0056] Figure 2 (a) is a schematic diagram of the error distance in the artificial synthetic network WS. Figure 2 (b) is a schematic diagram of the error distance in the artificial synthetic network BA. Figure 2 (c) is a schematic diagram of the error distance in the real network Facebook. Figure 2 (d) in the figure is a schematic diagram of the error distance in the real network LastFM.

[0057] It can be seen that the tracing algorithm based on infection depth clustering (IDC) performs well in tracing under four different network structure topologies, and can find the real source within a limited number of steps. Among them, the performance is best on the small-world network WS, with a tracing accuracy of more than 25%, and a probability of nearly 80% to find the rumor source within one step, and a probability of about 98% to find the rumor source within two steps. At the same time, the IDC algorithm has a higher tracing accuracy on the BA network and LastFM network than the other three algorithms, and is slightly inferior to the CC algorithm on the WS network, but far exceeds the RP and BC algorithms. The tracing effect on the Facebook network is weaker than the other three algorithms, but it can be guaranteed to find the rumor source within 3.

[0058] Comparison of average error distances of different algorithms in different network topologies, such as Figure 3 shown.

[0059] It can be seen that in the BA network, the tracing algorithm based on infection depth aggregation (IDC) has the smallest average error distance, which is better than the other three tracing algorithms. In the WS network, the average error distance of the IDC algorithm is slightly greater than 1 step, second only to the CC algorithm; except for the RP algorithm, the average error distances of the other three are all around 1 step, and the tracing performance is better. In the LastFM network, the average error distances of several tracing algorithms are very close, all around 2 steps. In the Facebook network, the average error distance of the IDC algorithm is higher than that of the other three algorithms, and its performance is relatively weak. However, regardless of the topology, the average error distance of the IDC algorithm fluctuates within 2 steps.

[0060] Furthermore, in this embodiment, the tracing method based on infection depth aggregation is used to show the evolution of the average error distance with the change of infection scale under different network topologies. Figure 4 As shown in the figure, the evolution of the average error distance as the average degree of the network changes is shown in the figure. Figure 5 As shown in the figure; the changing trend of the average error distance under different infection rates in different network topologies is shown in the figure Figure 6 shown.

[0061] It can be seen that the method of this embodiment can identify the source of rumors within a few steps, and the comprehensive tracing accuracy and average error distance are excellent and stable in different network topologies.

[0062] In an embodiment of the present invention, an electronic device is also provided, including: one or more processors; a storage device on which one or more programs are stored; when the one or more programs are executed by the one or more processors, the one or more processors implement the tracing method based on infection depth aggregation described in any of the above embodiments.

[0063] In an embodiment of the present invention, a computer-readable storage medium is further provided, on which a computer program is stored. When the program is executed by a processor, the steps of any one of the tracing methods based on infection depth aggregation in the above embodiments are implemented.

[0064] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A rumor tracing method based on infection depth aggregation, characterized in that: The steps include: S1. Based on the infection base graph network before and after the rumor spread, obtain the corresponding infection subgraph and calculate the infection degree of each node in the infection subgraph; S2, traverse the nodes of the infected subgraph and add the nodes with an infection degree of 1 to the suspicious set of rumor sources; S3, traverse the suspicious set of rumor sources, perform breadth-first search on the nodes in the suspicious set, and calculate the infection depth of each node; S4, calculate the aggregation degree of the suspicious concentrated nodes and add them to the suspected source point infection aggregation degree set; S5. In the infection concentration of suspected source points, select the node with the largest infection concentration as the presumed rumor source point.

2. The rumor tracing method based on infection depth aggregation according to claim 1 is characterized in that: Assume that the infection base graph before and after the rumor spread is G(V,E), and the infection subgraph is G i (V i ,E i ), any node Calculate the infection degree Inr(v) of node v, the formula is as follows: Among them, V, V i They represent the set of nodes in the network before and after infection, E and E i Respectively represent the set of edges in the network before and after infection, k v They represent the degree of node v in the infected subgraph and infected base graph respectively; if Inr(v)=1, add node v to the suspicious set S of the rumor source.

3. The rumor tracing method based on infection depth aggregation according to claim 2 is characterized in that: In step S3, the nodes in the suspicious set S are traversed, and for any node v∈S, a breadth-first search BFS(G,v) is performed to find the first node u that satisfies Inr(u)<1, and the infection depth starting from node v is calculated. The formula is as follows: dep i (v)=d(v,u)-1 Among them, d(v,u) is the shortest path length from node v to node u.

4. The rumor tracing method based on infection depth aggregation according to claim 3 is characterized in that: In step S4, for any node v∈S in the suspicious set, the infection concentration Cr(v) of node v is calculated, and the formula is as follows: Among them, g i l (v) represents the number of infected nodes with a step length of l from node v; the step length l is set according to the shortest path length from node v. The infection concentration of each node is calculated, and the infection concentration set C of the suspected source point is added. The node with the largest infection concentration in C is selected. As the source of the speculated rumor.

5. An electronic device, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors implement the rumor tracing method based on infection depth aggregation as described in any one of claims 1 to 4.

6. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the program is executed by a processor, the steps in the rumor tracing method based on infection depth aggregation described in any one of claims 1 to 4 are implemented.

Images