Method and system for identifying dominant path in multi-source scene based on perturbation response model

Abstract

The invention discloses a dominant path identification method and system in a multi-source scene based on a perturbation response model, and relates to the technical field of computers. According to the method, the influence of the link state change on the steady state of the system is dynamically analyzed based on a perturbation response mode instead of depending on a static index, the method is closer to a real interaction mechanism of the network, and the mechanism modeling precision is improved. By traversing the link state, the actual contribution of each link to the response of the target node is quantified, and the complex interaction of weak connection and multi-link coordination is prevented from being ignored. Through the process that the system is evolved to the steady state again after disturbance, flow distribution and competition at the position of a shared node are naturally modeled instead of depending on local optimal addressing, the global reasonability of dominant path recognition is improved, and the problem of multi-source target resource competition is solved. The method is suitable for various networks, and has a core support effect on construction of a propagation risk early warning model based on network science, implementation of a key node intervention strategy and optimization of an intelligent governance mechanism of a social information system.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and particularly to a method and system for identifying dominant paths in a multi-source scenario based on a perturbation response model. Background Art

[0002] A multi-source - target scenario refers to an application scenario that includes multiple pairs of source nodes and target nodes. Identifying dominant paths in a multi-source - target scenario means: in a scenario where multiple source nodes and multiple target nodes conduct information interaction, filtering out the paths that mainly carry information flows during the directional propagation of network information flows. It is mainly applied to fields such as public health early warning and network structure optimization in social networks.

[0003] In the task of identifying dominant paths, the general idea of existing methods is to calculate certain metrics to obtain the importance of links, and then use a local optimal addressing method to find target nodes. However, these methods have defects such as insufficient mechanism modeling, neglecting weak connections in dominant paths, and insufficient modeling of complex interactions. Moreover, there are also modeling difficulties in the task of identifying dominant paths in a multi-source - target scenario, such as resource competition easily occurring at shared nodes.

[0004] Therefore, the existing technology still needs to be improved and developed. Summary of the Invention

[0005] The technical problem to be solved by the present invention is, in view of the above-mentioned defects of the existing technology, to provide a method and system for identifying dominant paths in a multi-source scenario based on a perturbation response model, aiming to solve the problems of insufficient mechanism modeling in the existing dominant path identification method and resource competition easily occurring at shared nodes.

[0006] The technical solution adopted by the present invention to solve the problem is as follows: In a first aspect, an embodiment of the present invention provides a method for identifying dominant paths in a multi-source scenario based on a perturbation response model, the method including: Determine the node indices of multiple pairs of source nodes and target nodes in the target network, and input the graph structure data of the target network and the node indices into the perturbation response model; Determine the dominant paths in a multi-node interaction scenario through the perturbation response model; The perturbation response model is used for: for each link in the target network, under two link states where the propagation effect of the link is normal and stopped respectively, perturb each source node in sequence, and make the target network re-evolve from the initial steady state to a new steady state, obtaining two new steady states; determining the traffic carried by the link in the information propagation process of different source nodes according to the response amplitudes of each target node in the initial steady state and the two new steady states, obtaining the link traffic carrying distribution data of each source node; and determining the dominant path according to the link traffic carrying distribution data corresponding to each source node.

[0007] In one implementation, under two link states where the propagation effect of the link is normal and stopped respectively, perturb each source node in sequence, and make the target network re-evolve from the initial steady state to a new steady state, obtaining two new steady states, including: In each link state, apply a perturbation with a specific amplitude to one source node of the target network in the initial steady state to make the target network enter a non-steady state; By solving the ordinary differential equation corresponding to the information flow dynamics equation, re-evolve the target network in the non-steady state to a new steady state; the information flow dynamics equation is used to describe the node interaction behavior in different scenarios.

[0008] In one implementation, by solving the ordinary differential equation corresponding to the information flow dynamics equation, including: By using the Runge-Kutta algorithm, solve the ordinary differential equation corresponding to the information flow dynamics equation.

[0009] In one implementation, according to the response amplitudes of each target node in the initial steady state and the two new steady states, determining the traffic carried by the link in the information propagation process of different source nodes, obtaining the link traffic carrying distribution data of each source node, including: Obtain the response amplitudes of each target node in each new steady state through the response matrix; the response matrix is used to describe: the response amplitudes between different node pairs after the target network is perturbed in the steady state, and the response amplitudes between different node pairs after the propagation effect of each link stops; Compare the response amplitudes of each target node in the new steady state and the two new steady states, and calculate the difference in the response amplitudes of each target node; According to the difference in the response amplitudes of each target node obtained after perturbing each source node, determine the traffic carried by each link in the information propagation process of different source nodes, obtaining the link traffic carrying distribution data of each source node.

[0010] In one embodiment, the response matrix is an N*N matrix, where N represents the number of rows or columns; each element of the response matrix represents the response relationship of a node pair; each row corresponding to the source node in the response matrix represents the difference in response amplitude between the other nodes when reaching the new steady state and when in the initial steady state after a perturbation is applied from the source node.

[0011] In one embodiment, determining the dominant path according to the link traffic carrying distribution data respectively corresponding to each source node includes: Determining the traffic carrying capacity of each link in the target network according to the link traffic carrying distribution data respectively corresponding to each source node; Determining the dominant path according to the traffic carrying capacity of each link.

[0012] In one embodiment, determining the dominant path according to the traffic carrying capacity of each link includes: Sorting the traffic carrying capacity of each link to obtain a target sequence; Performing a binary search operation according to the target sequence, and determining a target threshold through the result of the binary search operation; Screening out a number of target links according to the target threshold, and determining the dominant path according to each target link; the dominant path can connect all source nodes and target node pairs.

[0013] In a second aspect, an embodiment of the present invention further provides a dominant path identification system in a multi-source scenario based on a perturbation response model. The system includes: A data input module, configured to determine the node indexes of multiple pairs of source nodes and target nodes in the target network, and input the graph structure data of the target network and the node indexes into the perturbation response model; The perturbation response model is configured to, for each link in the target network, sequentially apply a perturbation to each source node in two link states where the propagation effect of the link is normal and stopped, and make the target network re-evolve from the initial steady state to a new steady state to obtain two new steady states; determining the traffic carried by the link in the information propagation process of different source nodes according to the response amplitudes of each target node in the initial steady state and the two new steady states, to obtain the link traffic carrying distribution data of each source node; determining the dominant path according to the link traffic carrying distribution data respectively corresponding to each source node.

[0014] In a third aspect, an embodiment of the present invention further provides a terminal, which includes a memory and one or more processors; the memory stores one or more programs; the programs include instructions for executing the method for identifying the dominant path in a multi-source scenario based on the perturbation response model as described in any one of the above; the processor is used to execute the programs.

[0015] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which multiple instructions are stored, and the instructions are suitable for being loaded and executed by a processor to implement the steps of the method for identifying the dominant path in a multi-source scenario based on the perturbation response model as described in any one of the above.

[0016] Advantages of the present invention: The embodiment of the present invention dynamically analyzes the impact of link state changes on the system steady state based on the perturbation response method, rather than relying on static indicators, which is closer to the real network interaction mechanism and makes the mechanism modeling more accurate. By traversing the link states and quantifying the actual contributions of each link to the response of the target node, the complex interaction of weak connections and multi-link cooperation is avoided. Through the process of the system re-evolving to the steady state after perturbation, the traffic distribution and competition at the shared nodes are naturally modeled, rather than relying on local optimal addressing, which improves the global rationality of dominant path identification and solves the problem of multi-source target resource competition. The method of the present invention is applicable to various networks and has a core supporting role in constructing a propagation risk early warning model based on network science, implementing key node intervention strategies, and optimizing the intelligent governance mechanism of social information systems. Description of the Drawings

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0018] Figure 1 It is a flowchart of the method for identifying the dominant path in a multi-source scenario based on the perturbation response model provided by the embodiment of the present invention.

[0019] Figure 2 It is a module diagram of the system for identifying the dominant path in a multi-source scenario based on the perturbation response model provided by the embodiment of the present invention.

[0020] Figure 3 It is a schematic block diagram of the terminal provided by the embodiment of the present invention. Detailed Embodiments

[0021] The present invention discloses a method and system for identifying dominant paths in a multi-source scenario based on a perturbation response model. To make the objectives, technical solutions, and effects of the present invention clearer and more definite, the following further elaborates on the present invention with reference to the accompanying drawings and by way of examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0022] Those skilled in the art of this technology can understand that unless specifically stated otherwise, the singular forms "a", "an", "the", and "said" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the specification of the present invention means the presence of the described features, integers, steps, operations, but does not exclude the presence or addition of one or more other features, integers, steps, operations. It should be understood that the phrase "and / or" used herein includes all or any unit and all combinations of one or more of the associated listed items.

[0023] Those skilled in the art of this technology can understand that unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as the general understanding of those of ordinary skill in the art to which the present invention belongs. It should also be understood that terms such as those defined in a general dictionary should be understood to have a meaning consistent with the meaning in the context of the prior art, and will not be interpreted with an idealized or overly formal meaning unless specifically defined as herein.

[0024] In view of the above defects of the prior art, the present invention provides a method for identifying dominant paths in a multi-source scenario based on a perturbation response model. The method includes: determining the node indices of multiple pairs of source nodes and target nodes in a target network, and inputting the graph structure data of the target network and the node indices into the perturbation response model; determining the dominant paths in a multi-node interaction scenario through the perturbation response model. The perturbation response model is used to: for each link in the target network, respectively apply perturbations to each source node in turn under two link states where the propagation effect of the link is normal and stopped, and evolve the target network from the initial steady state to a new steady state again, obtaining two new steady states; according to the response amplitudes of each target node in the initial steady state and the two new steady states, determine the traffic carried by this link in the information propagation process of different source nodes, obtaining the link traffic carrying distribution data of each source node; determine the dominant paths according to the link traffic carrying distribution data corresponding to each source node respectively. The present invention dynamically analyzes the influence of link state changes on the system steady state based on the perturbation response method, rather than relying on static indicators, which is closer to the real interaction mechanism of the network, making the mechanism modeling more accurate. And by traversing the link states, the actual contribution of each link to the response of the target node is quantified, avoiding ignoring the complex interaction of weak connections and multi-link cooperation. In addition, through the process of the system evolving to the steady state again after perturbation, the traffic distribution and competition at the shared nodes are naturally modeled, rather than relying on local optimal addressing, improving the global rationality of dominant path identification and solving the problem of multi-source target resource competition. The method of the present invention can be applied to various networks in reality, especially providing a new solution idea for the process of finding dominant paths in a multi-source - target scenario. It has a core supporting role in constructing a propagation risk warning model based on network science, implementing key node intervention strategies, and optimizing the intelligent governance mechanism of social information systems.

[0025] As Figure 1 shown, the method specifically includes the following steps: Step S100: Determine the node indices of multiple pairs of source nodes and target nodes in the target network, and input the graph structure data of the target network and the node indices into the perturbation response model.

[0026] Specifically, the target network in this embodiment can be any network that needs to identify dominant paths. Obtain the graph structure data of the target network, as well as the node indices of the source nodes and target nodes that need to perform information interaction, and input the graph structure data and node indices into the pre-established perturbation response model.

[0027] Step S200: Determine the dominant paths in a multi-node interaction scenario through the perturbation response model; The perturbation response model is used for: for each link in the target network, perturbing each source node in turn under two link states where the propagation effect of the link is normal and stopped respectively, and evolving the target network from the initial steady state to a new steady state to obtain two new steady states; determining the traffic carried by the link in the information propagation process of different source nodes according to the response amplitudes of each target node in the initial steady state and the two new steady states, so as to obtain the link traffic carrying distribution data of each source node; and determining the dominant path according to the link traffic carrying distribution data corresponding to each source node respectively.

[0028] Generally speaking, this embodiment redefines the penetration mode of information flow and perturbs the target network that was originally in a steady state through the perturbation response model. The perturbation will break the original steady-state balance, and after re-evolving to the steady state, the dominant path in the information flow propagation process is determined. Perturbation analysis can track the information propagation between interacting particles. The stability of a nonlinear system is evaluated by observing its response to local perturbations, which can gradually converge to the original stable point, diverge into instability, or exhibit cyclic or chaotic behavior. During the process of perturbation and steady-state evolution, for each link, perturbations are applied to a certain source node under two states of normal propagation and stopped propagation respectively, so that the network evolves from the initial steady state to a new steady state, and the steady-state responses under the two states are obtained. By comparing the difference in the response amplitudes of the target node under the two steady states, the traffic carrying situation of each link corresponding to the source node is determined, and the link traffic carrying distribution data is obtained. The link traffic carrying distribution data can reflect the actual contribution of each link to the information transmission between the source node and the corresponding target node. Finally, the link traffic carrying distribution data of all source nodes are integrated, and by analyzing the traffic distribution under multi-source interaction, the dominant path carrying the key traffic is screened out. This embodiment uses a dynamic analysis method combining link state control and perturbation response observation to quantify the actual role of each link in multi-source transmission, avoiding relying on static indicators or local optimal assumptions, so as to accurately capture the dominant path of multi-node interaction in a complex network.

[0029] For example, there are multiple nodes in the target network, and the link between two nodes is a link or an edge. For a network in a steady state, multiple pairs of source nodes and target nodes in the information interaction process are determined. Each time starting from the initial steady state, a slight perturbation is added to each source node in turn. The information propagation activities of each link are stopped in turn, that is, the information propagation activity of one link is stopped each time, and then it evolves to a new steady state again. For each node, the amplitude of the node response perturbation is obtained by taking the difference between the response amplitudes of the node in each new steady state and the initial steady state. By comparing the response amplitude differences of all target nodes before and after the link stops the information propagation activity, the traffic distribution carried by all links for the scenario of transmitting from a specific source node to all target nodes is obtained, that is, the link traffic carrying distribution data of the specific source node is obtained. The link traffic carrying distribution data can be in vector form, that is, a single source node can correspond to a vector describing the traffic carried by the link. Finally, the overall link traffic carrying distribution data in the multi-source - target scenario is obtained. Subsequently, by setting a reasonable threshold, the dominant paths can be further screened out.

[0030] In one implementation, under two link states where the propagation effect of the link is normal and stopped respectively, a perturbation is applied to each of the source nodes in turn, and the target network is re-evolved from the initial steady state to a new steady state, obtaining two new steady states, including: In each link state, a perturbation with a specific amplitude is applied to one of the source nodes of the target network in the initial steady state to make the target network enter an unsteady state; By solving the ordinary differential equation corresponding to the information flow dynamics equation, the target network in the unsteady state is re-evolved to a new steady state; the information flow dynamics equation is used to describe the node interaction behavior in different scenarios.

[0031] Generally speaking, the perturbation response model includes an information flow dynamics equation, which is used to describe the node interaction behavior in different scenarios and can adapt to the scenario diversity in the information propagation process of multiple pairs of source nodes and target nodes. Among them, the information flow dynamics equation consists of two data items. One data item considers how the node evolves after receiving information itself, and the other data item describes how the node is affected by the information source itself. Different information reception behaviors can be described by selecting appropriate non-linear terms.

[0032] Specifically, when the target network is in the initial steady state, a perturbation with a specific amplitude (such as injecting a specific amplitude of traffic) is applied to a certain source node to break the system balance and make it enter the non-steady state. Through the information flow dynamics equation, the dynamic interaction rules between nodes in the network are characterized. Solve this system of equations to simulate the gradual evolution of the network from the initial state after perturbation to the new steady state under the non-steady state. For each link, repeat the above process respectively in two states: normal propagation of the link and stop propagation of the link. By comparing the response differences of the target node in the new steady states of the two states, the actual contribution (such as traffic carrying capacity) of this link to the information transmission between the source node and the target node can be quantified.

[0033] In other words, by quantifying the carrying role of the path in the information propagation process, the dominant path can be determined. However, when the source node and the target node are determined, the complexity of calculating the carrying role of all paths is too high. Therefore, in this embodiment, quantifying the carrying role of the path is transformed into quantifying the traffic carrying role of each link, so that only all links in the target network need to be traversed. Specifically, the traffic carrying role of the link can be quantified by stopping the transmission activity of the link in the information propagation process. Its essence is to transform the implicit role of the link (such as traffic carrying capacity) into a measurable steady-state response difference through perturbation, evolution, and observation.

[0034] It should be noted that stopping the transmission activity of the link only performs functional restrictions on a certain edge or a certain link in the target network, rather than deleting the link from the target network, nor restricting the activities of the two nodes connected by the link. The two nodes can still interact with other nodes through other links.

[0035] For example, consider a node and link network . Define independent information sources, that is, the source node ; independent receiving targets, that is, the target node .

[0036] The information flow and its penetration behavior are represented by the following dynamic equation, which is an ordinary differential equation: ; Among them, , respectively represent the opinion value or information level of node , at a certain moment; represents the value changed by node within the change step time; Denotes the change step size, indicating a very short time. The first data item , which is used to reflect how the node itself evolves after receiving information; the second data item , which is used to reflect how the node itself is affected by the information source, and different information reception behaviors can be described by selecting appropriate non-linear terms; Denotes when and carry out the non-linear mapping behavior of information interaction; Denotes and When carrying out information interaction, it actively docks the non-linear mapping behavior of the received information interaction. This behavior can be regarded as mapping its own information and the information of node to a couplable space; Denotes the rate of information transfer from node to node , which can be unified as the elements of the adjacency matrix or other weights can be used instead.

[0037] In actual application scenarios, according to different dynamic equations, , and can be changed, and then the node interaction modes in different scenarios can be simulated, the information flow propagation modes in different scenarios can be found, and thus the dominant paths in different modes can be found through evolution.

[0038] According to the dynamic equation, when the system reaches a steady state, the derivative is 0, as shown in the following formula: ; The steady-state solution form of the dynamic equation is: ; Among them, is the result obtained by taking the inverse function of , and denotes the intermediate variable; Denotes the in-degree of node . The steady state of the node is jointly determined by its weighted degree and the dynamic equation.

[0039] Subsequent experimental results are obtained based on the following information flow dynamic equation, which is similar to the aforementioned dynamic equation, but , and in it are set according to specific application scenarios: ; Among them, denotes at time The opinion value or information level; denote the node at a moment the square of the opinion value or information level of the node.

[0040] In one implementation, by solving the ordinary differential equation corresponding to the information flow dynamics equation, including: By using the Runge-Kutta algorithm, solve the ordinary differential equation corresponding to the information flow dynamics equation.

[0041] Specifically, in the steady state, the calculation result of the ordinary differential equation is 0. When a perturbation is added to the source node, the steady state of the entire system will be broken, and the ordinary differential equation is no longer 0. The Runge-Kutta algorithm is a numerical method for solving ordinary differential equations, which improves the accuracy of the approximate solution through multi-stage slope estimation. In this embodiment, the Runge-Kutta algorithm is selected to re-evolve the unstable system to the steady state.

[0042] In one implementation, according to the initial steady state and the response amplitudes of each target node in the two new steady states, determine the traffic carried by the link during the information propagation of different source nodes, and obtain the link traffic carrying distribution data of each source node, including: Through the response matrix, obtain the response amplitude of each target node in each new steady state; the response matrix is used to describe: the response amplitude between different node pairs after the target network is perturbed in the steady state, and the response amplitude between different node pairs after each link stops propagating. Compare the new steady state and the response amplitudes of each target node in the two new steady states, and calculate the response amplitude difference of each target node; According to the response amplitude difference of each target node obtained after each source node applies a perturbation, determine the traffic carried by each link during the information propagation of different source nodes, and obtain the link traffic carrying distribution data of each source node.

[0043] Generally speaking, the perturbation response model also includes a response matrix. In this embodiment, the response matrix is introduced to describe the response amplitude of nodes to perturbations. The response matrix can not only be used to describe the response amplitude between different node pairs in a steady-state network under perturbations, but also be used to describe the response amplitude between different node pairs after a certain link stops propagating.

[0044] Specifically, perturbations are applied to the source nodes in sequence. After evolving to a steady state using the Runge-Kutta method, the opinion values of each node reach a new steady state. At this time, the response matrix is calculated. The size of the response matrix is: the number of nodes * the number of nodes; the elements in each row of the response matrix are used to reflect: when the node pointed to by the row index is added as a source node for perturbation, the response amplitude or change amplitude of other nodes. In other words, the response matrix is used to store the response amplitudes, and the elements in the response matrix are essentially the quantification of the influence of perturbation propagation between nodes, reflecting the direct or indirect influence on the information transmission between the source node and the target node in the normal transmission state or the stopped transmission state of the link.

[0045] In an actual application scenario, taking a link as an example, perturbations are applied to a certain source node in two link states where the propagation effect of the link is normal and stopped respectively, so that the target network evolves from the initial steady state to a new steady state again. For each target node, the response amplitudes in the two new steady states are obtained through the response matrix, and the response amplitude in the initial steady state is obtained, and the following formula is calculated: (the response amplitude in the first new steady state - the response amplitude in the initial steady state) - (the response amplitude in the second new steady state - the response amplitude in the initial steady state) = the response amplitude difference of the target node finally. Add up the response amplitude differences of the corresponding target node indices to obtain the traffic carried by this link in the information propagation process of this source node; or for this source node, the traffic carrying capacity propagated to all target nodes through this link. In other words, the traffic carrying capacity of each link is related to the total traffic actually carried by this link in the transmission from the source node to all target nodes.

[0046] By determining the traffic carried by each link in the information propagation process of different source nodes, the link traffic carrying distribution data of each source node can be obtained. The link traffic carrying distribution data can be a vector, and each element corresponds to a link, and its value is the traffic carried by this link in the information propagation process of this source node, or for this source node, the traffic carrying capacity propagated to all target nodes through this link.

[0047] In this embodiment, a quantitative relationship between the link state and the steady-state response is constructed through the response matrix. Using the response amplitude difference between the normal transmission state and the stopped transmission state, the contribution of a single link to the perturbation propagation of multiple target nodes is converted into a measurable traffic carrying value. Thus, a quantitative basis at the mechanism level is provided for subsequent dominant path identification.

[0048] For example, it is possible to trace how local perturbations of nodes in steady-state activities affect the information of all remaining nodes in the system, and find the degree of influence of the source node on other nodes in information propagation. Introduce the response matrix to describe the global response amplitude driven by perturbations, and the specific element form of ; wherein, represents the differential form of represents the differential form of represents the natural logarithm function; represents that the percentage change of node results in the percentage change of node , which can capture the change ratio before and after the information propagation from node to node . This dimensionless method of dividing the change amount by the original amount makes the response coefficient applicable to node activities of different magnitudes (for example, and may differ by several orders of magnitude), thus being applicable to various scenarios.

[0049] In the target network, summing the absolute values of the steady-state changes of other nodes caused by applying a perturbation to node can obtain the total amount of response to the perturbation. What it reflects is the global response to the perturbation : ; For example, in a steady-state network, each node has its own information level value, expressed as a vector steady_state_old. In the steady state, the calculation result of the ordinary differential equation is 0. When a perturbation is added to the source node, the steady state of the entire system will be broken and the differential equation will no longer be 0. After re-iterating to the steady state through the Runge-Kutta algorithm, the information level values in the entire network will change. Assume that in the new steady state, the steady-state vector of the node is steady_state_new1. Then stop the propagation activity of a certain link and evolve the system again from the steady_state_old state (add the same perturbation to the original source node) until the system reaches the new steady state again. Assume that when the new steady state is reached again, the steady-state vector of the node is steady_state_new2.

[0050] The way to quantify the carrying role of this link in the information propagation process is: sum(abs(steady_state_new1 - steady_state_new2)); wherein, sum represents the summation operation, and abs represents the operation of taking the absolute value.

[0051] In one implementation, the response matrix is an N*N matrix, where N represents the number of rows or columns; each element of the response matrix represents the response relationship of a node pair; each row corresponding to the source node in the response matrix represents the difference in response amplitude between the other nodes when reaching the new steady state and the initial steady state after applying a perturbation from the source node.

[0052] Specifically, the response matrix is the only source for calculating the traffic carrying capacity of a link. The response matrix is an N*N matrix, and each row represents the result of taking the absolute value of the difference between the other nodes reaching the new steady state and the original old steady state after adding a perturbation from the source node. In a multi-source - target scenario, only the rows of elements corresponding to the source nodes in the corresponding matrix need to be analyzed.

[0053] For example, when facing a multi-source - target pair scenario, to quantify the traffic carrying capacity of a link, it is not necessary to sum up the entire rows of the response matrix. Instead, first determine the rows corresponding to the source node indices in the response matrix, and then determine and sum up the elements in the columns corresponding to the target node indices in these rows.

[0054] Take the link as an example. To quantify its traffic carrying capacity during information interaction, the network can be evolved to an initial steady state, assumed to be old_steady. Then, a perturbation is applied to a certain source node, and the network is evolved to a new steady state new_steady; then, stop the transmission activity of the link . On the basis of old_steady, add the same perturbation again and re-evolve to a new steady state new_steady_without_ .

[0055] Compare the differences regarding the target nodes between new_steady and new_steady_without_ , and sum up these differences. Thus, the expression for calculating the traffic carrying capacity of a link based on the response matrix is as follows: ; where represents the traffic carrying capacity of the link .

[0056] However, to calculate the traffic carrying capacity of all links, the time-consuming of the evolution process is relatively long. For example, if the target network has E links, it means that E evolutions are required.

[0057] This embodiment shortens the time consumed by the system steady-state evolution through a formula calculation method. First, the calculation formula for the response matrix is: ; where Indicates the response amplitude that affects the node indirectly through the node and needs to be solved using the chain rule; Indicates the link in the process of transmitting the amplitude of the traffic carried by the disturbance incoming from the node ; ; Among them, Indicates the calculation method for solving ; Indicates the derivative of Indicates the derivative of Indicates substituting into the value of the function; Indicates the information level value of the node (the intermediate node in the propagation process).

[0058] The traffic carrying capacity of a specific link in the information propagation process can be directly calculated through the following formula: ; Among them, is the source node; is the target node; is the number of source nodes; is the number of target nodes; Indicates the response amplitude that affects the node through the node ; Indicates the response amplitude that affects the node through the node . Through this formula, the traffic carrying capacity of each link can be calculated with only two evolutions, so that the method of this embodiment can be applied to large-scale networks.

[0059] In one implementation, determining the dominant path according to the link traffic carrying distribution data respectively corresponding to each of the source nodes includes: Determining the traffic carrying capacity of each link in the target network according to the link traffic carrying distribution data respectively corresponding to each of the source nodes; Determining the dominant path according to the traffic carrying capacity of each link.

[0060] Specifically, by applying a specific perturbation to each source node, a vector describing the traffic carried by the link of the source node can be obtained, that is, the link traffic carrying distribution data. Therefore, a single source node can correspond to a vector describing the link traffic carrying. By averaging the vectors of all source nodes, the link traffic carrying distribution in the multi-source - destination scenario can be obtained. Finally, according to the link traffic carrying distribution in the multi-source - destination scenario and a reasonable threshold, the dominant path can be screened out.

[0061] It should be noted that the perturbation is a permanent perturbation. Most related systems are near a permanent critical point and are insensitive to instantaneous perturbations in the long run. Such perturbations decay over time and rarely penetrate deep into the network. Therefore, the amount of information passing through a certain node depends more on the stability of the system rather than the dynamic characteristics of the node.

[0062] In one implementation, determining the dominant path according to the traffic carrying capacity of each link includes: Sorting the traffic carrying capacities of each link to obtain a target sequence; Performing a binary search operation according to the target sequence, and determining a target threshold through the result of the binary search operation; Filtering out several target links according to the target threshold, and determining the dominant path according to each of the target links; the dominant path can connect all source nodes and target node pairs.

[0063] Generally speaking, the traffic carrying capacity of each link is calculated through a perturbation response model. The traffic carrying capacity is a quantitative value of the actual contribution of its information transmission process in the multi-source - destination scenario. The higher the value, the more traffic the link carries and the more crucial its role in the perturbation propagation. Sort all links in descending order of traffic carrying capacity to form a sequence. The binary search operation is to quickly locate a reasonable threshold in the traffic carrying capacity sequence, that is, to obtain the target threshold. This makes the links above this threshold can not only cover the main links of the perturbation propagation but also avoid including too many secondary links.

[0064] Specifically, the design principle of the target threshold is: to ensure that in the information propagation process, the network reconstructed by the links with traffic carrying capacity greater than the threshold can connect all source nodes and target nodes together. In other words, the target threshold needs to satisfy that the network reconstructed by the target links screened according to this threshold can connect all source nodes and target node pairs. After sorting according to the traffic carrying capacity, then gradually test through the binary search method to find a reasonable threshold, that is, the target threshold. If you expect to change the retention ratio of the target links, it can also be achieved by adjusting the threshold.

[0065] In practical applications, links with a traffic-carrying capacity greater than the target threshold will initially be identified as the links included in the dominant path, i.e., the target links. A network is reconstructed based on each target link, and the links corresponding to the nodes with a node degree of 1 in this network are deleted to ensure that there are no nodes with a node degree of 1 (except the source node and the target node) in this network. Here, the node degree is an indicator of how many neighbors a node has. The reason for deleting the links corresponding to the nodes with a node degree of 1 is that for traffic to flow, there must be both ingress and egress, so the links corresponding to the nodes with a node degree of 1 will not be the dominant links.

[0066] If the network reconstructed based on the target links screened out based on the target threshold can connect all the source nodes and target nodes, increase the threshold, reduce the proportion of target links, and test the connectivity again; if the network reconstructed based on the target links screened out based on the target threshold cannot connect all the source nodes and target nodes, reduce the threshold, increase the proportion of target links, and test the connectivity again. The optimal target threshold is to find a threshold that can just connect all the source nodes and target nodes. This threshold can be determined according to the return value of the binary search algorithm. The binary search algorithm will find an index of a sequence, and this index is the optimal result. Therefore, the sequence obtained by sorting the traffic-carrying capacities of each link is input into the binary search algorithm, and the element value corresponding to the sequence index returned by the binary search algorithm is used as the final target threshold.

[0067] Based on the above embodiments, the present invention also provides a dominant path recognition system in a multi-source scenario based on a perturbation response model, as Figure 2 shown. The system includes: A data input module 01, configured to determine the node indices of multiple pairs of source nodes and target nodes in the target network, and input the graph structure data of the target network and the node indices into the perturbation response model; The perturbation response model 02 is configured to, for each link in the target network, sequentially apply perturbations to each source node in two link states: normal propagation effect and stopped propagation effect of the link, and make the target network re-evolve from the initial steady state to a new steady state to obtain two new steady states; determine the traffic carried by this link in the information propagation process of different source nodes according to the response amplitudes of each target node in the initial steady state and the two new steady states, to obtain the link traffic-carrying distribution data of each source node; and determine the dominant path according to the link traffic-carrying distribution data respectively corresponding to each source node.

[0068] Based on the above embodiments, the present invention also provides a terminal, and its principle block diagram can be as Figure 3As shown. The terminal includes a processor, a memory, a network interface, and a display screen connected via a system bus. Among them, the processor of the terminal is used to provide computing and control capabilities. The memory of the terminal includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the terminal is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, it implements a method for identifying the dominant path in a multi-source scenario based on a perturbation response model. The display screen of the terminal can be a liquid crystal display screen or an electronic ink display screen.

[0069] Those skilled in the art can understand that Figure 3 the block diagram of the principle shown in is only the block diagram of the partial structure related to the solution of the present invention, and does not constitute a limitation on the terminal to which the solution of the present invention is applied. The specific terminal may include more or fewer components than those shown in the figure, or combine some components, or have a different component layout.

[0070] In one implementation, one or more programs are stored in the memory of the terminal, and are configured to be executed by one or more processors. The one or more programs include instructions for performing a method for identifying the dominant path in a multi-source scenario based on a perturbation response model.

[0071] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided by the present invention can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or an external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0072] In summary, the present invention discloses a method and system for identifying dominant paths in a multi-source scenario based on a perturbation response model, which relates to the field of computer technology. The method includes: determining the node indices of multiple pairs of source nodes and target nodes in a target network, and inputting the graph structure data of the target network and the node indices into the perturbation response model; determining the dominant paths in a multi-node interaction scenario through the perturbation response model; the perturbation response model is used for: for each link in the target network, respectively applying perturbations to each source node in turn under two link states where the propagation effect of the link is normal and stopped, and evolving the target network from the initial steady state to a new steady state again, to obtain two new steady states; determining the traffic carried by the link in the information propagation process of different source nodes according to the response amplitudes of each target node in the initial steady state and the two new steady states, to obtain the link traffic carrying distribution data of each source node; determining the dominant paths according to the link traffic carrying distribution data respectively corresponding to each source node. The present invention dynamically analyzes the influence of link state changes on the system steady state based on the perturbation response method, rather than relying on static indicators, which is closer to the real interaction mechanism of the network and makes the mechanism modeling more accurate. And by traversing the link states, the actual contribution of each link to the response of the target node is quantified, avoiding ignoring the complex interaction of weak connections and multi-link cooperation. In addition, through the process of the system evolving to the steady state again after perturbation, the traffic allocation and competition at the shared nodes are naturally modeled, rather than relying on local optimal addressing, improving the global rationality of dominant path identification and solving the problem of multi-source target resource competition. The method of the present invention can be applied to various networks in reality, especially providing a new solution idea for the process of finding dominant paths in a multi-source - target scenario. It plays a core supporting role in constructing a propagation risk early warning model based on network science, implementing key node intervention strategies, and optimizing the intelligent governance mechanism of social information systems.

[0073] It should be understood that the application of the present invention is not limited to the above examples. For those of ordinary skill in the art, improvements or transformations can be made according to the above description, and all such improvements and transformations should fall within the protection scope of the appended claims of the present invention.

Claims

1. A method for identifying dominant paths in a multi-source scenario based on a perturbation response model, characterized in that The method includes: Determining the node indices of multiple pairs of source nodes and target nodes in the target network, and inputting the graph structure data of the target network and the node indices into the perturbation response model; Determining the dominant path in the multi-node interaction scenario through the perturbation response model; The perturbation response model is used for: for each link in the target network, under two link states where the propagation effect of the link is normal and stopped respectively, perturbing each of the source nodes in sequence, and evolving the target network from the initial steady state to a new steady state again, obtaining two new steady states; according to the response amplitudes of each target node in the initial steady state and the two new steady states, determining the traffic carried by the link during the information propagation of different source nodes, obtaining the link traffic carrying distribution data of each source node; and determining the dominant path according to the link traffic carrying distribution data corresponding to each source node respectively.

2. The method for identifying the dominant path in a multi-source scenario based on the perturbation response model according to claim 1, characterized in that, Perturbing each of the source nodes in sequence under two link states where the propagation effect of the link is normal and stopped respectively, and evolving the target network from the initial steady state to a new steady state again, obtaining two new steady states, includes: Under each link state, applying a perturbation with a specific amplitude to one of the source nodes of the target network in the initial steady state to make the target network enter an unsteady state; Re-evolving the target network in the unsteady state to a new steady state by solving the ordinary differential equation corresponding to the information flow dynamics equation; the information flow dynamics equation is used to describe the node interaction behaviors in different scenarios.

3. The method for identifying the dominant path in a multi-source scenario based on the perturbation response model according to claim 2, wherein Solving the ordinary differential equation corresponding to the information flow dynamics equation includes: Solving the ordinary differential equation corresponding to the information flow dynamics equation through the Runge-Kutta algorithm.

4. The method for identifying the dominant path in a multi-source scenario based on the perturbation response model according to claim 1, wherein Determining the traffic carried by the link during the information propagation of different source nodes according to the response amplitudes of each target node in the initial steady state and the two new steady states, obtaining the link traffic carrying distribution data of each source node, includes: Obtaining the response amplitudes of each target node in each new steady state through the response matrix; the response matrix is used to describe: the response amplitudes between different node pairs after the target network is perturbed in the steady state, and the response amplitudes between different node pairs after the propagation effect of each link stops; Comparing the response amplitudes of each target node in the initial steady state and the two new steady states, and calculating the response amplitude differences of each target node; Determining the traffic carried by each link during the information propagation of different source nodes according to the response amplitude differences of each target node obtained after perturbing each source node, obtaining the link traffic carrying distribution data of each source node.

5. The method for identifying the dominant path in a multi-source scenario based on the perturbation response model according to claim 4, wherein The response matrix is an N*N matrix, where N represents the number of rows or columns; each element of the response matrix represents the response relationship of a node pair; each row corresponding to the source node in the response matrix represents the response amplitude difference between the other nodes when reaching the new steady state and when in the initial steady state after perturbing from the source node.

6. The method for identifying the dominant path in a multi-source scenario based on the perturbation response model according to claim 1, characterized in that Determining the dominant path according to the link traffic carrying distribution data respectively corresponding to each of the source nodes includes: Determining the traffic carrying capacity of each link in the target network according to the link traffic carrying distribution data respectively corresponding to each of the source nodes; Determining the dominant path according to the traffic carrying capacity of each link.

7. The method for identifying the dominant path in a multi-source scenario based on the perturbation response model according to claim 6, characterized in that Determining the dominant path according to the traffic carrying capacity of each link includes: Sorting the traffic carrying capacity of each link to obtain a target sequence; Performing a binary search operation according to the target sequence, and determining a target threshold through the result of the binary search operation; Screening out a number of target links according to the target threshold, and determining the dominant path according to each of the target links; the dominant path can connect all source node and target node pairs.

8. A dominant path recognition system in a multi-source scenario based on a perturbation response model, characterized in that, The system includes: A data input module, configured to determine the node indexes of multiple pairs of source nodes and target nodes in the target network, and input the graph structure data of the target network and the node indexes into the perturbation response model; The perturbation response model is configured to, for each link in the target network, sequentially apply perturbations to each source node in two link states where the propagation effect of the link is normal and stopped, and make the target network re-evolve from the initial steady state to a new steady state to obtain two new steady states; determining the traffic carried by the link in the information propagation process of different source nodes according to the response amplitudes of each target node in the initial steady state and the two new steady states, to obtain the link traffic carrying distribution data of each source node; determining the dominant path according to the link traffic carrying distribution data respectively corresponding to each of the source nodes.

9. A terminal, characterized in that, The terminal includes a memory and more than one processor; the memory stores more than one program; the program includes instructions for executing the dominant path identification method in a multi-source scenario based on the perturbation response model as described in any one of claims 1-7; the processor is configured to execute the program.

10. A computer-readable storage medium having multiple instructions stored thereon, characterized in that, The instructions are applicable to be loaded and executed by the processor to implement the steps of the dominant path identification method in a multi-source scenario based on the perturbation response model as described in any one of claims 1-7.

Images