A service range maximization method based on a variant tim algorithm

Abstract

The application provides a service range maximization method based on a variant TIM algorithm, comprising the following steps: obtaining a target service network with hub node topology characteristics, randomly assigning each node in the service network, and obtaining the influence parameter, the susceptibility parameter and the node degree of each node; calculating the propagation probability of each edge based on the influence parameter of each node and the susceptibility parameter of the corresponding node of the edge; combining the trained variant TIM algorithm to determine the target node group, and determining the target service range based on the target node group, so as to obtain a better coverage range of information propagation.

Description

Technical Field

[0001] This invention relates to the field of network propagation dynamics technology, and in particular to a method for maximizing service range based on a variant of the TIM algorithm. Background Technology

[0002] A service network is an interconnected system comprised of service providers and users. Within this network, participants collaborate, allocate resources, and share information to supply and acquire services, thereby improving service efficiency, optimizing service quality, and enhancing overall service effectiveness. Service scope, as one of the core indicators of a service network, represents the scale of users a single service provider can cover. Within this framework, the core optimization objective of a service network can be summarized as: maximizing service scope through an optimal node selection strategy under the constraint of limited service provider resources. This problem is isomorphic to the influence maximization problem in network science—that is, selecting a fixed number of seed nodes in a given network to optimize the coverage of information propagation. Currently, information propagation maximization algorithms based on node selection strategies are mainly divided into three categories: greedy algorithms, heuristic algorithms, and community detection algorithms. However, most existing information propagation maximization algorithms only identify seed nodes based on network topology characteristics, without identifying the influence of different nodes themselves, resulting in a relatively small coverage of information propagation. Summary of the Invention

[0003] In view of this, the present invention provides a method for maximizing service range based on a variant TIM algorithm.

[0004] According to a first aspect of the present invention, a method for maximizing service coverage based on a variant TIM algorithm is provided, comprising:

[0005] Obtain a target service network with hub node topology characteristics, wherein each node is connected to at least one other node;

[0006] Each node in the service network is randomly assigned a value to obtain the influence parameter, susceptibility parameter, and node degree of each node. The influence parameter represents the ability of this node to influence other nodes, and the susceptibility parameter represents the degree to which this node is influenced by other nodes.

[0007] Based on the influence parameters of each node and the susceptibility parameters of the nodes corresponding to the edges, the propagation probability of each edge is calculated.

[0008] The influence parameters, susceptibility parameters, node degree, and propagation probability of the corresponding edges of each node are imported into the trained variant TIM algorithm to determine the target node group, and the target service range is determined based on the target node group.

[0009] In some embodiments, the propagation probability of each connection is the product of the influence parameter of the first node at both ends of the connection and the susceptibility parameter of the second node.

[0010] In some embodiments, the training steps of the variant TIM algorithm include:

[0011] Obtain a training node network with multiple training nodes, wherein each training node is connected to at least one other training node;

[0012] Calculate the target lower bound;

[0013] The number of targets is calculated based on the target lower bound and the preset calculation coefficients, and the number of reverse reachable sets of the target is generated. The calculation coefficients are related to the size coefficient and error parameters of the training node network.

[0014] From the target number of reverse reachable sets, select the first node with the highest influence score, and delete the reverse reachable sets that include the first node;

[0015] Repeat the step of filtering out the first node with the highest influence score from the target number of reverse reachable sets until a preset number of first nodes are obtained;

[0016] Import a preset number of first nodes into a preset independent cascaded model and calculate the first coverage area;

[0017] If the first coverage area is less than the preset range threshold, update the calculation coefficients and re-execute the steps of calculating the number of targets based on the target lower bound and the preset calculation coefficients, and generating the target number of reverse reachable sets;

[0018] If the first coverage area is not less than the preset range threshold, determine the current calculation coefficients to obtain the trained variant TIM algorithm.

[0019] In some embodiments, the step of selecting the first node with the highest influence score from the target number of backward reachable sets includes:

[0020] Obtain the node degree and influence parameters for each first node;

[0021] The influence score of each first node is obtained by multiplying the node degree and the influence parameter.

[0022] In some embodiments, the calculation steps for the target lower bound value include:

[0023] Continue building a reverse reachable set from scratch until the average width of the generated reverse reachable sets is less than a preset width threshold, thus obtaining the first reverse reachable set and the first lower bound value;

[0024] By continuously constructing multiple reverse reachable sets from the first reverse reachable set, a second reverse reachable set and a target lower bound are obtained.

[0025] In some embodiments, a preset number of first nodes are imported into a preset independent cascaded model, and a first coverage area is calculated, including:

[0026] Set all training nodes in the training node network to an inactive state, and set the first node to an active state;

[0027] Set an initial time point, and start infection based on the initial time point, until no new infected nodes are added in the current time step, then stop adding time steps and calculate the first coverage area.

[0028] In some embodiments, the steps of performing the infection include:

[0029] Calculate the infection rate for each activated training node and the inactive training nodes connected to it, and perform infection based on the infection rate, where the infection rate is the product of the influence parameter of any activated training node and the susceptibility parameter of any inactive training node connected to any activated training node.

[0030] At least one embodiment of the present invention obtains a target service network with hub node topology characteristics, randomly assigns values ​​to each node in the service network to obtain the influence parameter, susceptibility parameter and node degree of each node; calculates the propagation probability of each connection based on the influence parameter of each node and the susceptibility parameter of the node corresponding to the connection; combines a trained variant of the TIM algorithm to determine the target node group, and determines the target service range based on the target node group, thereby obtaining a better information propagation coverage. Attached Figure Description

[0031] Figure 1 This is a flowchart illustrating a service range maximization method based on a variant TIM algorithm, provided in some embodiments of this specification.

[0032] Figure 2 This is a diagram illustrating the node activation state of a service range maximization method based on a variant TIM algorithm provided in some embodiments of this specification.

[0033] Figure 3 This is a schematic diagram of a network including 8 nodes and 7 connecting edges, provided by some embodiments of this specification;

[0034] Figures 4a to 4f This is a schematic diagram illustrating the differences between the variant TIM algorithm provided in some embodiments of the specification and the traditional TIM algorithm. Detailed Implementation

[0035] Many specific details are set forth in the following description to provide a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0036] The terminology used in one or more embodiments of this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the one or more embodiments of this specification. The singular forms “a” and “the” as used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items. The modifications “a” and “a plurality” as used in this disclosure are illustrative and not restrictive, and those skilled in the art will understand that they should be understood as “one or more” unless the context clearly indicates otherwise.

[0037] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this specification, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0038] First, the terms and concepts used in one or more embodiments of this specification will be explained.

[0039] TIM Algorithm: Timsort.

[0040] Current technologies primarily select seed nodes based on network structural characteristics or the frequency of a node's appearance in the reverse reachability set (indirectly utilizing structural characteristics). The basic assumption is that a node's centrality index is positively correlated with its propagation capability. However, this single-dimensional evaluation method has significant limitations. In real-world network propagation scenarios, the connection strength between nodes is a key factor influencing information diffusion, directly determining the probability of information transmission between nodes. Specifically, tightly connected node pairs have a higher probability of information propagation. Therefore, relying solely on the centrality index of structural characteristics may lead to an overestimation of a node's actual propagation capability; this evaluation bias is particularly pronounced in practical applications such as service networks.

[0041] See Figure 1 , Figure 1A flowchart is shown of a service range maximization method based on a variant TIM algorithm according to some embodiments of this specification, specifically including the following steps.

[0042] Step 101: Obtain the target service network with hub node topology characteristics, wherein each node is connected to at least one other node.

[0043] In some embodiments, the execution entity (such as a pre-defined computing device) of a service range maximization method based on a variant TIM algorithm can connect to the target device via a wired or wireless connection, and then obtain the target service network with hub node topology characteristics, wherein each node is connected to at least one other node. The aforementioned wireless connection methods may include, but are not limited to, 3G / 4G / 5G / 6G connections, WiFi connections, Bluetooth connections, WiMAX connections, Zigbee connections, UWB (ultra-wideband) connections, and other currently known or future-developed wireless connection methods.

[0044] Hub node topology characteristics refer to the features of nodes that play a key connecting role in a network topology, characterized by high connectivity, strong propagation ability, and central location, effectively influencing the state transmission of other nodes. A target service network refers to a specific service architecture composed of multiple nodes, each with at least one connection relationship. A node can refer to a basic unit in a network. As an example, a target service network could be a telecommunications network, a logistics distribution network, or other networks with hub node topology characteristics.

[0045] Step 102: Randomly assign values ​​to each node in the service network to obtain the influence parameter, susceptibility parameter, and node degree of each node. The influence parameter represents the ability of this node to influence other nodes, and the susceptibility parameter represents the degree to which this node is influenced by other nodes.

[0046] Step 103: Calculate the propagation probability of each edge based on the influence parameter of each node and the susceptibility parameter of the corresponding node. The corresponding node can refer to the node at the other end of a line connecting to a node. As an example, the propagation probability of each edge is the product of the influence parameter of the first node at both ends of the edge and the susceptibility parameter of the second node. It should be noted that when calculating the propagation probability between two nodes, one node is considered as the propagating node and the other as the receiving node. Therefore, the calculation is based on the influence parameter of the propagating node and the susceptibility parameter of the corresponding node.

[0047] Step 104: Import the influence parameters, susceptibility parameters, node degree, and propagation probability of corresponding edges of each node into the trained variant TIM algorithm to determine the target node group, and then determine the target service area based on the target node group. The target node group can refer to the set of nodes with the largest calculated coverage area. The target service area can refer to the largest coverage area corresponding to the target node group.

[0048] In some optional implementations, the training steps of the variant TIM algorithm include: obtaining a training node network with multiple training nodes, wherein each training node is connected to at least one other training node; calculating the target lower bound; calculating the target number based on the target lower bound and preset calculation coefficients, generating a target number of reverse reachable sets, wherein the calculation coefficients are related to the size coefficient and error parameters of the training node network; selecting the first node with the highest influence score from the target number of reverse reachable sets, and deleting reverse reachable sets that include the first node; repeating the step of selecting the first node with the highest influence score from the target number of reverse reachable sets until a preset number of first nodes are obtained; importing the preset number of first nodes into a preset independent cascaded model and calculating the first coverage area; if the first coverage area is less than a preset range threshold, updating the calculation coefficients and re-executing the step of calculating the target number based on the target lower bound and preset calculation coefficients, and generating a target number of reverse reachable sets; if the first coverage area is not less than the preset range threshold, determining the current calculation coefficients, and obtaining the trained variant TIM algorithm.

[0049] Training nodes refer to network nodes used during the algorithm training phase to construct reverse reachability sets and verify model performance. Their connections simulate the structure of a real service network. The target lower bound refers to a dynamically calculated reference threshold during the generation of the reverse reachability set, ensuring that the sampling results meet the minimum coverage requirement. Calculation coefficients refer to dynamic parameters related to network size and error tolerance, controlling the number of reverse reachability sets generated to balance computational accuracy and efficiency. The reverse reachability set refers to the set of nodes obtained by traversing the network backward from the target node, used to quantify the coverage potential of node propagation influence. The size coefficient refers to a parameter reflecting the total number of network nodes and connection density, used to adjust the algorithm's computational complexity. The error parameter refers to the allowable deviation range between the algorithm's output and the theoretical optimal value, affecting the rigor of the sampling process. The first node refers to the candidate node with the highest influence score in the reverse reachability set, which is preferentially added to the target node group as a seed node. The independent cascading model refers to a classic model simulating information propagation, verifying coverage effects by activating nodes to gradually infect neighboring nodes. The first coverage area can refer to the proportion of nodes actually affected by the current seed node group through the independent cascade model. The coverage threshold can refer to the preset standard for acceptable coverage effect, used to determine whether the algorithm training should be terminated.

[0050] By meticulously defining network parameters and algorithm variables, a complete optimization loop is constructed, from theoretical calculations to practical verification. A dynamic adjustment mechanism enables the algorithm to adapt to network environments of varying sizes, while a multi-level verification system ensures that the output always aligns with the core objective of maximizing service coverage, providing reliable decision support for resource deployment in complex networks.

[0051] In some alternative implementations, the step of filtering the first node with the highest influence score from the target number of backward reachable sets includes: obtaining the node degree and influence parameter of each first node; calculating the product of the node degree and influence parameter to obtain the influence score of each first node.

[0052] In some optional implementations, the steps for calculating the target lower bound include: continuously building a reverse reachable set from zero until the average width of the generated reverse reachable sets is less than a preset width threshold, thus obtaining a first reverse reachable set and a first lower bound; and re-performing the step of continuously building multiple reverse reachable sets from the first reverse reachable set to obtain a second reverse reachable set and a target lower bound.

[0053] like Figure 2 As shown, in some optional implementations, a preset number of first nodes are imported into a preset independent cascaded model, and the first coverage is calculated, including: setting all training nodes in the training node network to an inactive state (e.g., Figure 2 (The blue nodes in the image), the first node is set to active (e.g., blue nodes). Figure 2 (orange nodes in the diagram); set an initial time point, start infection based on the initial time point, and execute infection based on time steps until no new infected nodes are added in the current time step, then stop adding time steps and calculate the first coverage area.

[0054] The effectiveness of node selection is verified by simulating the real propagation process. Its core benefits are reflected in three aspects: First, the time-step infection mechanism accurately simulates the dynamics of information diffusion, ensuring the authenticity of coverage calculations. Second, the global initialization of inactive states and the gradual activation strategy avoid evaluation biases caused by state confusion in traditional methods. Finally, the design using no new infections as the convergence condition not only fully captures the propagation path but also significantly reduces invalid calculations, making the verification process both rigorous and efficient. Overall, this implementation provides a quantifiable and reproducible standardized verification framework for node influence assessment.

[0055] In some alternative implementations, the infection process includes: calculating the infection rate for each activated training node and the inactive training nodes connected to it, and performing infection based on the infection rate, wherein the infection rate is the product of the influence parameter of any activated training node and the susceptibility parameter of any inactive training node connected to any activated training node.

[0056] The following example compares the differences between the variant TIM algorithm and the traditional TIM algorithm in selecting seed nodes. Since the variant TIM algorithm differs from the traditional TIM algorithm only in the node selection phase, this example only demonstrates the parts where the differences exist. Figure 3 The network shown includes 8 nodes and 7 edges. The influence and susceptibility of the nodes are assigned as follows: Figure 3 As shown, for node a, (0.1, 0.2) indicates that the node's influence metric is 0.1 and its susceptibility metric is 0.2; the other nodes are similar. Assume that the frequencies of nodes a, b, c, d, e, f, g, and h in the reverse reachable set are 2, 3, 1, 5, 5, 2, 4, 3 respectively, and their degree is 1, 1, 5, 1, 1, 3, 1, 1 respectively; the influence metrics corresponding to these 8 nodes in the network are 0.1, 0.2, 0.4, 0.2, 0.5, 0.3, 0.6, and 0.4 respectively.

[0057] Traditional TIM algorithm:

[0058] The influence scores for nodes a through h are calculated as 0.1, 0.2, 2, 0.2, 0.5, 0.9, 0.6, and 0.4 respectively. Considering the number of times a node appears in the reverse reachable set, the final influence scores for nodes a through h are: 0.2, 0.6, 2, 1, 2.5, 0.6, 2.4, and 1.2. If only one seed node is selected, then node e should be selected, as its influence score is the highest at 2.5.

[0059] The TIM algorithm variant provided by this invention:

[0060] After selecting seed nodes in six networks using a variant of the TIM algorithm, the propagation range of the seed nodes was statistically analyzed using an independent concatenation model and compared with the results of the classic TIM algorithm. First, a service network was constructed using six open-source network datasets. Then, two behavioral characteristics of the nodes were assigned values ​​using a uniform distribution of [0, 1], and the propagation probability of each edge in the network was calculated based on the node's behavioral characteristics. Taking nodes i and j as an example, if the influence characteristic of node i is 0.3 and the susceptibility characteristic of node j is 0.5, then the propagation probability... =0.15. Then, the influence score for each node is calculated; for node i, its influence score is... =Node Degree *Influence Parameters If node i has a degree of 10 and an influence parameter of 0.5, then its =5. The network structure characteristics, edge weights, and node influence scores are input into the variant TIM algorithm to determine the final seed node group. Then, the obtained seed node group is input into the independent cascade model to begin propagation. Information starts from the seed node, and the propagation probability is determined by the seed node's influence characteristic and the susceptibility characteristics of its neighboring nodes. Assuming node i is a seed node with an influence characteristic of 0.5, and its neighbor node j has a susceptibility characteristic of 0.6, then node i influences node j with a probability of 0.3. If node j is influenced, its state changes from inactive to active; otherwise, j will not be influenced by node i. Finally, when there are no more newly added active nodes in the network, propagation stops, and the nodes in the active state are counted. Furthermore, to verify the effectiveness of the proposed variant TIM algorithm, the final propagation range of the variant TIM algorithm and the classic TIM algorithm under different seed node sizes is compared. The propagation capabilities of the variant TIM algorithm and the classic TIM algorithm are compared as follows: Figures 4a to 4f As shown in the figure (orange represents the variant TIM algorithm), it can be seen that, when selecting different numbers of seed nodes, the variant TIM algorithm always selects the seed node group with the largest final propagation range.

[0061] The beneficial effects of one of the embodiments in this specification include at least the following: by obtaining a target service network with hub node topology characteristics, randomly assigning values ​​to each node in the service network to obtain the influence parameter, susceptibility parameter, and node degree of each node; calculating the propagation probability of each connection based on the influence parameter of each node and the susceptibility parameter of the corresponding node of the connection; and determining the target node group by combining the trained variant TIM algorithm, and determining the target service range based on the target node group, thereby obtaining a better information propagation coverage.

[0062] The preferred embodiments disclosed above are merely illustrative of this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this invention. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. A method for maximizing service range based on a variant TIM algorithm, characterized in that, include: Obtain a target service network with hub node topology characteristics, wherein each node is connected to at least one other node; Each node in the service network is randomly assigned a value to obtain the influence parameter, susceptibility parameter, and node degree of each node. The influence parameter represents the ability of this node to influence other nodes, and the susceptibility parameter represents the degree to which this node is influenced by other nodes. Based on the influence parameters of each node and the susceptibility parameters of the nodes corresponding to the edges, the propagation probability of each edge is calculated. The influence parameters, susceptibility parameters, node degree, and propagation probability of the corresponding edges of each node are imported into the trained variant TIM algorithm to determine the target node group, and the target service range is determined based on the target node group. The training steps of the variant TIM algorithm include: Obtain a training node network with multiple training nodes, wherein each training node is connected to at least one other training node; Calculate the target lower bound; The number of targets is calculated based on the target lower bound and the preset calculation coefficients, and a set of reverse reachable targets is generated. The calculation coefficients are related to the size coefficient and error parameters of the training node network. From the target number of reverse reachable sets, select the first node with the highest influence score, and delete the reverse reachable sets that include the first node; Repeat the step of filtering out the first node with the highest influence score from the target number of reverse reachable sets until a preset number of first nodes are obtained; The preset number of first nodes are imported into a preset independent cascaded model to calculate the first coverage area; If the first coverage area is less than the preset range threshold, update the calculation coefficients and re-execute the step of calculating the number of targets based on the target lower bound and the preset calculation coefficients, and generating the target number of reverse reachable sets; If the first coverage area is not less than the preset range threshold, determine the current calculation coefficients to obtain the trained variant TIM algorithm; The step of selecting the first node with the highest influence score from the target number of reverse reachable sets includes: Obtain the node degree and influence parameters for each first node; Calculate the product of the node degree and the influence parameter to obtain the influence score of each first node; The process of importing the preset number of first nodes into a preset independent cascaded model and calculating the first coverage area includes: Set all training nodes in the training node network to an inactive state, and set the first node to an active state; Set an initial time point, and start infection based on the initial time point, until there are no new infected nodes in the current time step, then stop adding time steps and calculate the first coverage area.

2. The method according to claim 1, characterized in that, The propagation probability of each connection is the product of the influence parameter of the first node at both ends of the connection and the susceptibility parameter of the second node.

3. The method according to claim 1, characterized in that, The calculation steps for the target lower bound value include: Continue building a reverse reachable set from scratch until the average width of the generated reverse reachable sets is less than a preset width threshold, thus obtaining the first reverse reachable set and the first lower bound value; The process of continuously constructing multiple reverse reachable sets from the first reverse reachable set is performed to obtain the second reverse reachable set and the target lower bound.

4. The method according to claim 1, characterized in that, The steps for performing the infection include: Calculate the infection rate for each activated training node and the inactive training nodes connected to it, and perform infection based on the infection rate, wherein the infection rate is the product of the influence parameter of any activated training node and the susceptibility parameter of the inactive training nodes connected to any activated training node.

Images