Search

CN-121981715-A - Competitive overlapping community influence maximization method

CN121981715ACN 121981715 ACN121981715 ACN 121981715ACN-121981715-A

Abstract

The invention discloses a competitive overlapping community influence maximization method, which is characterized in that a competitive independent cascade extension model is constructed, an inter-node activation probability calculation mechanism is introduced, node similarity, interaction strength and user interest preference are comprehensively considered, multi-information competition propagation modeling is realized, meanwhile, a node comprehensive influence assessment method is provided based on overlapping community structure and reverse reachable set sampling, high influence nodes are accurately identified by combining the number of third-order neighbors and community overlapping degree, seed node selection is optimized through important community screening and seed budget allocation strategies, information propagation efficiency is improved, the method is remarkably superior to the existing algorithm on a real social network data set, and excellent performances are shown in the aspects of influence propagation range and calculation efficiency, and the method is suitable for the fields of social network analysis, information popularization, recommendation systems and the like.

Inventors

  • CHEN JIMING
  • Liu Miaoxia

Assignees

  • 江苏大学

Dates

Publication Date
20260505
Application Date
20260127

Claims (6)

  1. 1. A competitive overlapping community influence maximization method is characterized by comprising the following steps: step 1, acquiring input data, namely acquiring a social network diagram Wherein A set of nodes is represented and, Collecting interaction data among nodes, user interest labels and other information to construct a complete social network data set; Step 2, constructing a competitive independent cascade extension model, wherein the model supports simultaneous propagation of two competitive information sources A and B, and each node has three states, namely an A active state, a B active state or an inactive state; Calculating the activation probability among nodes, wherein the activation probability is comprehensively determined by three factors including node similarity, node interaction strength and activation probability of user interests; step 4, carrying out overlapping community division on the social network based on the SLPA algorithm to obtain an overlapping community structure; Step 5, calculating the comprehensive influence of each node, wherein the comprehensive influence comprises the frequency of occurrence of the node in a reverse reachable set, the overlapping degree of communities to which the node belongs and the number of third-order neighbors of the node; And 6, selecting the high-influence node as a seed node set according to the community importance and the seed budget allocation strategy.
  2. 2. A method for maximizing the impact of competing overlapping communities as in claim 1, wherein in step 2, nodes select sources of information with higher probability of activation when competing messages A and B attempt to activate inactive nodes at the same time.
  3. 3. A method for maximizing competitive overlapping community influence as defined in claim 1, wherein in step 3, The calculation formula of the inter-node activation probability is as follows: Wherein, the The strength of interaction of the nodes is represented, The similarity of the nodes is indicated and, Representing the probability of activation based on the user's interests, And Is a weight factor, meets ; The node similarity The method consists of social similarity and community similarity, and the calculation formula is as follows: Wherein, the In the event of a social similarity, Is community similarity; The social similarity The calculation formula is as follows: The community similarity The calculation formula is as follows: Wherein, the Representing nodes Is used to determine the neighbor set of a neighbor, Representing nodes A set of communities to which the present invention pertains, Representing nodes Is used to determine the neighbor set of a neighbor, Representing nodes A community collection to which the community collection belongs; The node interaction strength The calculation formula is as follows: Wherein, the Representing nodes And (3) with The number of interactions between the two, And The partitions represent the minimum and maximum interaction times in the network; activation probability based on user interests The calculation formula is obtained through calculation of an LDA topic model: Wherein, the Representing that information B satisfies user interests Is a function of the probability of (1), Is expressed in user interest Lower node Influencing nodes Is a probability of (2).
  4. 4. A method for maximizing competitive overlapping community influence as defined in claim 1, wherein in step 5, The comprehensive influence of the nodes The calculation formula is as follows: Wherein, the Indicating the frequency of occurrence of the node in the reverse reachable set, The degree of overlap of the communities is indicated, Representing the number of neighbors of the third order, And Is a weight coefficient, and is respectively set to 0.4 and 0.4; the frequency of occurrence of the node in the reverse reachable set The calculation formula is as follows: Wherein, the Representing nodes Is a set of reachable points of the (c), Is an indication function; The community overlapping degree The calculation formula is as follows: Wherein, the Representing nodes Number of communities to which they belong.
  5. 5. The method for maximizing competitive overlapping community influence as claimed in claim 1, wherein in step 6, community importance assessment is constructed, community scale and cross-community connection capacity are combined, important overlapping communities are screened, seed budgets are reasonably distributed based on community relative scale, seed nodes are ensured to be distributed in communities with the highest propagation potential, in each important community, nodes with the highest influence are selected to serve as seed nodes according to node comprehensive influence ranking until the distributed seed budgets are exhausted, and influence maximization is achieved.
  6. 6. The method for maximizing competitive overlapping community influence of claim 5, wherein said community importance assessment index is: Wherein, the Representing the number of overlapping nodes of the community with other communities, Representing community scale; the seed budget allocation formula is as follows: Wherein, the Representation allocation to communities Is used for the seed quantity of the seed, Budgets for the total seeds.

Description

Competitive overlapping community influence maximization method Technical Field The invention relates to the technical field of social network analysis, in particular to a competitive overlapping community influence maximization method. Background The impact maximization problem is one of the core problems in social network analysis, and aims to select the most influential node set to maximize the information propagation range. Traditional influence maximization algorithms mainly focus on the propagation of a single information source, and neglect the complexity of multi-information source competition propagation in a real social network. Meanwhile, the existing algorithm is generally based on a non-overlapping community model, and the overlapping characteristic that nodes possibly belong to a plurality of communities at the same time cannot be fully considered. In a competitive environment, there is an interaction suppression between different information sources, and the activation state of a node is affected by a plurality of pieces of competitive information. In addition, the ubiquitous overlapped community structure in the social network makes the information propagation path more complex, and the traditional method is difficult to accurately evaluate the real influence of the nodes. Therefore, how to effectively model the competition propagation mechanism and comprehensively consider the overlapped community structure becomes an important challenge for the research of the influence maximization. Disclosure of Invention The invention aims to provide a competitive overlapping community influence maximization method aiming at the defects of the prior art. The technical scheme for solving the problems is that the competitive overlapping community influence maximization method specifically comprises the following steps: step 1, acquiring input data, namely acquiring a social network diagram WhereinA set of nodes is represented and,Collecting interaction data among nodes, user interest labels and other information to construct a complete social network data set; Step 2, constructing a competitive independent cascade extension model, wherein the model supports simultaneous propagation of two competitive information sources A and B, and each node has three states, namely an A active state, a B active state or an inactive state; Calculating the activation probability among nodes, wherein the activation probability is comprehensively determined by three factors including node similarity, node interaction strength and activation probability of user interests; step 4, carrying out overlapping community division on the social network based on the SLPA algorithm to obtain an overlapping community structure; Step 5, calculating the comprehensive influence of each node, wherein the comprehensive influence comprises the frequency of occurrence of the node in a reverse reachable set, the overlapping degree of communities to which the node belongs and the number of third-order neighbors of the node; And 6, selecting the high-influence node as a seed node set according to the community importance and the seed budget allocation strategy. Further, in step 2, in the competitive independent cascade extension model, when competing information a and B attempt to activate inactive nodes at the same time, the nodes select an information source with a higher activation probability. Further, in the step3, The calculation formula of the inter-node activation probability is as follows: Wherein, the The strength of interaction of the nodes is represented,The similarity of the nodes is indicated and,Representing the probability of activation based on the user's interests,AndIs a weight factor, meets; The node similarityThe method consists of social similarity and community similarity, and the calculation formula is as follows: Wherein, the In the event of a social similarity,Is community similarity; The social similarity The calculation formula is as follows: The community similarity The calculation formula is as follows: Wherein, the Representing nodesIs used to determine the neighbor set of a neighbor,Representing nodesA set of communities to which the present invention pertains,Representing nodesIs used to determine the neighbor set of a neighbor,Representing nodesA community collection to which the community collection belongs; The node interaction strength The calculation formula is as follows: Wherein, the Representing nodesAnd (3) withThe number of interactions between the two,AndThe partitions represent the minimum and maximum interaction times in the network; activation probability based on user interests The calculation formula is obtained through calculation of an LDA topic model: Wherein, the Representing that information B satisfies user interestsIs a function of the probability of (1),Is expressed in user interestLower nodeInfluencing nodesIs a probability of (2). Further, in step 5, The comprehensive influence of t