Research On Maximizing Dynamic Influence In Social Networks

Due to the continuous spread of the network,social networks are dynamic.Dynamic social networks meet different needs of people.Scholars have also discussed and researched social network-related issues from different perspectives,leading to a very practical topic of maximizing influence.The definition of maximizing influence is to select k nodes so that the source information can spread to the largest extent in the social network.Because traditional influence maximization does not take into account the dynamic characteristics of social networks,such influence maximization is not accurate.Maximizing the dynamic influence in social networks refers to considering the dynamic characteristics of the network,selecting the most accurate seed nodes,and then propagating them accordingly,so that the source information is more widely spread.This article has conducted in-depth research on the issue of maximizing dynamic influence in social networks,mainly including the following two aspects.First,changes in the geographic location of users lead to the dynamic nature of social networks.Most users in social networks carry smartphones to share their geographic location.Therefore,there is an emerging demand for users in social networks to promote regional advertising,where target users are located in a spatial area designated by an advertiser.In order to meet this emerging demand,we need to retrieve the seed set with the most influence in a specific region rather than the entire network,which is not possible with traditional influence maximization algorithms.However,the users in this area are not static and will change the social network in the area with the change of location.Therefore,the issue of maximizing the dynamic influence in the area has become our focus.This paper proposes a region-based dynamic influence maximization algorithm,which has NP-hard characteristics and monotonic submodularity.The algorithm first uses a quadtree to save node position information.After selecting the target area,the nodes are filtered by the stored information in the quadtree.Then,a corresponding tuple is generated for the nodes in the corresponding area.The tuple contains the target node,the coordinates of the target node,the propagation relationship function between the target node and other nodes,and the generated sketch.Each node in the sketch can Reach the target node.When the location information of the nodes in the target area changes,we can update the tuples in real time.Finally,the seed nodes in the target area are calculated and the influence is diffused.Experiments on real-world social network datasets verify that our proposed region-based influence maximization algorithm is more efficient than existing influence algorithms and has accuracy.Secondly,in real life,businesses often need to advertise their products.However,existing algorithms for maximizing influence do not take into account theme factors,and users have different degrees of preference for different themes.Therefore,only choosing users who are very interested in the target topic will make the product's influence spread more widely.However,everyone's interests may not remain the same,and the coefficient of interest on different topics is constantly changing.Therefore,this paper proposes a theme-based dynamic influence maximization algorithm.The algorithm has NP-hardness and monotonicity of function under the topic-based propagation model.The topic-based dynamic impact maximization algorithm first filters the data set based on the target topic.Then use tuples to store the topic information and interest vectors of the nodes,and select the seed nodes,and then use the topic propagation model to conduct influence transmission.When the nodes in the filtered set and the theme of the nodes change,the tuples are updated in real time to ensure that the most representative seed nodes can be selected.The experimental results show that the theme-based dynamic influence maximization algorithm proposed in this paper has obvious advantages in efficiency and accuracy.