Research On Influence Maximization Of Multi-information Source Propagation Based On Reverse Reachable Sets

At present,with the wide promotion and application of social media,social network has become an indispensable bridge in people's communication.As an important topic of social network research,influence maximization focuses on the two aspects of wide spread and high efficiency,and is applied in practice as an advanced marketing means.The work carried out by most researchers is mainly focused on optimizing the influence maximization algorithm and constructing a more realistic information dissemination model.However,most of the existing influence maximization algorithms are not suitable for large-scale social networks because of their high time complexity or limited spread of influence.In addition,the existing information dissemination model is only aim at the transmission of single information source,that is,the dissemination of information between users is single and independent.But in the real social network,nodes are affected by a variety of interactive information sources at the same time.Based on the above content,this paper mainly starts from the actual situation of the social network,and studies the communication model of the joint influence of multiple information sources in the social network and the influence maximization algorithm to achieve better timeliness under this model.The main work of this paper is summarized as follows:(1)In this paper,combined with the reverse reachable set sampling method,a Dynamic-Reverse Influence Sampling influence maximization algorithm based on the independent cascade model is proposed.According to the characteristic that the influence propagation function of the algorithm satisfies monotonicity and submodularity,the judgment condition for generating the critical value of random reverse reachable set is set.Automatic debugging generates a certain number of reverse reachable sets,and then the algorithm uses the maximum coverage method to select seed nodes.(2)The MILT(Multiple Information Linear Threshold)model which proposed in this paper considers that the global propagation probability of each edge is affected by the information influence rate and the propagation probability between nodes on the basis of the linear threshold model,and combines the MI-IC model to simulate the influence propagation between nodes.At the same time,the influence maximization algorithm of DR-MILT(Dynamic Reverse-MILT Algorithm)algorithm under MILT model is given by combining D-RIS influence maximization algorithm.Finally,through the experimental verification,it is proved that the modeling of the propagation model of the influence of multiple information sources is more realistic,and the DR-MILT algorithm based on MILT model also has better timeliness.