The Research On Influence Maximization Based On TWC Model

With the rapid and vigorous development of information technology in 21^stcentury,the ties between objects also become closer.A huge and complex social network has been formed among people,as well as countless connections between affairs.Studying and analyzing these networks,understanding their transmission modes,are of great theoretical significance and practical value in advertising,marketing,information monitoring,infectious disease control and so on.Among them,the discussion of network influence maximization is an important hotspot,which is,selecting K nodes as seed nodes to influence the network.The algorithms used either have the larger propagation area,or better efficiency.At present,influence maximization algorithms are divided into influence ranking-based algorithms and Monte Carlo-based algorithms.Most of them are designed on independent cascade model(ICM)and linear threshold model(LTM).However in practice,the transmission is much more complex.In some circumstances we should not only consider whether the nodes can propagate,but also if other nodes tend to accept the influence.Based on this,this paper establishes the T model containing propagation rates as well as tendency,and combines with the weight cascade model(WCM).Therefore a new propagation model:TWC model is proposed.We study this model based on 4 probability models and seven real data sets.The results are as follows:Firstly,in order to solve the screening criteria,the concept of self/area tendentiousness is proposed.By sorting the degree value and tendency of each node,updating the tendency value in real time and decaying the degree,we propose a seed selection algorithm TDD which can be successfully applied to TWCM.Compared with other influence ranking algorithms,the propagation range is significantly improved.Secondly,TCS algorithm based on clique ranking is proposed.The algorithm screens the best clique in the network first,then selects the best node in the best clique.Compared with TDD algorithm,the influence range is further broadened.Finally,aiming at the high complexity of traditional Monte Carlo algorithm(CELF),we propose a new node sorting algorithm—TCPM-CELF.The first step is the candidate seed set screened by TCPM strategy,then the application of CELF.The experimental results show that the accuracy of the algorithm is much more higher than that of the traditional Monte Carlo algorithm,and the running time is obviously shortened.