New Optimal Algorithm On Information Diffusion Maximization Problem

With the rapid development of Internet technology,all kinds of mobile devices become more widespread such as mobile phones and tablet computers.The social network is more and more convenient to be used.And it presents more and more diversified and complicated.Internet marketing,electronic advertising and others,which depend on social networks,are also increasingly popular.Excavating the most influential users from social networks are more and more important.The information diffusion maximization algorithm is aimed to find the most influential nodes from social networks.These nodes form the initial transmission set,having the greatest influence in the whole social network.First of all,to make the selected nodes more effective,this paper puts forward a new optimal algorithm(NOA)on the information diffusion maximization problem by synthetically considering the theme feature and degree centrality.And this paper thinks that the nodes of having the biggest theme preference values and degrees are the most influential nodes.Secondly,the paper presents a way of expressing the user influence based on the user preference,in order to descirbe the user influence more accurately.Furthermore,in order to solve the sparse problem of the user rating matrix,the paper also adopts the latent semantic analysis algorithm(LSA)and the matrix filling technique.In addition,in order to compare with other methods more extensively and deeply and verify the effectiveness and superiority of the algorithm,the traditional information diffusion models extend to the topic-based linear threshold model(TLTM)and the topic-based independent cascade model(TICM)which can describe the information diffusion rule more accurately.Finally,this paper conducts a large number of experiments to verify the validity of the algorithm.The results show that NOA algorithm proposed in this paper is 5000 times faster than the climbing greedy algorithm in the worst case and also faster than the CELF algorithm.If we use the nodes selected through NOA algorithm as the initial spread set,the results have a significantly improvement than the climbing greedy algorithm in the diffusion range and the actual influence.And the results also show that the algorithm proposed in this paper is really efficient,accurate and stable.