The Research Of Influence Maximization Algorithm Based On Harmony Search

Since the issue of maximizing influence was proposed,many scholars have been researched and various algorithms have been proposed to solve this problem.Among them,greedy algorithms are the most common.However,due to the high computational complexity of greedy and its improved algorithms,they are not suitable for running on large-scale networks and can't meet people's requirements for time.However,some algorithms based on degrees have a fast running speed,but the results are not effectively guaranteed.The heuristic algorithm can achieve good results in the optimization of various complex problems.So we try using the harmony search,a new heuristic algorithm,to solve this question.The harmony search algorithm is a heuristic algorithm that simulates the music creation process.Compared with other heuristic algorithms,it has the advantages of fast convergence,good effect,and strong adaptability.However,applying the harmony search algorithm directly to the problem of maximizing influence can be found in two aspects.On the one hand,because the harmony search algorithm needs to evaluate the new solution each time it is generated to determine the direction of optimization,and the traditional assessment methods are all completed through simulation propagation,so each assessment will consume a lot of time.On the other hand,due to the large number of nodes,the optimization speed of the harmony search algorithm on the issue of maximizing influence is too slow.It requires a lot of iterations to generate a good solution,which in turn exacerbates the problem of slow operation.In this paper,the harmony search algorithm is improved for these two problems.For the first question,we by using EDV(Expected diffusion value),a method that activation expectations of the solution's neighbor nodes are calculated as indicators of the effect evaluation.,to be solved.Since the original method is only applicable to the ICM,this paper has made some extensions to it.And it can be applied to LTM and WCM.For the second question,the harmony search algorithm is optimized based on the degree of node ranking.First,construct a solution space sorted by degree of nodes,then remove some obviously impossible nodes to reduce the scope of the solution space,and then optimize by the harmony search algorithm.In order to further improve the optimization speed of the harmony search algorithm,the HMCR adopts the strategy of distribution adjustment,and adopts two ways of adjusting the solution space and randomly selecting K-step neighbors to perform the new solution.Through experiments,we found that compared with the original harmony search algorithm,in the three different propagation models used in this paper,the optimization speed of the improved algorithm has been greatly improved,and the effect of maximizing the influence has been well achieved.The effect shows that the improved algorithm has good adaptability.