Multi-stage Social Network Profit Maximization Algorithm Research

With the rapid development of social networks,the academic community has a strong interest in the exploration of profit maximization in product promotion and advertising marketing.Although the IC model and the LT model can be well simulated for the influence propagation in real society.However,for the problem of profit maximization in real marketing,these two models cannot accurately describe them.Therefore,there is an urgent need to develop a more realistic communication model and to study algorithms that are closer to real social networks.The problem of profit maximization in social networks can be described as: under a specific propagation model and certain cost constraints,find a set of nodes S so that the set can generate the maximum range of influence after the spread of the spread,and then obtain The biggest profit.To this end,this paper proposes a communication model that is closer to the reality,and studies the problem of profit maximization in static social networks and dynamic networks.The main results of this paper are:1.In the traditional static social network research work,it ignores the reality that the marketing of new products is actually staged,and also neglects that the activation node can repeatedly activate the inactive nodes multiple times during the marketing process.Therefore,based on Ebbinghaus' s "standard forgotten curve" theory,this paper proposes an influence-distribution model IV-MV(Influence Value-Memory Value)based on cumulative memory,and based on this,based on cumulative memory MMP(Memory-stage Maximization Profit),and proved that the problem is NP-hard problem.Experimental results on multiple real data sets show that the MMP algorithm is superior to the existing greedy algorithm in terms of final impact range and running time.2.Although research on profit maximization has been fully developed in recent years,few people have studied the issue of profit maximization in dynamic social networks.In this paper,we propose an effective algorithm DMP(Dynamic Maximization Profit)to solve the profit maximization in a constantly changing dynamic network.We use the idea of "divide and conquer" to divide the whole network into n regional blocks,and consider that some nodes are at the boundary,and can move from the A region to the B region,thus affecting the two regions A and B.But the traditional approach does not take into account this part of the node.To this end we introduce the problem of maximizing profits in dynamic social networks.Consider the stable nodes in the region,find out the seed nodes,consider the boundary nodes in the region,and find the seed nodes from them.Finally,consider the profits brought by the two types of seed nodes,and select the top k to bring the maximum profit.Node.The experimental results on the actual social network dataset show that the DMP algorithm has achieved good results in terms of final impact range and running time.