The Design And Implementation Of Cost-based Advertising Strategy In Social Networks

With the rapid development of the Internet,online advertising gets increasingly wide attention and utilization in all field.However,the extensive advertising often causes inconvenience for users.It also wastes advertisers' money and cannot achieve the desired effect.The expansion of social networks brings revolutionary impacts on the online advertising.Personalized advertising can be applied based on users'information in social networks.Meanwhile,the viral marketing that based on social networks use oral propagation to make information cover most of the network,which is the prototype of the influence maximization problem.In the most studies of influence maximization,they not only ignore the cost of choosing the initial seed node,but also ignore the impact of node theme attribute on information dissemination.Based on the above consideration,we proposes a cost-based advertisement problem that takes both the cost of selecting a node and the theme attribute of the node into account and aims at choosing an appropriate initial seed node to achieve the best possible advertising effect based on limited advertising budget.According to the proposed problem,firstly,establishing the model of nodes cost by combining the network marketing platform with the reality;secondly,building the model of influence probability between nodes by using the degree of preference of the node for certain information(the similarity between node theme attributes and advertising subjects)and the intimacy degree between nodes(the attention degree between nodes);finally,proposing a new measure of node influence based on the influence probability between nodes and the cost of nodes.Then,designing an algorithm of selecting seed node based on average probability-Average Probability Algorithm(AvePA).Based on the above research results,a prototype system of cost-based advertising in social networks is designed and implemented.In this paper,experiments based on six datasets and extended independent cascade model are conducted to verify our algorithms' effectivity.The results show that:(a)For avoiding the overlapping scope problem,it is effective to introduce the coverage distance into the algorithms;(b)For the effect of influence,on the undirected topology,performance of AvePA is better than other algorithms and the advantage is more obvious with the advertising budget increasing.On the directed topology,performance of AvePA is superior to other algorithms and it is more stable;(c)For the time efficiency,time efficiency of AvePA is better when comparing with other algorithms.Considering the effect of influence and time efficiency,overall performance of AvePA is optimal.