张凯伦1 汪超2  王璐1

Zhang Karen 1 Wang Chao 2 Wang Lu 1

（1.安徽工业大学，马鞍山市 243002）

（2.安徽工程大学，芜湖市 241000 ）

(1.Anhui University of Technology，Ma’anshan  243002，China)  

(2.Anhui Polytechnic University，Wuhu  241000，China)

摘要：影响力最大化问题的主要目标是挖掘社交网络中影响力排名前k个用户，使影响力发挥至最大效用。为了提高识别用户的准确率，提出一种基于BP神经网络的影响力最大化算法。该算法基于数据驱动方法利用BP神经网络优化预测模型，并使用改进的樽海鞘优化算法识别有效节点。在BA网络中的实验结果表明，该算法优于其他几种影响力优化算法。 

Abstract: The main goal of the influence maximization problem is to tap the top k-users in the social network in terms of influence to maximize the effectiveness of the influence. To improve the accuracy of identifying users, an influence maximization algorithm based on BP neural network is proposed. The algorithm was based on a data-driven approach using BP neural networks to optimize the prediction model and used an improved salps optimization algorithm to identify valid nodes. Experimental results in an artificial BA network showed that this algorithm outperformed several other influence optimization algorithms.

关键词：影响力最大化 社交网络 数据驱动 优化算法

Key words: Influence maximization  Social Network  Data-driven  Optimization algorithm