基于重叠社区搜索的传播热点选择方法

随着社交网络的蓬勃发展，信息传播问题由于具有广泛的应用前景而受到广泛关注，影响力最大化问题是信息传播中的一个研究热点.它致力于在信息传播过程开始之前选取能够使预期影响力达到最大的节点作为信息传播的初始节点，并且多采用基于概率的模型，如独立级联模型等.然而，现有的影响力最大化解决方案大多认为信息传播过程是自动的，忽略了社交网站平台在信息传播过程中可以起到的作用.此外，基于概率的模型存在一些问题，如无法保障信息的有效传播、无法适应动态变化的网络结构等等.因此，本文提出了一种基于重叠社区搜索的传播热点选择方法，该方法通过迭代式推广模型根据用户行为反馈逐步选择影响力最大化节点，使社交网站平台在信息传播过程中充分发挥控制作用，并提出一种新型的基于重叠社区结构的方法来衡量节点影响力，根据这种衡量方式来选择传播热点.本文提出了解决该问题的两种精确算法，包括一种基本方法和一种优化方法，以及该问题的近似算法.并通过大量实验验证了精确及近似算法的效率和近似算法的准确率以及迭代式传播热点选择方法的有效性.

Approach for Hot Spread Node Selection Based on Overlapping Community Search

With the development of social network, information diffusion problem has received a lot of attention because of its extensive application prospects, and influence maximization problem is a hot topic of information diffusion. It is aiming at selecting nodes which could maximize the expected influence as initial nodes of information diffusion, and most work adopts probabilistic models such as independent cascade model. However, most existing solutions of influence maximization see the information diffusion process as an automatic process, and ignore the role of social network websites during information diffusion process. Besides, the probabilistic models have some issues, for example, it cannot guarantee the information to be delivered effectively, and it cannot adapt the dynamic networks and so on. Therefore, in this paper we propose an approach of hot spread node selection based on overlapping community search. This approach selects influence maximized nodes step by step through the iterative promotion model according to users'' behavior feedback, and makes the social network websites play the controller role during information diffusion process, we also propose a new method to measure the influence of nodes, which is based on overlapping community structure, and utilize this measure method to select hot spread nodes. In this paper, two exact algorithms are proposed including a basic algorithm and an optimized algorithm, and also an approximate algorithm is proposed. Comprehensive experiments are conducted and demonstrate performance and accuracy of both exact and approximate algorithms, and the effectiveness of the iterative hot spread node selection method.