Web病毒式营销核心群体挖掘与推荐策略

Web病毒式营销已经成为电子商务领域中的重要营销策略, 核心群体在其中发挥着重要的作用。为了挖掘核心群体并对其进行商品推荐, 在Web客户信任网络(customer trust network, CTN)的基础上考虑了信任度、评价分数以及推荐次数等因素定义了影响度的概念, 提出了以影响度为基础的节点网络影响集的构建方法以及基于网络影响集的核心群体挖掘算法MCGNIS（mining core group based on network-influence set）, 并以挖掘出的核心群体为对象建立了基于网络影响集的推荐模型RCGNIS(recommending model for core group based on network-influence set）, 设计了相应的推荐算法来计算商品对核心群体的可推荐度。实验证明, 以节点网络影响集为基础挖掘出的核心群体在Web客户信任网络中具有较高的网络覆盖率（network-coverage, NC）, 推荐模型RCGNIS具有很好的推荐准确性, 同时又保持了推荐的多样性。

Mining and recommending strategies based on core groups in Web viral marketing

The Web viral marketing is rapidly becoming an important marketing strategy in the field of e-business. The core groups play an important role in the Web viral marketing. For mining core groups and recommending the commodity to them reasonably and effectively, this paper defined the concept of influence degree with some factors such as trust degree, evaluation score and recommended number of times on the basis of Web customer trust network, introduced the methods to build network-influence set of node on the basis of influence degree and an algorithm called BUNES (building up a network-influence set) firstly; and proposed a recommendation model called RMNES (recommending model based on network-influence set) and the corresponding algorithm for calculating the recommending degree. The experimental results show the core groups based on the network-influence set on the CTN has the advantage of NC, the RCGNIS has preferably accuracy while maintaining the diversity of the recommending.