社会化商务中消费者感知推荐信任的聚类方法研究

新兴社会化商务社会中人与人之间的高交互性及推荐信息的海量化和高动态性,对平台分析消费者感知信任提出了新的挑战。然而,对推荐信息进行聚类难以体现消费者主观性及主体间关系。本文将感知推荐信任聚类问题转化为复合网络划分问题,将主观逻辑方法与基于Normal矩阵的谱平分方法相结合构建社会化商务中消费者感知推荐信任的聚类方法。首先,将推荐信息转化为感知推荐信任,然后,从社交网络中抽取感知推荐信任相似度与关系亲密度网络,并构建Normal矩阵用谱平分方法进行划分。最后,通过多组仿真实验证明了该方法的实用性和有效性。该方法能够为社会化商务中消费者信任的分析提供新视角,为平台制定精准化营销策略提供支持。

Clustering Method of Consumers' Perception Recommendation Trust in Social Commerce

The massive and highly dynamic recommendations, as well as the high interactivity among people in the emerging social commerce, have posed a new challenge to the platform in terms of analyzing consumer's perceived trust. Nevertheless, the method of clustering recommendations has its limitation in reflecting consumer's subjectivity and social relationship. In order to solve the problem, this study intends to transform the cluster of recommendations into the partition of a compound network. In this case, a clustering method for consumer's perceived trust in social commerce is elaborated by combining subjective logical method with spectral bisection algorithm based on the Normal matrix. In the proposed method, recommendation information is firstly turned into perceived recommendation trust. Afterwards, similarities between perceived recommendation trust and intimacy degree network are extracted from the social network, and a Normal matrix is generated and then portioned with spectral bisection algorithm. Finally, practicability and effectiveness is proved by several simulation experiments. The model proposed by the study will throw new light on the analysis of consumer's perceived trust and provide precise marketing strategies to the platform.