Construction of Brand Community Overlap Based on Ensemble Link Prediction Algorithm

摘要

Brand community marketing provides great convenience for companies to accurately target dispersed customers and audiences and carry out targeted marketing activities. The overlapping of friends in the brand community is important for cultivating members' interaction and improving the vitality of the entire network. The meaning. The Scoring Link Prediction Algorithm (SLPA) is the main method for constructing overlapping circle of friends, but traditional SLPA needs a lot of expert experience to select the appropriate algorithm according to the network characteristics, thus limiting its application effect in practice. This paper proposes a hierarchical cluster ensemble model which is based on knowledge granulation (HCEKG) to build a brand community overlap friend circle. The experimental results of 971 community networks in the twitter network show that the proposed HCEKG has more accurate and reliable prediction performance.