The review scoring results in large-scale buzz marketing sites can greatly affect actual purchase activities of many users. In this paper, since the scoring tendency for an item usually changes over time due to several reasons, we propose a method for visualizing its scoring stream data as a timeline based on switching regimes. Namely, by assuming that . fundamental scoring behavior of users in each regime obeys a multinomial distribution model, we first estimate the switching time steps and the model parameters by maximizing the likelihood of generating the observed scoring stream data, and then produce a timeline and its associated dendrogram as our final visualization results by calculating the probability function from the estimated switching regimes. In our experiments using not only synthetic stream data generated from a known ground truth model but also real scoring stream data collected from a Japanese buzz marketing site, we show that our proposed method can produce accurate and interpretable visualization results for such stream data.
展开▼
