关键词：语义变化;概率模型;模型期望最大化（EM）
摘 要：We present a new model of semantic change that traces the evolution of a word’s meaning through a phylogeny. By mapping meaning changes to Word Net paths, we are able to identify common kinds of semantic change. We find optimal weights for our model using expectation maximization (EM), and use the results to conjecture about the relative frequency of different kinds of semantic change.