Word clustering for parallelism in Classical Chinese poems
John Lee, Mengqi Luo · 2016
This paper explores the use of statistical methods to describe the phenomenon of parallelism in Classical Chinese poems. We apply a graph-based clustering method to automatically induce word clusters from a corpus of poems. We describe several methods for computing similarity scores. We compare these methods by evaluating the quality of the induced clusters, with respect to a semantic taxonomy for parallelism proposed by a literary scholar.