Corpus-based Extraction of Japanese Compound Verbs
James Breen, Timothy J. Baldwin · 2009
We describe two methods for Japanese compound verb (JCV) extraction, based on synthesis and pattern matching over the Google Japanese n-gram corpus. We de-vise a number of filters to boost the preci-sion of the corpus-based method, and eval-uate the two methods based on a sample of JCVs occurring in varying frequency bands. We also investigate the distribution of JCV token frequency, and the type fre-quency of their components. 1