no code implementations • PACLIC 2018 • Ryosuke Miyazaki, Mamoru Komachi
Previous approaches to training syntax-based sentiment classification models required phrase-level annotated corpora, which are not readily available in many languages other than English.