A simple and efficient dialogue generation model incorporating commonsense knowledge

Geonyeong Son, Misuk Kim

Abstract: Highlights•We propose a dialogue generation model using commonsense knowledge embedding.•Performance improved when commonsense knowledge embedding was applied.•A desirable ratio of commonsense embeddings has been detected in various models.•A commonsense embedding ratio causing a drastic drop in performance was detected.•The method can be applied to English and Korean which underlines its extensibility.