Emotion-based story event clustering

Abstract

In this paper we explore how events can be represented and extracted from text stories, and describe the results from our simple experiment on extracting and clustering events. We applied k-means clustering algorithm and NLTK-VADER sentiment analyzer based on Plutchik's 8 basic emotion model. When compared with human raters, some emotions show low accuracy while other emotion types, such as joy and sadness, show relatively high accuracy using our method.