Personality Prediction of Narrative Characters from Movie Scripts

Abstract

An NLP model that understands stories should be able to understand the characters in them. To support the development of neural models for this purpose, we construct a benchmark, Story2Personality. The task is to predict a movie character’s personality based on the narratives of the character in the movie script. Experiments show that our task is challenging for the existing text classification models, as none is able to largely outperform random guesses. We further proposed a multi-view model to use both verbal and non-verbal descriptions for personality prediction, which gives improvement compared to using only verbal descriptions. The uniqueness and challenges in our dataset call for the development of narrative comprehension techniques from the perspective of understanding characters.

Tasks

Reproductions