MBTI Personality Prediction for Fictional Characters Using Movie Scripts

TL;DR

This paper introduces a new benchmark dataset for predicting fictional characters' personality types from movie scripts, highlighting the challenge for current models and proposing a multi-view approach that improves prediction accuracy.

Contribution

The paper presents the Story2Personality benchmark and a novel multi-view model that combines verbal and non-verbal descriptions for improved personality prediction.

Findings

Existing models perform no better than random guessing.

Multi-view model improves prediction accuracy.

Dataset reveals challenges in narrative-based personality understanding.

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 MBTI or Big 5 personality types based on the narratives of the character. 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 for personality prediction using both verbal and non-verbal descriptions, 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.