Happy Together: Learning and Understanding Appraisal From Natural Language

TL;DR

This paper develops deep neural network models to learn and understand appraisal components like agency and sociality from happy language, achieving high accuracy on the HappyDB dataset.

Contribution

It introduces novel embedding methods and compares neural models with traditional approaches for analyzing happiness-related language.

Findings

Achieved 87.97% accuracy on agency detection

Achieved 93.13% accuracy on sociality detection

Demonstrated superiority of neural models over traditional methods

Abstract

In this paper, we explore various approaches for learning two types of appraisal components from happy language. We focus on 'agency' of the author and the 'sociality' involved in happy moments based on the HappyDB dataset. We develop models based on deep neural networks for the task, including uni- and bi-directional long short-term memory networks, with and without attention. We also experiment with a number of novel embedding methods, such as embedding from neural machine translation (as in CoVe) and embedding from language models (as in ELMo). We compare our results to those acquired by several traditional machine learning methods. Our best models achieve 87.97% accuracy on agency and 93.13% accuracy on sociality, both of which are significantly higher than our baselines.