Humor in Word Embeddings: Cockamamie Gobbledegook for Nincompoops

TL;DR

This paper demonstrates that humor-related aspects of words can be captured by simple linear directions in word embeddings, revealing individual and group differences in humor perception through computational analysis.

Contribution

It introduces a novel approach to modeling humor in NLP by linking humor features to word embedding directions and analyzing individual and demographic humor preferences.

Findings

Word vectors encode multiple humor aspects.

Individual humor preferences can be predicted from word vectors.

Different demographic groups show distinct humor preferences.

Abstract

While humor is often thought to be beyond the reach of Natural Language Processing, we show that several aspects of single-word humor correlate with simple linear directions in Word Embeddings. In particular: (a) the word vectors capture multiple aspects discussed in humor theories from various disciplines; (b) each individual's sense of humor can be represented by a vector, which can predict differences in people's senses of humor on new, unrated, words; and (c) upon clustering humor ratings of multiple demographic groups, different humor preferences emerge across the different groups. Humor ratings are taken from the work of Engelthaler and Hills (2017) as well as from an original crowdsourcing study of 120,000 words. Our dataset further includes annotations for the theoretically-motivated humor features we identify.