Towards LLM-Generated Affective Phrase Sets for Text Entry Evaluation

Publication Type:

Conference Paper

Source:

Proceedings of the International Conference on Mobile and Ubiquitous Multimedia, Association for Computing Machinery, New York, NY, USA (2024)

ISBN:

9798400712838

URL:

https://doi.org/10.1145/3701571.3703384

Keywords:

LLMs, Phrase sets, Smartphones, text entry, Transcription tasks

Abstract:

We present a novel method to generate affective phrase sets using Large Language Models (LLMs), for use in text entry study transcription tasks. These phrases could be used to adjust participant emotions, and enable performance assessment under various affective conditions. We found that the affective phrases impacted participants’ valence, but not arousal or text entry performance. Though partially succesful, our work provides a foundation for exploring the relationship between emotion and text entry performance in lab studies.

Download full text: