It is the sentiment that counts: comparing sentiment analysis tools for estimating affective valence in dream reports.
Garrett Baber, Nancy Hamilton, Jeffrey M. Girard, Jamie M Cohen, Matthew Gratton, Samantha Ellis, Eliza Hemmer
Sleep December 1, 2024 DOI: 10.1093/sleep/zsae210 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Peer reviewed |
|---|---|
| Key points | Proposes that sleep and dream researchers can use computational sentiment analysis to automate estimation of dream affect from dream journal text, thereby circumventing the time-intensive manual annotation and coder training that self-reported diary data currently require. |
Abstract
Despite advances in sleep and dream science via neuroscientific methods, self-reported diary data (e. g. dream journals) remain indispensable for capturing theory-relevant phenomenological aspects of dreams including their emotional content [1, 2]. Manual annotation of such data; however, is time-intensive and typically requires human coders to be trained to reliability. To circumvent these bottlenecks, sleep, and dream researchers can apply computational sentiment analysis techniques to automate the estimation of dream affect (DA) from dream journal text.