all AI news
The Detection and Understanding of Fictional Discourse. (arXiv:2401.16678v1 [cs.CL])
cs.CL updates on arXiv.org arxiv.org
In this paper, we present a variety of classification experiments related to
the task of fictional discourse detection. We utilize a diverse array of
datasets, including contemporary professionally published fiction, historical
fiction from the Hathi Trust, fanfiction, stories from Reddit, folk tales,
GPT-generated stories, and anglophone world literature. Additionally, we
introduce a new feature set of word "supersenses" that facilitate the goal of
semantic generalization. The detection of fictional discourse can help enrich
our knowledge of large cultural heritage archives …
array arxiv classification cs.cl datasets detection discourse diverse fanfiction fiction generated gpt literature paper reddit stories trust understanding world