Reproduce “Automated Storytelling via Causal, Commonsense Plot Ordering” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Prithviraj Ammanabrolu, Wesley Cheung, William Broniec, Mark O. Riedl
Researchers at the Georgia Institute of Technology developed C2PO, a system for automated story generation that leverages "soft causal relations" derived from commonsense knowledge to enhance narrative coherence. Human evaluations demonstrated C2PO's stories were significantly preferred for causal plausibility and overall quality compared to baselines in both mystery and fairy tale genres.
Key insights
- Soft causal relations, based on commonsense understanding of character goals and event enablement, are effective in bridging the gap between strict logical planning and implicit neural generation for narrative coherence.
- Incorporating explicit commonsense reasoning significantly enhances both local and global causal coherence, which directly correlates with overall perceived story quality.
- High-quality automated storytelling necessitates a synergistic combination of domain-specific commonsense reasoning and a storytelling-specific inductive bias.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.