An Interpretable Linguistically-Grounded Analysis of Scene Saliency in Movie Screenplays
Per-scene UID summaries carry no independent signal once length is controlled, but at the token level salient scenes match non-salient material in the bulk while specifically suppressing the rarest, hardest-to-predict tokens in the upper tail.
ProofPaper ↗
Key points
- The UID picture is layered in long-form discourse.
- UID therefore lives at the within-scene token level rather than as a per-scene aggregate.
- Salient scenes also show a moderate and sustained cast of central characters (following an inverted-U pattern), strong third-person usage, syntactically simpler language, and more abstract vocabulary about motivation and conflict.
- LGSM exploits this signature and improves recall on salient scenes over a strong semantic baseline while yielding better downstream summaries.
Sources (1)
- [1]An Interpretable Linguistically-Grounded Analysis of Scene Saliency in Movie ScreenplaysarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 11, 11:36 AM
Per-scene UID summaries carry no independent signal once length is controlled, but at the token level salient scenes match non-salient material in the bulk while specifically suppressing the rarest, hardest-to-predict tokens in the upper tail.
The UID picture is layered in long-form discourse.
Extractive summary: sentences quoted from the sources.