Saved in:
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2601.08510 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913162154999808 |
|---|---|
| author | Tian, Qiuyu Liu, Zequn Li, Yiding Chen, Fengyi Kong, Youyong Guo, Fan Li, Yuyao Shen, Jinjing Xie, Zhijing Luo, Yiyun Zhang, Xin Xia, Yingce |
| author_facet | Tian, Qiuyu Liu, Zequn Li, Yiding Chen, Fengyi Kong, Youyong Guo, Fan Li, Yuyao Shen, Jinjing Xie, Zhijing Luo, Yiyun Zhang, Xin Xia, Yingce |
| contents | Movie screenplays are rich long-form narratives that interleave complex character relationships, temporally ordered events, and dialogue-driven interactions. While prior benchmarks target individual subtasks such as question answering or dialogue generation, they rarely evaluate whether models can construct a coherent story world and use it consistently across multiple forms of reasoning and generation. We introduce STAGE (Screenplay Text, Agents, Graphs and Evaluation), a unified benchmark for narrative understanding over full-length movie screenplays. STAGE defines four tasks: knowledge graph construction, scene-level event summarization, long-context screenplay question answering, and in-script character role-playing, all grounded in a shared narrative world representation. The benchmark provides cleaned scripts, curated knowledge graphs, and event- and character-centric annotations for 150 films across English and Chinese, enabling holistic evaluation of models' abilities to build world representations, abstract and verify narrative events, reason over long narratives, and generate character-consistent responses. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_08510 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | STAGE: A Full-Screenplay Benchmark for Reasoning over Evolving Storie Tian, Qiuyu Liu, Zequn Li, Yiding Chen, Fengyi Kong, Youyong Guo, Fan Li, Yuyao Shen, Jinjing Xie, Zhijing Luo, Yiyun Zhang, Xin Xia, Yingce Computation and Language Artificial Intelligence Movie screenplays are rich long-form narratives that interleave complex character relationships, temporally ordered events, and dialogue-driven interactions. While prior benchmarks target individual subtasks such as question answering or dialogue generation, they rarely evaluate whether models can construct a coherent story world and use it consistently across multiple forms of reasoning and generation. We introduce STAGE (Screenplay Text, Agents, Graphs and Evaluation), a unified benchmark for narrative understanding over full-length movie screenplays. STAGE defines four tasks: knowledge graph construction, scene-level event summarization, long-context screenplay question answering, and in-script character role-playing, all grounded in a shared narrative world representation. The benchmark provides cleaned scripts, curated knowledge graphs, and event- and character-centric annotations for 150 films across English and Chinese, enabling holistic evaluation of models' abilities to build world representations, abstract and verify narrative events, reason over long narratives, and generate character-consistent responses. |
| title | STAGE: A Full-Screenplay Benchmark for Reasoning over Evolving Storie |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2601.08510 |