SummScreen: A Dataset for Abstractive Screenplay Summarization
Mingda Chen, Zewei Chu, Sam Wiseman, Kevin Gimpel
Abstract
We introduce SUMMSCREEN, a summarization dataset comprised of pairs of TV series transcripts and human written recaps. The dataset provides a challenging testbed for abstractive summarization for several reasons. Plot details are often expressed indirectly in character dialogues and may be scattered across the entirety of the transcript. These details must be found and integrated to form the succinct plot descriptions in the recaps. Also, TV scripts contain content that does not directly pertain to the central plot but rather serves to develop characters or provide comic relief. This information is rarely contained in recaps. Since characters are fundamental to TV series, we also propose two entity-centric evaluation metrics. Empirically, we characterize the dataset by evaluating several methods, including neural models and those based on nearest neighbors. An oracle extractive approach outperforms all benchmarked models according to automatic metrics, showing that the neural models are unable to fully exploit the input transcripts. Human evaluation and qualitative analysis reveal that our nonoracle models are competitive with their oracle counterparts in terms of generating faithful plot events and can benefit from better content selectors. Both oracle and non-oracle models generate unfaithful facts, suggesting future research directions. 1 * Work done while the author was at the University of Chicago 1 Code, data, and trained models are available at https: //github.com/mingdachen/SummScreen [ The apartment ] Sheldon : What color would you like to be ? Leonard : Well , I 'd like to be green , but you know you always take it . Sheldon : That 's not true . Any color 's fine with me . Yeah , I could be aa combination of blue and yellow . Leonard : Blue and yellow make green . Sheldon : Well , then it 's settled . Penny : Hi . Ready to go ? Sheldon : Oh , good news , we ordered lunch , so we can all stay here and play Lord of the Rings Risk . Amy : Sheldon , we said that we would play games with you tonight . Sheldon : Oh , no , we 'll still be playing it tonight , this game can easily take eight hours . Penny : Sweetie , you really thought I 'd want to do this ? Leonard : No . Penny : Well , did you tell him that ? Leonard : Yes . Penny : Did you say it out loud with words ? Leonard : No . Penny : I do n't want to spend the whole day playing a board game . … Sheldon and Leonard are happy playing a board game until Amy and Penny say they are tired of doing what the guys want …
Ask about this paper
Your agent reads all of it.
Lune indexed this paper to the last equation, along with the top-tier papers that cite it. Ask a question and the answer quotes them.
Your agent calls
Luneget_paper_fulltext
Free to start. No credit card required.
Terminal
Install the CLIlune papers fulltext 8ce9e7a1-fe7a-4aee-ab22-b644f80ea0fbCited by top-tier papers50
- Chain of Agents: Large Language Models Collaborating on Long-Context TasksYusen Zhang, Ruoxi Sun, Yanfei Chen, Tomas Pfister et al.NeurIPS 2024 · 297 citations
- Unlimiformer: Long-Range Transformers with Unlimited Length InputAmanda Bertsch, Uri Alon, Graham Neubig, Matthew GormleyNeurIPS 2023 · 176 citations
- DialogLM: Pre-trained Model for Long Dialogue Understanding and SummarizationMing Zhong, Yang Liu, Yichong Xu, Chenguang Zhu et al.AAAI 2022 · 150 citations
- Long Is More for Alignment: A Simple but Tough-to-Beat Baseline for Instruction Fine-TuningHao Zhao, Maksym Andriushchenko, Francesco Croce, Nicolas FlammarionICML 2024 · 96 citations
- InstructRetro: Instruction Tuning post Retrieval-Augmented PretrainingBoxin Wang, Wei Ping, Lawrence McAfee, Peng Xu et al.ICML 2024 · 75 citations
Builds on4
- BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and ComprehensionMike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad et al.ACL 2020 · 1,224 citations
- Dialogue-Based Relation ExtractionDian Yu, Kai Sun, Claire Cardie, Dong YuACL 2020 · 106 citations
- Abstractive Snippet GenerationWei-Fan Chen, Shahbaz Syed, Benno Stein, Matthias Hagen et al.WWW 2020 · 33 citations
- Screenplay Summarization Using Latent Narrative StructurePinelopi Papalampidi, Frank Keller, Lea Frermann, Mirella LapataACL 2020 · 8 citations
Related papers
- A Modular Approach for Multimodal Summarization of TV ShowsLouis Mahon, Mirella LapataACL 2024 · 1 citation
- EmailSum: Abstractive Email Thread SummarizationShiyue Zhang, Asli Celikyilmaz, Jianfeng Gao, Mohit BansalACL 2021
- RNSum: A Large-Scale Dataset for Automatic Release Note Generation via Commit Logs SummarizationHisashi Kamezawa, Noriki Nishida, Nobuyuki Shimizu, Takashi Miyazaki et al.ACL 2022
- Storytelling with Dialogue: A Critical Role Dungeons and Dragons DatasetRevanth Rameshkumar, Peter BaileyACL 2020 · 37 citations
- STORYSUMM: Evaluating Faithfulness in Story SummarizationMelanie Subbiah, Faisal Ladhak, Akankshya Mishra, Griffin Adams et al.EMNLP 2024 · 2 citations
