MENSA: Movie Scene Saliency Dataset

Dataset Summary

The dataset, MENSA (Movie Scene Saliency Dataset) is from the paper "Select and Summarize: Scene Saliency for Movie Script Summarization", and consists of movie scripts and their corresponding summaries. Each scene in the movie script is annotated with scene saliency labels. The training set contains silver labels, which are automatically generated, while the validation and test sets contain human-annotated gold labels.

MENSA is also named after my mother, Meena Saxena, as a small tribute to her.

Dataset Structure

The dataset is divided into three parts:

• Training Set: Contains movie scripts and summaries with silver scene saliency labels.
• Validation Set: Contains movie scripts and summaries with human-annotated gold scene saliency labels.
• Test Set: Contains movie scripts and summaries with human-annotated gold scene saliency labels.

License

Creative Commons Attribution Non Commercial 4.0

Citation

@misc{saxena2024select,
      title={Select and Summarize: Scene Saliency for Movie Script Summarization}, 
      author={Rohit Saxena and Frank Keller},
      year={2024},
      eprint={2404.03561},
      archivePrefix={arXiv},
      primaryClass={id='cs.CL' full_name='Computation and Language' is_active=True alt_name='cmp-lg' in_archive='cs' is_general=False description='Covers natural language processing. Roughly includes material in ACM Subject Class I.2.7. Note that work on artificial languages (programming languages, logics, formal systems) that does not explicitly address natural-language issues broadly construed (natural-language processing, computational linguistics, speech, text retrieval, etc.) is not appropriate for this area.'}
}

@inproceedings{saxena-keller-2024-select,
    title = "Select and Summarize: Scene Saliency for Movie Script Summarization",
    author = "Saxena, Rohit  and
      Keller, Frank",
    editor = "Duh, Kevin  and
      Gomez, Helena  and
      Bethard, Steven",
    booktitle = "Findings of the Association for Computational Linguistics: NAACL 2024",
    month = jun,
    year = "2024",
    address = "Mexico City, Mexico",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2024.findings-naacl.218",
    pages = "3439--3455",
}