Vist-Edit is the dataset of ACL 2019 short paper "Visual Story Post-Editing". Please do not redistribute without our permission. Thanks.
Arxiv: https://arxiv.org/abs/1906.01764
ACLWeb: https://www.aclweb.org/anthology/P19-1658
Crowd-AI Lab: https://crowd.ist.psu.edu/crowd-ai-lab.html
Each Json file contains a machine-generated story (from AREL[1] or GLAC[2]), and five human-edited stories.
Filename: <photo_sequence_id>.json
photo_sequence_id: The concatenation of the photo sequence ids.
photo_sequence_ids: The list of photo sequence ids.
photo_sequence_urls: The list of photo urls.
auto_story_text_normalized: The automatic story generated by the model. Normalized by NLP tools.
edited_story_text: A list of stories.
worker_id: The de-identified id of the worker who edited the story.
edited_story_text: The original text edited by the worker.
normalized_edited_story_text_sent: The normalized edited_story_text.
data_split: train / dev / test.
model: The model used for generation.
note: Notes.
album_id: The album id of the photos.
Human evaluation results of different models on AREL & GLAC. The six evaluation aspects are used in the Visual Storytelling paper [3].
team: The concatenation of the photo sequence ids.
story_id: The list of photo sequence ids.
album_id: The list of photo urls.
worker_id: The de-identified id of the worker who rated the story.
focused: "Focus".
coherent: "Structure and Coherence".
share: " "I Would Share".
human: "This story sounds like it was written by a human".
grounded: "Visually-Grounded"
detailed: "Detailed"
[1] Wang, Xin, et al. "No metrics are perfect: Adversarial reward learning for visual storytelling." arXiv preprint arXiv:1804.09160 (2018).
[2] Kim, Taehyeong, et al. "GLAC Net: GLocal Attention Cascading Networks for Multi-image Cued Story Generation." arXiv preprint arXiv:1805.10973 (2018).
[3] Huang, Ting-Hao Kenneth, et al. "Visual storytelling." Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2016.