This is the page for the Danmuku research of AIM3 Lab.
Live video commenting, commonly known as "danmaku" or "bullet screen", is an emerging feature on online video sites, which allows viewers to post real-time comments anonymously to fly across the screen like bullets.
Danmaku have some unique interactive features: (1) Pin to specific moment of video. (2) Involve rich multimodal information interaction. (3) Present a group chatting scenario which contains various interaction forms among viewers.
Picture of characteristic
VideoIC is large scale video interactive comments dataset introduced in VideoIC: A Video Interactive Comments Dataset and Multimodal Multitask Learning for Comments Generation(ACM MM 2020).
(1)Large in scale
VideoIC consists of 4951 videos spanning 557 hours.
(2)Board Categories
Videos are collected from popular categories on the ‘Bilibili’ video streaming website.
(3) High comments density:
VideoIC contains more than 5 million comments, with 1077 comments per video on average. The figure above shows the Distribution of the number of comments of videos in the VideoIC dataset.
To protect the copyright, the videos should be downloaded by yourself using the aid we provided, the comments can be download after signing the license
License Agreement.pdf
Please sign the license and send it to chenjieting1208@163.com and we will provide the link of data.
@inproceedings{wang2020videoic,
title={VideoIC: A Video Interactive Comments Dataset and Multimodal Multitask Learning for Comments Generation},
author={Wang, Weiying and Chen, Jieting and Jin, Qin},
booktitle={Proceedings of the 28th ACM International Conference on Multimedia},
pages={2599--2607},
year={2020}
}
@article{陈洁婷:167,
author = {陈洁婷, 王维莹, 金琴},
title = {弹幕信息协助下的视频多标签分类},
publisher = {计算机科学},
year = {2021},
journal = {计算机科学},
volume = {48},
number = {1},
eid = {167},
numpages = {7},
pages = {167},
keywords = {分类;多标签;弹幕;视频;标签关系;多模态},
url = {http://www.jsjkx.com/CN/abstract/article_19684.shtml},
doi = {10.11896/jsjkx.200800198}
}
If you have any questions about Video dataset, please contact us by chenjieting1208@163.com or wy.wang@ruc.edu.cn.