Beyond Entertainment: Unpacking Danmaku and Comments’ Role of Information Sharing and Sentiment Expression in Online Crisis Videos
This blog post summarizes the above paper about how danmaku and video comments, two different types of video commentary, play different roles in crisis communication. The work is conducted by Changyang He, Lu He, Tun Lu, and Bo Li. This paper will be presented at the 24th ACM Conference on Computer-Supported Cooperative Work and Social Computing (CSCW 2021) and will be published in The Proceedings of the ACM on Human Computer Interaction. Here is a link to the paper.
Can video viewers also become video content contributors?
Video commentary is an indispensable component for video viewers to communicate with others, share relevant information, and express emotions. Conventional video comments are often placed in a comment box under videos, making watching videos and accessing video comments two separate and asynchronous procedures. Danmaku, an emerging commentary interface that has gained increasing popularity especially in Japan and China, makes things different. It floats above videos for several seconds, which influences viewers’ perceptions synchronously.
With live and synchronized danmaku, the video content that viewers see is collectively composed by the original video creator(s) and collaborative contributions of danmaku posts. Such video content “co-creation” by viewers brings potential values in crisis communication, where accurate, timely, and situational information is needed while misinformation is rampant. For example, the danmaku in the white box in the above figure, “You should cover your nose when wearing the mask”, provides a timely correction of the video character’s misbehavior.
Most previous works regard video creators as the sole source of information and take viewers only as someone to engage. Under this model, danmaku was only investigated as a tool for entertaining and engaging. Actually, danmaku also affords viewers’ synchronous participation and collaborative contributions. In other words, danmaku enables viewers to become not only viewers, but also video content contributors. Such collaborative work was of great significance during crises, which helps to collaboratively disseminate situational information and mutually offer emotional support. Previous work largely investigated the collaborative work in crisis communication on traditional social media platforms like Facebook and Twitter, but none focused on the online video settings, an increasingly important channel in crises. This work thus aims to investigate how viewers leverage video commentary to collaboratively contribute to crisis videos beyond entertainment. As a further step, we also explore how the unique features of danmaku such as synchronicity and diverse styles, may characterize crisis communication in danmaku with special patterns compared to traditional comments.
How we conducted the research
We investigate viewers’ contributions from the perspectives of information sharing and sentiment expression, which are two core components in crisis communication. We proposed three research questions regarding the prevalence, emotion categories and information themes, as well as the unique patterns of danmaku and comments in crisis communication.
To answer these research questions, we adopted a mixed-methods approach incorporating NLP and qualitative analysis. We collected 776,865 danmaku and 1,539,629 comments from 879 COVID-19-related videos on Bilibili, the largest danmaku video website in China. We built two machine learning classifiers to quantify the prevalence of information sharing and sentiment expression in danmaku and comments. We applied qualitative thematic analysis to uncover information themes, and built a quantitative emotion classifier to identify the emotion categories. To uncover the unique patterns, we systematically reviewed danmaku and comments samples with corresponding video context, and assisted with quantitative analysis to validate the empirical findings.
What we found
Danmaku and comment satisfy different informational and emotional needs:
- Sentiment expression was more prevalent in danmaku, while information sharing was much more prevailing in video comments
- Viewers leveraged danmaku more to collectively vent positive emotions, and relied on comments more to individually disclose negative criticisms
- More domain knowledge, as well as information focusing on specific video sections, was observed in danmaku, while more information on general video topics was found in comments
Danmaku affords various hypertext crisis communication patterns based on its temporal, spacial and style nature:
- Danmaku with unique display styles, such as the bright-colored, big-font, or fixed-at-the-top danmaku, had a higher information sharing proportion
- Danmaku afforded timely information supplement and regulation. Danmaku immediately following the corresponding scene could effectively reveal misbehavior in videos, supplement relevant knowledge for confusing points, and explain interesting scene details.
- Viewers also used danmaku for collective emotion resonance and interactive emotion communication in crisis
What we contributed to the HCI and CSCW community
- understanding viewers’ contributions to crisis videos through commentary instead of only being engaged
- proposing comprehensive taxonomies for information themes and emotion categories of commentary in crisis videos
- discovering unique hypertext patterns of crisis communication in danmaku and comments
- highlighting design implications for video commentary to facilitate crisis communication
In conclusion, this work reveals the unique benefits of synchronous commentary interface in crisis videos, and sheds light on richer and more intelligent commentary interfaces that satisfy variant sentiment expression and information sharing needs of viewers during crises. For more details, please refer to our paper and feel free to contact me!
Full citation: Changyang He, Lu He, Tun Lu, and Bo Li. 2021. Beyond Entertainment: Unpacking Danmaku and Comments’ Role of Information Sharing and Sentiment Expression in Online Crisis Videos. Proc. ACM Hum.-Comput.Interact.5, CSCW2, Article 411 (October 2021), 27 pages. https://doi.org/10.1145/3479555