Douban Movie | rebuild trust of movie scoring by re-design & data visualization
In current Information Age, we have undergone tremendous changes in the way we receive, process and trust information. Trust crisis breaks out in movie scoring in the case of Douban Movie recently. In this project, I explore three issues: why movie scoring trust is broken, how to use data intelligence to promote the growth of valuable movie comments, how to keep the growth of comment environment. By redesigning, I try to rebuild trust of Douban Movie as “the spiritual corner” for users and movie lovers.
Thesis project, 2017 | 6 months
B.E. Digital Media Design
College of Design & Innovation, Tongji University, China
Skill: data analysis, data visualization, user experience design, user interface design
《数读信息爆炸时代的电影评分信任危机 — 以豆瓣电影平台为例的改良性设计》
BACKGROUND Credit Crisis of Douban Movie
Douban Movie is a Chinese social website about films, one part of Douban website, which has a big range of user groups. Its movie scoring is famous for credibility and the scoring histogram.
At the end of 2016, trust crisis broke out in movie scoring of Douban Movie. It was criticized by The People's Daily (the biggest newspaper group in China) due to the extremely low scores of a certain film. It entails a wide range of discussion. Many users furiously voted one star to that movie on Douban, while others questioned the credibility of the ratings.
RESEARCH How We Lose Trust In Movie Scoring
To get a clear image of this case, I capture film reviews and scoring of 11,600 users by Python from this movie, which also includes the voting date and the number of thumbs-up.
After visualizing these data, I map them with the social discussion of this case to get the relationship between scoring and public opinion.
What’s more, with the word occurrence frequency statistics, I get a better opinion and emotion analysis of different scoring groups.
EASILY AFFECTED BY SOCIAL DISCUSSION
The voting percentage of one star has a sudden rising after the criticism of newspaper, while some people are trying to vote five star to balance scoring. Furthermore, the box office doesn’t change much. That is, its scoring could be quite emotional.
TOO MANY WORTHLESS COMPLAINING COMMENTS
With the word occurrence frequency statistics of different scoring groups, I get many keywords of complaining about the director and actors, or simple evaluation such as “bad movie”. Such reviews aren’t informative and useful for other users. Thus, users can’t get valuable film reviews quickly or generate emotional resonance with reviews.
TOO COMPLICATED INFORMATION STRUCTURE
By analysizing the information structure of Douban Movie, users can reach 51 different pages within 2 clicks from home page, which is too complicated to use. It has so many functions and data, but some is rarely accessible.
QUESTIONNAIRE & INTERVIEW Who are target users? What're their needs?
The questionnaire is designed for 4 purposes:
1. how users look at movie scoring and reviews
2.how to combine users’ data with their behavior
3. define values and problem areas of improvement
4. define target users, persona and brand value
Since 80s~90s are the most active and widest range of users, I try to analysis and understand their needs to Douban Movie. Finally, I find that the recommendation and management functions are the most needed demand. (as shown on the right diagram)
473 questionnaires are collected, covers age from 60s~90s. 294 of them are 80s~90s. 269 of them are users of Douban Movie, among whom 80% active users are 80s~90s.
The way we get movie information and watch movies have changed a lot. Different apps on the cell phone participates in different viewing phases (before, during, and after watching movie) . Thus, it has a good database of movie preference.
The generating speed of information is far beyond our ability to accept and analyze information. Visualized data helps people get meaningful data efficiently and know more about themselves.
by Movie Reviews
When we decide whether to watch this movie, acquaintance's opinion is still the most influential factor. However, the meaning of acquaintance has changed to whom has the same values and preferences with us, rather than narrow friends in reality.
Intelligent Recommendation Based on Personal Preference
Improve the recommendation on home page according to the personal data network of movie
Valuable recommandation from the preference of friends on Douban Movie
Focus on watching movie as a private experience, build movie’s circle of friends
Improvement of the existing short review recommendation
Filter valuable reviews which comes from Douban movie friends who has similar preference with me
Promote valuable reviews with discussion, such as most controversial or most popular reviews currently
Most controversial review currently
Data visualization of own activity about movie
Shape own portrait of watching movie, then get better recommendation
Record my emotional experience of movie
Visualization of movie scoring and reviews, then get a better understanding of the whole picture