Data Mining of Bilibili Introduction Blibili is China's YouTube, the most well-known video barrage site in China, created in June 2009. As of January 2019, Bilibili had 93 million monthly active users, 82% of whom were young people born after 1990. 5Bilibili features a real-time comment feature that floats above the video, which fans call "dan mu", a feature that makes video highly shareable and interactive. Blibili is currently one of the birthplaces of many popular Chinese vocabulary. In this report, we will investigate how Blibili collects, uses and shares user data, as well as the benefits and disadvantages of information mining to users and Bilibili company. Data Mining ▪ Users are required to provide personal information when registering an account and using the Bilibili service, including 3name, gender, date of birth, ID number, passport name, passport number, phone number, email address, shipping address, bank Card information and related additional information. ▪ During the user's use of the service, Bilibili will automatically retain the user's browser and computer information, location information and device information. ▪ Bilibili will invite users to participate in surveys on Bilibili products and services to obtain data information. ▪ Bilibili 2may collect and use user information through cookies and web beacons and store this information as log information. At the same time, 6cookies and web beacons may collect non-personally identifiable information about the user, analyze how the user uses the service, send advertisements to the user that may be of interest to them, or evaluate the effectiveness of the advertising service. ▪ Bilibili uses its 2own cookies and web beacons to provide users with a more personalized user experience and services, and to remember the user's identity, analyze the user's use of the Blibili service, and optimize ads for three purposes. ▪ Information from third parties: User data is also shared with affiliates and partners of Bilibili, including user 8identity information, behavioral information, transaction information, and device information. Bilibili will aggregate this information to improve the website service. Alleged Benefits 1. Recommendation system According to the viewers’ searching history, browsing history and rating history, bilibili will analysis the preference of viewers, and display the same or similar type of videos on the top of viewers’ homepage. Therefore, the users of bilibili can find the videos which they may like quickly and easily, which can increase more choices for the users when they kill time on bilibili. In addition, the advertisement will be provided to suitable people according to this system, which can improve the effectiveness and cost of sponsors, and provide suitable products for people. 2. The most popular videos Bilibili provides many leaderboards in the website, which will show the videos which has been watched or commended most times among different categories within a week, a month and a year. The users can acquire the information about these videos and watch them, which may increase their interest on bilibili. For uploaders, they can know about the preferences of users by these leaderboards, which can give them some advices on improving their videos, and encourage them to upload more videos. 3. Rating system From September 2017, the memberships of bilibili can rate the videos which they have watched from 1 to 10, and the rating system concludes the average rate. Besides, there are many rules which formulated by bilibili about this system, like the qualification of whom can rate and how to guarantee the authenticity of these rates. This rating system allows the users to know the common impression about every video before they watch, so that they can decide whether to watch the video. Furthermore, this system can have a positive impact on improving the quality of videos, and it also provides data for bilibili to create a better surveillance. 4. Danmu Danmu, also called bullet screen, is a pretty popular comment-sharing video platform in China, especially on bilibili. Unlike the traditional video website, “Danmu”, 1the comments floating on the video in the snapshot, moves as video plays, which means viewers can watch the video while reading other viewers’ comments at the same time. By this method, viewers can create a 1large volume of funny or sarcastic comments on videos by themselves, and other viewers can enjoy these comments. It is the reason why many videos can be watched on other website, but most people choose to use bilibili. Potential harms 1.The ‘information cocoon’ effect is enhanced Bilibili’s ‘personalized recommendation’ model exacerbates the negative impact of ‘information cocoon’ (people will be attracted to information they are interested in, which will lead them to gradually lose access to different things and the ability to understand different things) effect. It collects the user's search history, clicks and other personal information for statistics and analysis. Finally, it will regularly push relevant videos according to the habits and hobbies of users. After a long time, people's attention will only be limited to these aspects. 2. User privacy is leaked Detailed personal information must be provided when a user creates a bilibili’s account. For example, name, date of birth, address, passport number, email address and other private information. When providing this information, the user is guaranteed that these will be safely stored on the platform. But in 2018, many bilibili’s users found that their account can be logged in to “Quick Video (Video Player Software under 360 Brands)”, and both personal information and uploaded videos can be found in it. 3. User data is abused Bilibili collects user search records, viewing history and other data. And provide it to third-party partners. For example, some users found that shopping websites automatically recommended 19th-century style clothing when they watched the related videos of 19th-century European clothing, which makes users feel that their information is being abused and their privacy is violated. 4. User's digital labor is exploited In the Bilibili user's use agreement that users allow to share the content and information they publish on the platform to related companies working with bilibili. Rey (2012) found that digital exploitation has already occurred when 7social media users do not receive any monetary compensation from the digital content they create. Bilibili takes profits from the digital work generated by users through sharing with other companies, but the workers do not get any form of profit. Conclusion 41. Bilibili is a video sharing website themed around animation, comic and game based in China; 2. Bilibili asks users to provide massive personal data, which would be used in different ways, and sometimes it may be harmful to their privacy; 3. Bilibili will aggregate the data from users to improve the website service, so many users tend to enjoy the benefits or conveniences afforded by it; 4.Due to the popularity of bilibili in China, it is necessary to establish a useful and effective system on data regulation.