Sina Weibo and Twitter: Comparing data and conversation structures


Launched in July 2006, Twitter is an online social network that popularized the concept of “microblogs”. Conciseness was defined at 140 characters by the founders of Twitter and this was chosen as a practical way to update one’s Twitter account from a mobile phone, using short message service (SMS).

In terms of characters, conciseness does not have the same meaning in Western alphabet-based languages as it does in Far-eastern ones like Chinese. A Chinese ideogram may be analogous to a word and an amalgam of four of them may often contain as much meaning as a sentence in an European language. The phenomenon of microblogs in China are transforming the landscape of information diffusion in the country, for instance with the Yihuang (宜黄) self-immolation case in September 2010.

Screenshot-李开复的微博 新浪微博-隨時隨地分享身邊的新鮮事 - Google Chrome
Figure 1. Former Google China chief and Internet entrepreneur Kai-fu Lee is one of the ten most popular “Weiboers” in mainland China with 2,472,508 followers in December 2010.

Chinese Twitters

In the midst of the July 2009 Xinjiang unrest, the Chinese government decided to block access to Twitter and Fanfou, then the leading Twitter clone and now recently re-established in mainland China. Internet giants Sina and Tancent (QQ) immediately filed the void and started offering microblog services to their millions of users in mainland China. In November 2010, there were already over 40 million users signed up to the Sina Weibo microblogging service.

Sina Weibo focused its strategy on becoming a text-based broadcast entertainment media, offering exclusive contents from celebrity microbloggers from across the sinosphere.

Basic elements of Twitter and Sina Weibo and the effects on conversation

“Atom” comes from the Greek atomos, which means uncuttable, or indivisible. In the study of microblogs, the atom is the microblog entry, commonly called a “tweet” in Twitter. It is a message that can serve one of three different functions: it may be an ordinary message (tweet), a repetition of another user’s tweet (retweet) or a reply to another user’s tweet.

Instead of using “RT” to signal a retweet, the Sina Weibo user writes “//”, followed with the retweeted user’s name. Behind these vocabulary trivialities, the structure of the group conversation is in fact dramatically different in practice. Users of the American microblogging service often deviate from the adopted syntax (by using “via @somebody”) or employ legacy Twitter clients that may not appropriately mark an entry as a retweet. Sina Weibo instead makes a good case of preserving original postings.

On Sina Weibo’s official interfaces (Web and mobile), the equivalent of a Twitter retweet is instead shown as two amalgamated entries: the original entry and the current user’s actual entry — which is a commentary on the original entry, often with a mention of his sources, if the original entry was obtained from an intermediary.

Screenshot-Detail小S的微博 新浪微博-隨時隨地分享身邊的新鮮事 - Google Chrome Figure 2. This is a status update made by Taiwanese celebrity user Dees Hsu and shown on her user timeline. In this example from Sina Weibo, the user quotes a entry by singer A-Mei Chang (Note 1) who herself retweeted an entry originally written by Liu Hanya (Note 2). The two are distinct Weibo entries, displayed as an amalgamated entry on Dees Hsu’s user timeline, along with comments and number of reposts of Dees’ repost (Note 3). The subtility of this conversation may be lost on Twitter, where the different parts have to be retrieved over several user timelines. (see: http://t.sina.com.cn/1704116960/wr0sRJbMeG)

Sina also borrows from its experience as one of China’s biggest blogging service providers by introducing a crucial data structure absent from Twitter: the comment.

Twitter conversation is achieved with combinations of hashtags and search, or with successive replies. On Sina Weibo, these methods are also present but comments add the familiarity and structure to microblog conversation not naturally found on Twitter.

Putting a user entry side by side with the original entry also has the effect of centralizing conversation on one original entry. With Sina’s focus on celebrity users, these VIP users’ timeline may often resemble a television variety show where acquainted characters directly speak and mention each other, while the masses write in the comments section and repost to their networks.

Pictures posted on Weibo are also directly hosted by Sina and directly linked as an entry property, rather than taking up space in the text field. It is now evident that combined with the fact that they are written in Chinese language, entries on Weibo convey a lot more information in a single entry topping 140 characters.

Through the analysis of both networks, Twitter’s model naturally suffers from being the first on the market. While keeping the underlying data structure intact, it may be possible to tweak the user interfaces to change the perception of conversation, which is what some third-party visualisations like Moritz Stefaner’s Revisit attempt to do.

The generic nature of the tweet is Twitter’s most defining characteristic and has shaped the social media’s chaotic but multi-directional conversation, relative to Sina Weibo tendency towards centralization.

Of lists and tags

In attempting to group users, both networks have chosen different strategies. Twitter has lists that a user creates to classify the people he reads into “reading lists”, which may then be shared. Sina Weibo instead has a system of self-assigned tags.


Figure 3. Tags let users classify themselves by interests, age groups, origin, etc.

Based on the names of the lists and co-membership of lists across the network — across different users, that is — Twitter can guess what users are considered to share something in common. The system employed by Sina Weibo groups users by interests, therefore allowing suggestion of other users with a similar or compatible set of potential users to follow. Combined with the set of entries reposted by a Sina Weibo user, it becomes increasingly easy to categorize people into personas.

An outlook on visualisations

Sina Weibo’s more rigid structure for conversation means cleaner and higher-quality data, which is a blessing for future visualisations.# Unlike Twitter, Weibo has also codified the geographic location of its users: on signup, a user must select his province and city (or foreign country only, if living abroad) of residence from a finite list. We could naturally think of a very precise map where the circulation of information is modelled.

Maps detailing social networks may also be drawn using information derived from Sina Weibo’s API. Followers and friends information are evident pieces of data to consider, but relationships may also be quantified by the strength of conversation linking pairs of users, through their number of mutually shared entries, comments and mentions on each other’s timeline.

Sets of similar users by tags and entries reposted, combined with geographical location, can also be of great value to Sina and its commercial partners. From a media studies perspective, it would be invaluable to observe the patterns of reposting and characterize types of individuals instrumental to the propagation of news across the Internet.

In the future, we would envisage interfaces for online social networks to navigate more seamlessly the social interactions, giving us more facility to view, discover and assess the relevance of a piece of information based on the number of references throughout the social network.


This article was written for the Information Architecture and Visualisation course at the Hong Kong Polytechnic University School of Design where the author is pursuing a graduate degree in interaction design. He also works full-time just across Victoria Harbour as a technical researcher in digital media at the Journalism and Media Studies Centre at the University of Hong Kong.

A version of this article appeared on the China Media Project website

3 responses to “Sina Weibo and Twitter: Comparing data and conversation structures”

  1. […] but much on my Chinese senses, but I have found Weibo to be a very interesting experience. I found an article which explains the data structures really nicely. It makes for an appealing user experience. On […]

  2. Micah S says:

    The elephant in the room being that the centralized conversation model is much easier to censor under government directive. Instead, it’s touted as a feature.

  3. Tao says:

    Micah: Is that really necessary to interprete this design choice politically?

Leave a Reply

Your email address will not be published. Required fields are marked *