Social networks in China have been targeted by Chinese sensors and told to curb discussions of banned topics, according to a new study.
Released Friday, the study found that messages containing banned terms tended to be deleted from Sina Weibo. Its researchers added that China had tuned its censorship activity to be more aggressive in places where political unrest was high.
Conducted by David Bamman, Brendan O'Connor and Noah Smith from the Language Technologies Institute at Carnegie Mellon University, the study analyzed 57 million messages sent between Jun. 27 and Sep. 30 last year through the Chinese microblogging site. Three months later, the trio returned to Sina Weibo to check and assess messages no longer available on the site to identify terms which caught the censors' attention.
They found that the censorship efforts were similar to the system overseeing China's Web access, also dubbed the Great Firewall, which is known to stop the Chinese people from visiting certain sites outside China, as well as to return no results for banned search terms, censor online chat, and vet blogs.
Censored terms including well-known controversial topics had been subjected to censorship when they emerged as code words for "planned protests". These terms included Falun Gong, a movement banned by the Chinese government, human rights activists Ai Wei Wei and Liu Xiaobo, and Lianghui, a term which referred to a joint meeting of China's parliament and political advisory body.
Sina Weibo's censorship system had also reacted quickly when words or phrases assumed a political meaning. For example, the word "lianghui" had been sensitive when it was used as a code word for "planned protest".
Automated terms had been used to eliminate some messages while others had been deleted manually, noted Bamman, a PhD student at the university and one of the study's authors. While weibos, or microblogs, containing sensitive terms were seldom deleted, anecdotal evidence that showed certain messages were targeted had been overwhelming, he observed.
The study also found significant variation on how how active the system was in different parts of China. In Tibet, where political unrest was high, 50 percent of messages had been deleted, compared to 12 percent in Beijing, and 11 percent across China.
However, not all censored terms had been political in nature. Just like in the United States, spam and pornographic messages had also been subjected to deletion.
Additionally, after the nuclear disaster in Japan last year, weibos with politically innocuous terms such as iodized salt and radioactive iodine had high deletion rates. This was the result of government efforts to quash false rumors about the nuclear accident causing salt contamination, stated researchers of the study.
In a separate report, Sina Weibo said 60 percent of its microblog users would have successfully registered their real identities by Mar. 16, which was the deadline designated by the Chinese government for microbloggers to authenticate their accounts.