By Kim Da-ye
A group of Korean researchers said Friday they have developed technology to filter rumors from postings on social networking sites (SNS).
Researchers at the Graduate School of Culture Technology at the Korea Advanced Institute of Science and Technology (KAIST) analyzed more than 100 cases of rumors and non-rumors spread on Twitter in the U.S. between 2006 and 2009, according to KAIST.
“This research integrates statistics and mathematics with theories from sociology and psychology. We drew characters of rumors by analyzing an abundance of data,” Cha Mee-young, an assistant professor at KAIST who led the team, said.
“It is still difficult to classify rumors at the very early stage of their dissemination. But it is possible to do so, using big data, once some time has passed.”
By discovering clear characteristics of rumors ― how long they spread for, who disseminates them and which words are frequently used, the team could correctly identify 90 percent of unfounded claims as rumors.
Among unverified information circulating on Twitter that the researchers identified as rumors were allegations that U.S. President Barack Obama was a Muslim, the “antichrist” and illegally obtained his American citizenship.
Social networking sites including Twitter are helping spread messages easier than ever, but not all of them are true or well-intended. One clear characteristic of rumors is that they continue to spread over a long period of time. Information from news will widely spread in a short period of time, and is rarely mentioned by the media again. But unverified information tends to be repeatedly dealt with for several years.
Rumors are also circulated randomly by people who do not know each other while the spread of ordinary information can be tracked along the networks of online acquaintances.
Furthermore, some words associated with assumption and doubt are frequently seen in rumors. They include, “not sure,” “no idea if it works,” and “heard.”
Kwon Se-jeong, the first author of the paper and a PhD candidate at KAIST, said that he began the research after observing damage done by rumors to their victims including celebrities who committed suicide. Unproven folk remedies spreading online that would only deteriorate a patient’s condition also concerned Kwon.
“Rumors circulate for years, so we will be able to make a list of them,” Kwon said.
“I personally hope to identify wrong pieces of medical information and list them in order to help improve the health and welfare system.”
The research was jointly conducted by Jung Kyo-min, a professor at Seoul National University, and Microsoft Research Asia in China.
