研究發現 更新推特的可能是機器人

Twitter has more than 300 million monthly active users. But researchers have estimated that between about 30 million and 50 million of those are Twitter bots--automated accounts that do the bidding of their code--writing creators.

推特擁有超過3億的月度活躍用戶。但研究人員估計，其中約3000萬到5000萬是推特機器人，即執行代碼編寫人員指令的自動賬戶。

"There could newsbots, and there could be spam bots," said Zafar Gilani, a PhD student at the University of Cambridge in the U.K.

在英國劍橋大學就讀博士的扎法爾·吉拉尼說道：“可能是新聞機器人，也可能是垃圾郵件機器人。”

"Or there could be bots doing political infiltration, which is obviously bad. Or social infiltration which could be bad."

“或者可能是機器人在進行政治滲透，這顯然很糟糕。又或者是機器人在進行社會滲透，這也不好。”

Not all bots are bad. Some are just geeky, like a bot that describes imaginary exoplanets. Or another that tweets only prime numbers.

不過並不是所有機器人都是不好的。有一些機器人只是比較呆，比如有的機器人賬號會描繪虛構的系外行星。還有的賬號只發和質數有關的消息。

"It really depends on who the botmaster is and what are the intentions and what are the motivations."

“這其實取決於賬號操控者的身份及其意圖和動機。”

Gilani and his colleagues built an algorithm to single out bots from human accounts, using factors like tweet frequency or content, and how much users interacted with other users. And the system was able to tell bot from human 86 percent of the time.

吉拉尼和同事創建了一種能將機器人從真人賬號中區分出來的算法，這種算法依據的是發佈頻率或內容，以及該用戶與其他用戶的互動程度。該系統區分人類賬號和機器人賬號的準確率爲86%。

But in the case of celebrity accounts -- people with more than 10 million followers -- the bots and humans were harder to tell apart. Because both tend to tweet with more scheduled regularity than the average human. Both follow relatively few people. And both upload a lot of content.

但對於擁有超過1000萬粉絲的名人賬號，系統則很難區分是機器人還是人類。因爲二者更新的頻率要比普通人更規律。另外，二者關注的人都相對較少，而且都會上傳大量內容。

They differ in the details: celebrities don't post as many URLs luring people off Twitter. And they don't retweet as often as bots do.

不過二者在細節上存在區別：名人不會發太多鏈接誘使人們離開推特。他們也不會像機器人那樣頻繁轉發。

The researchers presented the findings at the International Conference on Advances in Social Networks Analysis and Mining in Sydney, Australia.

研究人員是在於澳大利亞悉尼舉辦的“社會網絡分析和挖掘進展國際會議”上發表的這一研究結果。