本文运用数据挖掘与分析方法，以Twitter上中国议题的分布与互动为分析对象，解 析社交机器人的舆论操纵行为，探究其行为模式及其与人类的交互关系。在抓取358,656 条推文、测量用户的机器人评分后发现，与中国相关的推文中有超过1/5疑似由机器人 用户发布。不同议题的自动化操纵程度存在差异。在用户互动网络中，机器人用户转 发、提及，但较少引用或回复。机器人可以成功地引发人类用户主动与之互动, 但人类 更倾向于与人类交互。研究认为，在传播内容上，社交机器人的存在可增加人类用户对 于特定信息的接触；在用户交互层面，社交机器人可以成功渗入社交网络，改变既有的 信息交互结构。

This research focuses on the automatic manipulation of public opinion towards “China” on Twitter. By analyzing 358,656 tweets and measuring the bot score of users involved, it fnds that 21.88% of tweets were posted by social bots. The economic resistance to China is more likely to come from human users, but a signifcant proportion of the criticism of China's political system and human rights situation is conducted by social bots. Interaction networks between users show that social bots tend to increase the visibility of information by retweeting tweets and mentioning other users, but rarely cite or reply. Bots can successfully trigger human users to conduct interactions with them, but humans are still more inclined to interact with humans. It argues that social bots can increase humans’ contact with certain information and social bots themselves can be involved in the social networks to change the given information interaction structure.