本研究采用数据挖掘和机器学习等大数据技术，通过设定相关关键词爬取“7.23动车事故”发生后40天内与该事故相关的94,562条微博并进行情感分析，以探讨网民对“人祸”的道德情绪变化特点，同时对不同群体情绪表达差异进行探讨。结果发现：（1）网民对于动车事故主要表达的道德情绪有：愤怒、鄙视、厌恶、同情和爱，且总体上愤怒、鄙视和厌恶三种情绪表达有趋同性，而同情和爱的表达程度较为一致；（2）包含不同道德基础的道德事件与不同的道德情绪相对应，结果与道德基础理论相一致；（3）对于愤怒、厌恶和鄙视，男性普遍有更高的表达倾向和表达程度，而女性更倾向于表达爱和同情且程度更高；（4）对于爱和同情，团体VIP用户组表达的可能性和强度都高于其他用户；个体VIP用户比非VIP用户更可能表达愤怒、鄙视和厌恶，而团体VIP用户表达这类情绪的强度最小。研究表明，虚拟网络环境中人们道德情绪变化特点依然符合道德基础理论，而不同群体在表达道德情绪时的差异性是对道德基础理论相关研究的补充。这说明大数据技术适用于社会心理学领域相关问题的研究，能够从大数据的视角出发对心理学经典理论进行验证和完善。

Weibo provides its users a platform to share opinions with each other and show their emotions towards issues at home and abroad. In the process, massive amounts of data are made, and then becomes the raw material of sentiment analysis. Previous studies mainly focus on how to develop better sentiment analysis technique in the fields of computer science and communication based on basic emotions. In the study, we shifted the attention to deeper description of the moral emotions that evoked by violating people’s moral foundations related to “7.23 Wenzhou Train Collision”. On the other hand, considering the particularity of network environment we also analyzed the difference between different groups like male and female, VIP and normal users.Firstly, we utilized Weibo API to get the Weibo Dataset. Specifically, from July 23rd, 2011 to September 1st, 2011 we used several developer IDs to keep grabbing the public timeline, which is a sample of the real time tweets. Then we used a set of keywords to obtain the tweets related to the 7.23 train accident. Finally we got 94,562 valid tweets, among which 21,466 tweets contain users’ information. Secondly, we conducted sentiment classification by labeling training dataset which was done by 41 experts and running KNN across the whole testing dataset. After that, all tweets in the dataset are assigned scores from 0 to 5 for each categories of sentiment and the sentiment evolution chart was drawn. Thirdly, we related the knee points of the chart to the moral events happened in the context of train accident to identify which emotion was evoked by certain event. Fourthly, we conducted logistic regression and Robust Maximum Likelihood Estimator (MLR) to analyze the different emotional expression of different groups to the perspective of both qualitative and quantitative.Results indicated that people mainly showed moral anger, contempt, disgust, compassion and love towards to the moral events that happened in the context of train accident. And different moral events that related to certain moral foundation were relevant to corresponding moral emotion. Men were more likely to express anger, disgust or contempt and the emotional intensity was higher than women, while women exceeded men at both possibility and emotional intensity in the expression of compassion and love. For compassion and love, organizational VIP users showed more possibility as well as intensity than individual VIP users and normal users. But when it came to anger, disgust or contempt, individual VIP users had more possibility and emotional intensity to express than other normal users and organizational VIP users.The results supported the Moral Foundation Theory, which violation of different moral foundation elicits corresponding moral emotions. Moreover, the study showed, at first time, some group difference of the expression of moral emotion in the networking environment. Data mining and machine learning technique are suitable for emotion research, even though many limitations exist.