Research on bias adjustment methods of data from social media

在这项研究中，我们计划开发基本方法来减少社交媒体数据中固有的偏见，例如消费者生成的媒体中的意见数据。更具体地说，我们开发了一些统计方法来减少这种偏差，将其视为样本选择偏差，因为消费者生成媒体中文本的发起者是从数据分析师想要了解的目标人群中非随机抽样的。其中一种建议的方法发表在统计学的顶级期刊（美国统计协会杂志）上。

In this research we planned to develop fundamental methods to reduce biases inherent in data from social media such as opinion data in consumer generated media. To be more concrete, we developed some statistical methods to reduce such biases by regarding them to be the sample selection bias in that the originators of the texts in consumer generated media are nonrandomly sampled from the targeted population the analysts of the data would like to know about. One of the proposed methods is published in the top journal in statistics (Journal of the American Statistical Association).