Online detection of public attitudes towards China's second-child policy: A linguistic analysis of social media responses

China has relaxed its family planning policy to allow all married couples to have a second child. To ensure a smooth implementation of the new policy (second-child policy), an efficient detection of public attitudes is necessary. However, traditional methods fail to meet the requirement. This paper aims to examine social media responses, and then uses of linguistic analysis methods to build computational models for automatic detection of attitudinal expressions in social media posts. In this study, a total of 17,305 Sina Weibo posts with keywords were collected and analyzed. First, a content analysis was performed on collected posts to determine whether each of them reflected positive, neutral, or negative attitudes. Second, a series of linguistic features were obtained from each post. Third, based on selected linguistic features, decision tree models were built for differentiating among posts indicating different types of attitudes (positive, neutral, and negative). Results showed that 22.30% of analyzed posts indicated negative attitudes. 62.88% of relevant posts reflected a negative desire for bearing a second child. "Cost and burden" was recognized as a major barrier to the birth of a second child (56.55%). The classification accuracy of decision tree models ranged from 71.6 to 74.4%. This paper supports the need for improving public attitudes towards the second-child policy, and confirms that the use of linguistic analysis method can add to the validity of online detection of public attitudes towards significant social events.