Sentimental Publics on Display

• In Event: Transforming Subjects into Objects

Abstract

The image of the public has been technologically mediated throughout history. Today, what we witness is the capture and representation of the public through analytics of massive online trace data. This study examines the political and cultural implications of this development with a focus on Chinese sentiment analysis. Sentiment analysis, or the automatic recognition of emotion in texts, belongs to the burgeoning field of affective natural language processing (NLP) and affective computing more broadly. Conceptualizing emotions as verbal reflections, sentiment analysis typically operates by categorizing texts into positive, negative, or neutral affective states. In the digital age, it has become a common technique in the toolkit of opinion exploration and market analysis. This study examines the deployment of sentiment analysis in China’s fast growing “online public opinion analytics” industry, specifically how it is incorporated into its commercial software packages and usage protocols. This emerging regime converts and aggregates individual online expressions into pie charts of categorical affect, which are further circulated and interpreted as public opinion in government and media discourses. I'll situate this development in the larger cultural trend of viewing society as circuits of emotional energy (as opposed to a site of discursive contentions, for example), of which China presents merely one, albeit particularly unsettling, variation.

Author

• Angela Xiao Wu, New York University