How does the internet feel today? Sentiment analysis in the era of big data

In keeping with the recent spate of neuro-scientific research exploring the role played by emotion in the decision-making process, data miners have turned their attention to monitoring the ‘mood’ of the internet via real-time analysis of conversations, blog-posts, Tweets, and Facebook posts. So-called ‘sentiment analysis’ or ‘opinion mining’ represents an attempt to extract value from the large amounts of information that people post online on topics ranging from current events to shopping. As in other forms of marketing research the goal is not simply to monitor public opinion (or ‘sentiment’) but to develop strategies for managing it. This paper explores the ways in which search algorithms construct online sentiment based on interpretations of key words in online communications. The goal of the resulting forms of sentiment analysis is not to assess the feelings of individual users, but to create an aggregate overview of the ‘feeling tone’ online with the goal of both anticipating and managing the outcomes with which this can be correlated (the fate of a particular candidate at the polls, the sales of a certain product, etc.). Thus the notion of ‘sentiment’ functions analogously to that of ‘opinion’ in ‘public opinion polling’ – in principle, it promises access to an ‘underlying’ (or ‘invisible to the unaided eye’) shared property of the public, while in practice, it constructs a statistical artefact based on simple forms of averaging and aggregation. The practice portrays itself as a descriptive one, but, given the crudeness of the algorithms and the purposes for which they are mobilised, it is better described as a predictive one. The paper explores the claims that ‘opinion miners’ make on behalf of these emerging data-mining strategies and considers the ways in which sentiment analysis seeks to contribute to what Patricia Clough has described as the process whereby ‘affect and its modulation displace socialization and disciplining as central to sociality’