Practical and theoretical limitations of sentiment analysis on tweets, dreams and conversations

Sentiment analysis aims to analyse people’s opinions, emotions and attitudes towards a certain target such as a product, person or event. We have seen a huge increase in the popularity of automatic sentiment analysis in last two decades following the rise of digital communication and social media.We examine the concept of sentiment analysis and the relation between sentiment analysis and research on emotions in field of the humanities.We discuss the limitations of current methods and illustrate this on the basis of three very different types of case studies. We look at an example of opinion mining and stance detection in corona-related tweets, at positive and negative feelings expressed in dream reports and how sentiment can be detected in spoken conversations with a conversational agent.

These case studies allow us to reflect on one of the themes of this symposium, datafication and the mismatch between human sentiment expressions and what we can measure with current computational methods. 


Iris Hendrickx

Would you like to participate? Don't miss out... register here!

Digital Humanities Tilburg