Sentiment in Citizen Feedback: Exploration by Supervised Learning

Abstract

<i><p>Abstract: Web-based citizen feedback systems have become commonplace in cities around the</p><p> world, resulting in vast amounts of data. Recent advances in machine learning and natural</p><p> language processing enable novel and practical ways of analysing it as big data. This paper</p><p>reports an explorative case study of sentiment analysis of citizen feedback (in Finnish) by</p><p> means of annotation with custom categories (Positive, Neutral, Negative, Angry, Constructive</p><p> and Unsafe) and predictive modelling. We analyse the results quantitatively and qualitatively,</p><p> illustrate the benefits of such an approach, and discuss the use of machine learning in the</p><p> context of studying citizen feedback. Custom annotation is a laborious process, but it offers</p><p> task-specific adaptation and enables empirically grounded analysis. In this study, annotation</p><p> was carried out at a moderate scale. The resulting model performed well in the most frequent</p><p> categories, while the infrequent ones remained a challenge. Nonetheless, this kind of approach</p><p> has promising features for developing automated systems of processing textual citizen feedback.</p><p> </p><p><br /></p><p></p><p></p></i><p><br /></p>