Evaluating Citizens’ Sentiments in Smart Cities: A Deep Learning Approach

Abstract

Sentiment analysis of user-generated online content is crucial for smart city analytics and relevant social services. Researchers have relied mainly on textual sentiment analysis to develop systems to predict political elections, measure economic indicators, and so on. Recently, social media users are increasingly using images and videos to express their feelings and share emotions. Sentiment analysis of such large-scale visual content, such as those in image tweets, helps to obtain user sentiments toward events or topics and therefore complement textual sentiment analysis. Motivated by the need to leverage large scale yet noisy training data to solve the extremely challenging problem of face sentiment analysis, we employ Convolutional Neural Networks (CNN). We designed a suitable CNN architecture to classify facial emotions and analyze sentiments. We have conducted extensive experiments on labeled images. The results show that the proposed CNN achieved a very good performance in face sentiment analysis with 89.9% of F1-measure.