Measuring urban sentiments from social media data: a dual-polarity metric approach

Abstract

Urban sentiment, as people’ perception of city environment and events, is a direct indicator of the quality of life of residents and the unique identity of a city. Social media by which people express opinions directly provides a way to measure urban sentiment. However, it is challenging to depict collective sentiments when integrating the posts inside a particular place, because the sentiment polarities will eventually be neutralized and consequently result in misinterpretation. It is necessary to capture positive and negative emotions distinguishingly rather than integrating them indiscriminately. Following the psychological hypothesis that two polar emotions are processed in parallel and can coexist independently, a novel dual-polarity metric is proposed in this paper to simultaneously evaluate collective positive and negative sentiments in geotagged social media in a place. This new measurement overcomes the integration problem in traditional methods, and therefore can better capture collective urban sentiments and diverse perceptions of places. In a case study of Beijing, China, urban sentiments are extracted using this approach from massive geotagged posts on Sina Weibo, a Twitter-like social media platform in China, and then their spatial distribution and temporal rhythm are revealed. Positive sentiments are more spatially heterogeneous than negative sentiments. Positive sentiments are concentrated in scenic spots, commercial and cultural areas, while negative sentiments are mostly around transportation hubs, hospitals and colleges. Following the principle of sense of place, multi-source data are integrated to evaluate the effects of influencing factors. The variation of spatial factors aggravates the heterogeneity of urban sentiment. The discovered spatiotemporal patterns give an insight into the urban sentiment through online behaviors and can help to improve city functionality and sustainability.