Climate change and populists in geolocated Twitter

Abstract

Surveys have been one of the traditional tools to collect public opinions. However, social media are an important alternative to surveys, being a source of information easily available, in high volume and at low cost. There is plenty of literature dealing with the study of different social, political or environmental topics through social media such as Twitter. Climate change is one of these topics and has major relevance in our current society. In addition, politics is a common element of analysis in the platform. Nevertheless, there is not enough insight about the overall quantitative relevance of climate change compared with other topics such as politicians. Moreover, some of the literature focus specifically on geolocated tweets, which are a small fraction of the total posts generated. This work in progress deals with the identification and semantic analysis of geolocated posts in social media. We analyse and compare the presence of climate change with populist politicians in the platform. These political figures often have a controversial stance on climate change while enacting policies affecting millions of citizens. We aim to study how suitable is the platform for spatiotemporal analysis of public opinion on climate change, and how relevant is the topic on it compared to the presence of some populists. We also aim to provide guidance for further research based on geolocated tweets by estimating how much geolocated data is produced by which countries. More than 170 M geolocated tweets were extracted and analysed. Those tweets containing terms related to climate change in the official languages of the 14 most popular countries in the dataset, as well as the names of several politicians were filtered. Then, an analysis was performed to characterise the spatial and temporal global distribution of these posts during most of the past decade. This was compared with the dates of major events related with climate change and politics. Additionally, sentiment analysis was used to characterise the polarity of the posts. This paper presents an estimation of the relative presence of climate change in Twitter based on probably one of the largest geolocated tweets datasets existing.