Comparing Methods to Retrieve Tweets: a Sentiment Approach

In current times Internet and social media have become almost unavoidable tools to support research and decision making processes in various fields. Nevertheless, the collection and the use of data retrieved from these sources pose different challenges. In a previous paper we compared the efficiency of three alternative methods used to retrieve geolocated tweets over an entire country (United Kingdom). One method resulted as the best compromise in terms of both the effort needed to set it and the quantity/quality of data collected. In this work we further check, in term of content, whether the three compared methods are able to produce “similar information”. In particular, we aim at checking whether there are differences in the level of sentiment estimated using tweets coming from the three methods. In doing so, we take into account both a cross-section and a longitudinal perspective. Our results confirm that our current best option does not show any significant difference in the sentiment, producing scores in between the scores obtained using the two alternative methods. Thus, such a flexible and reliable method can be implemented in the data collection of geolocated tweets in other countries and for other studies based on the sentiment analysis.