Hashtags as an information source. Analysing tweets to map La Terra dei Fuochi

In this paper, we present the UNIOR Eye (Earth your estate) corpus made up of 228,412 tweets dealing with environmental crimes. The core of this research is the extraction of information from tweets according to hashtags and their frequency. Then, in an unsupervised scenario, the K-means clustering algorithm is used to automatically cluster the different contents of the tweets by keeping the hashtags hidden. The analysis of each cluster at the lexical level allows us to identify i) alert tweets, ii) political background tweets and iii) personal opinion tweets. Once the alerts tweets about La Terra dei Fuochi are identified (precisely 36,207), we first extract the exact location (preceded by a hashtag) where the crime took place and then establish the type of crime. This information is displayed on a map with the use of Carto.