Semiautomatic dictionary-based classification of environment tweets by topic
Michela Cameletti, Stephan Schlosser, Daniele Toninelli, Silvia Fabris · Aisberg (University of Bergamo) · 2020
▪ This method can be easily applied to any topic of interest.▪ Instead of using pre-set and already-available dictionaries, the user can create an ad-hoc and personalized dictionary.▪ The method does not relay on a single or on a short list of predefined keywords, the list can be expanded as necessary to be updated and renewed any time it is needed.▪ The method relies on a dictionary that is not based on just single words, but on combinations of words (i.e. on bigrams and trigrams), thus reducing the inclusion of non-pertinent tweets.