Disaster Tweets: Analysis from the Metaphor Perspective and Classification Using LLM's

Nowadays, social networks, specially Twitter (now X), allow the spread of information about all topics; since this platform is completely open, there is little to none restriction on what a user can post, hence, creating a lack of confidence and trust on the information available. However, the information on Twitter sometimes have hidden meanings, as the users use metaphors to define their ideas. This paper analyzes and classifies a set of texts labeled as disaster and non-disaster, where those labeled as non-disaster include metaphorical context, focusing on the metaphorical tweets and their interaction with large language models such as BERT, RoBERTa and DistilBERT. These experiments showed an improvement compared with the state-of-the-art approaches, demonstrating that these models capture proper metaphorical text representations.