Resumen:
In this doctoral thesis, text data extracted from Twitter conversations regarding a natural disasteris analyzed and modelled. In doing so, contributions in different areas emerge: novel Twitterconversation datasets, new tasks scenarios, machine learning models to automatically label thedata. The main goal is to develop a conversational model to help NGOs to cope with the overwhelmingamount of data in the form of conversations, enabling citizens to contribute more efficiently duringnatural disasters.