How far is the risk? Detecting socio-psychological processes from consciousness to denial in social media exchanges.

Within the framework of the computational social sciences, specifically in social psychology, one of the big issues is to automatically recognise socio-psychological processes in texts. There have been several attempts to identify such processes (e.g., the most well-known, emotions), although mostly with dictionary-based approaches and little still exists on the identification of psychological processes through machine learning, precisely because of the difficulty of the preliminary work required that is very specific to the application domain, namely the creation of a training set. The present contribution addresses the issue of operationalising socio-psychological processes as they are expressed in natural language to create categories for text classification through machine learning algorithms. In particular, within the framework of the online construction of risk (related to two global linked crises: climate change and the Covid-19 pandemic), it focuses on the implementation of a training set aimed at detecting processes which serve as barriers to implementing effective coping behaviours to face risks (i.e., moral justification, psychological distance, denial) in social media exchanges (i.e. Twitter). Problems and potentials of the adopted approach will be discussed, as well as the methodological and applicative implications of the study.