From consciousness to denial of risk in social media interactions: a model for automatic detection of socio-psychological processes

Socio-psychological processes such as denial, moral disengagement and psychological distance are expressed in natural language, functioning as barriers to effective behaviour in response to risk-exposing situations. Hence the importance of studying communication (e.g. through social media) in the construction of risk. This paper, starting from the theoretical conceptualisation of the processes mentioned above, describes the construction of a model based on categories of risk perception - from consciousness to denial - that allows the classification of texts through machine learning algorithms. Italian tweets related to the Covid-19 pandemic and climate change were collected and analysed in order to: a) identify the linguistic features (i.e. expressions in the language) of the categories of the model and b) create a training set (i.e. a set of analysed data) to train machine learning algorithms for classification (Support-Vector Machines and Random Forest) to detect psychological processes and test the effectiveness of the model itself. The results will be discussed focusing on the use of the proposed tool as a monitoring strategy and as a starting point for the implementation of effective risk communication.