engleski

Natural language processing for personality research: A case study of religiousness on the social media site Reddit

Recently, advances in natural language processing (NLP) and machine learning have enabled research on personality cues that are ubiquitous in social media texts. In this study, we investigate whether the relationship between religiousness and personality traits demonstrated in the literature can also be found in social media texts using NLP. We propose a method based on the implementation of the statement-to-item matching framework (Gjurkovic et al., 2022), which we use to filter over 14 million sentences written by users with self-reported Big5 scores from the PANDORA Reddit dataset to find the most informative self-disclosures of users about their religious affiliation. Based on the filtered sentences alone, raters were able to classify a subset of users (n=299) into religious and non-religious groups. Although median differences on personality traits were in the expected direction, these differences were not significant. We present implications, challenges, and opportunities for using this method in future studies.

natural language processing ; religiousness ; text analysis ; SIMPA framework

nije evidentirano