#WhyDidTheyStay: An NLP-driven approach to analyzing the factors that affect domestic violence victims
Abstract: This paper uses domain specific data extracted from publicly available posts on social media platforms, Reddit and Twitter, to analyze and provide insight into the various psychosocioeconomic (Schmidt and Willis, 2007) factors that affect domestic violence survivors and contribute to their reluctance to leave abusive relationships. We propose a methodology that employs an unsupervised machine learning algorithm and natural language processing to extract abstractive topics from a large corpus of unstructured text. Further, we determine the distribution of these factors across the psychosocioeconomic spectrum and provide a few recommendations for policymakers based on the results of our analysis.
Paper Link : http://www.winlp.org/wp-content/uploads/2021/11/winlp2021_37_Paper.pdf