Machine Learning and Social Research: Methodological Challenges and Innovative Applications
Panel Session

In recent years, Machine Learning (ML) has become a central tool in social sciences, offering advanced tools to analyse complex and multidimensional data, such as those from social media or IoT sensors (Mazzeo Rinaldi, F., Celardi, E., Miracula, V., & Picone, A., 2025) These methods allow the identification of hidden relationships and patterns, improving the predictive capabilities of social research. However, using ML raises methodological questions, such as the validity and generalizability of models and ethical issues related to the risk of algorithmic bias.
This session will explore how ML can be integrated into quantitative and qualitative approaches, innovating traditional analysis methods. Among the topics covered will be the applications of ML to build predictive models of complex phenomena, analyse unstructured data, and generate new hypotheses in large datasets (Felaco, Amato & Aragona, 2024). The session will provide an opportunity to reflect on the potential and limits of ML, promote an interdisciplinary dialogue, and contribute to methodological innovation in social sciences.

Submissions may address but are not limited to:

-Automated Data Processing: Using ML for data collection, cleaning, and imputing missing data to enhance reliability.
-Data Triangulation: Combining ML and qualitative methods, like sentiment analysis, to enrich research.
-Mixed Strategies: Integrating diverse datasets with algorithms to analyse complex social phenomena.
-Explainable AI: Applying XAI to interpret and increase transparency in complex models.
-Ethical Analysis: Addressing the ethical risks of black box models, especially for vulnerable groups.
-Language Models: Using LLMs to analyse public discourse, detect fake news, and study political rhetoric.

Keywords: Machine learning, Innovative methods, Explainable AI, Hybrid approaches

Francesco Amato, Università degli Studi di Napoli “Federico II”, francesco.amato2@unina.it, Italy
Vincenzo Miracula, Università di Catania, vincenzo.miracula@phd.unict.it, Italy