This repository contains the data, methodology, and analysis pipeline for the study: "Gender Bias in LLM-Generated University Recommendations: Multi-Layered Analysis via Forma Mentis Networks and Psycholinguistics".
This research investigates whether Large Language Models (LLMs) vary academic recommendations based on a user's declared gender. Using Mistral Medium 3.1 and Mistral Large 3.0 as case studies, we analyze how AI-mediated educational advice reflects societal stereotypes and structural linguistic asymmetries