The thesis investigates how social media recommendation algorithms contribute to echo chambers and user bias reinforcement. The study begins with a review of algorithmic mechanisms (collaborative, content-based, hybrid models) and societal impacts through academic literature. Relevant datasets from platforms such as Reddit are collected, explored, and cleaned. Simulations and sentiment analysis are conducted to evaluate algorithm performance in terms of content diversity and user overlap.