Addressing Gaps in Scholarship Recommendations: A Review Focusing on State

Abstract

Existing scholarship recommendation systems typically provide only acceptance or rejection outcomes, without offering explanations. This limitation is evident in Kerala, where students often struggle to interpret the numerous complex rules governing scholarships. This review examines existing recommendation approaches, focusing on methods that explain their decisions for merit-based and minority-based schemes. The review evaluates techniques such as rule-based systems, collaborative filtering, content-based filtering, and decision trees, assessing their strengths and weaknesses in providing feedback, personalizing recommendations, and handling challenges such as dispersed information and intricate eligibility criteria. A key observation is that current systems rarely explain rejection reasons or propose alternative opportunities tailored to specific regions like Kerala. The findings of this review aim to guide the development of a scholarship recommendation system using Python’s Scikit-learn, Pandas, and Streamlit. This system will apply an XGBoost-based decision tree ensemble to recommend Kerala-specific merit and minority scholarships, clearly explain rejections, and suggest alternative schemes.