This paper presents a comprehensive approach to enhance answer ranking in Marathi question answering systems through the integration of lexical-syntactic, semantic, and contextual similarity measures. The goal is to improve the accuracy and relevance of answer selection, advancing Marathi language processing and developing more effective question answering systems. The evaluation is conducted using the Mean Reciprocal Rank (MRR) metric, achieving an impressive MRR value of 0.91 with the weighted combination of similarity measures. These results emphasize the significance of considering multiple dimensions of similarity and utilizing appropriate weights to prioritize different features. The study contributes to the advancement of Marathi question answering systems by demonstrating the effectiveness of integrating lexical-syntactic, semantic, and contextual similarity measures, providing valuable insights for the development of precise and pertinent answer ranking techniques in Marathi language processing.

Question Answering, Answer Ranking, Similarity measures, Lexical-Syntactic Semantic and Contextual similarity, Information Retrieval, Natural Language Processing