The recommendations of academic papers for reading and understanding research conducted by others are crucial to starting research in the academic field. It enables researchers to discover studies that are pertinent to their interests or the topics they are studying. This study presents a methodology for recommending academic papers using a co-occurrence graph based on the relationships of repeated terms between the abstract in the paper and its references. The centroid is the term with the shortest average distance to all other terms in the graph. The abstract of the paper and the text query of the user are evaluated to determine the centroid values of the graph. These centroids are then utilized in the paper recommendation process. Applying the co-occurrence graph with these techniques aims to provide accurate and relevant recommendations that align with user expectations.