Recommendations systems have become an important tool for scientists because they simplify the process of discovering related work, without getting distracted by the enormous amount of research publications. Academic research on recommendations systems can be classified on two main categories: the content-based filtering approach, where the papers published by the authors are indexed and the TF-IDF algorithm is applied to calculate the weights for each one of the indexed terms and the collaboration filtering, in which authors are considered to prefer similar publications with authors with akin behavior. In the proposed method, we present a hybrid approach using both of the before mentioned approaches. We contribute to the collaborative filtering by implementing graph based analysis in order to define the importance of each indexed term.