Predicting Research Collaboration Trends Based On The Similarity Of Publications And Relationship Of Scientists

Nowadays, collaboration is indispensable in solving increasingly complex problems. In the academic context, research collaboration influences many aspects of research problems approached. The research collaboration is beneficial for scientists, especially early-career scientists, to determine potential successful collaborations. Predicting the trend of collaboration is an important step in improving the quality of research collaboration between scientists. In this study, we propose a method for predicting research collaboration trends by taking into account the research similarity and the relationship between scientists. The research similarity is computed by considering the author's profiles. The co-author graph is built to explore new collaborators based on the connections weigh between scientists. We are currently in the process of developing a real system and our system shows promising results in predicting the potential success collaborators.