Quo Vadis Computer Science? The topics of the influential papers during the period 2014-2021

Nicolas Tsapatsoulis, Harris Partaourides, Christos Christodoulou, Constantinos Djouvas

Abstract: In this study we analyze the titles of all influential publications in Computer Science, during the period of 2014-2021, indexed by Scopus, using an innovative graph-partitioning based topic modelling approach. We consider as influential publications the top 1% cited papers per year. The findings of this specific study are interesting per se, showing the dominance of recent advances in deep learning and neural networks, but our main contribution is the methodology we followed which can be easily applied to any other field and with different types of paper information such as abstracts, keywords, etc. Our approach involves the creation of a bipartite graph of paper ids and their title keywords and the projection on the keywords to find and visually present the identified topics.