Many studies have analyzed "direct" partnerships in co-authorship networks. One the other hand, the whole network structure, including "indirect" links between researchers, has not been sufficiently studied yet. This study aims at deriving knowledge about the communication structures regarding production of papers by analyzing the researchers' activities from different viewpoints considering roles in co-authorship networks. In this study, we compare the co-authorship networks between the theoretical and application areas in computer science. By applying the modified HITS algorithm to the co-authorship networks, we analyze for each researcher in the co-authorship networks (1) the degree of importance as the leader and (2) that as the follower. We further examine the correlation between these two viewpoints. This study has shown that the negative correlation between (1) and (2) is greater in the application area. It suggests that, in computer science, the two roles (i.e., the leader and the follower) are more clearly separated from each other in the application area than in the theoretical area.