Early detection of innovations from citation networks

In this paper, we performed a comparative study in two research domains to develop a method of early detection of seeds of innovations. We divided the papers in each research domain into clusters using the topological clustering method, tracked the evolution of the clusters and the positions of the papers in each cluster, and visualized citation networks with cluster name for each cluster. And we also investigated the correlation between future times cited and three measures of centrality, i.e., clustering centrality, closeness centrality, betweenness centrality, the effect of aging and of self-correlation of times cited. With these analyses, we proposed how to distinguish incremental and radical innovations, to detect emerging papers which could be seeds of radical innovations, and to predict the capability of academic papers to be cited in the future.