Extraction and visualization of citation network

The digital information is expanding with exponential curve. It becomes difficult to find the relevant information. For example, in scientific domain, to identify the most relevant paper for the focused paper, one needs to explore the focused paper´s citations using some citation index such as Google Scholar etc. However, the focused paper may receive 100´s and sometimes 1000´s of citations. With this huge number of citations, it becomes difficult for users to find the highly relevant papers with respect to the focused paper. While exploring the relevant papers from citations of the focused paper, the user may end up with finding only few. This is because of the fact that authors often cite a paper for providing a background study of the domain and hence cited and cited-by paper have no strong relationship. Currently users need to skim all citations to filter such papers. There is a dire need to develop a system that could automate the process of finding the most relevant papers from the citations. In the current work, we have proposed such a system which ranks the most relevant cited papers for a cited-by paper. The system conceptualizes citations as a graph model. Furthermore, the system employs number of weights to find the most relevant papers. Such weights include topic weight, keyword weight, and author weight. The edge between two nodes in the graph gets a maximum Weight if both are the most relevant papers. Subsequently, a visualization technique has been designed to visualize the graph. We have used the dataset of Journal of Universal Computer Science having 1400 papers and 6090 citations. The manual inspection has revealed that the top ranked papers were actually the most relevant papers.