Discovering Semantic Relatedness between Scientific Articles through Citation Frequency

Finding relationship between scientific articles is dire need of scholarly community. Current systems support the task to some extent, for example showing cited-by articles, related-articles etc. However, in many cases, these systems show a long list of related documents. For example, if a paper has been cited-by 500 papers, the Google Scholar shows the list of those 500 papers. The system itself is unable to recommend the most relevant cited-by articles for a cited article. This hinders the researchers to skim all of the cited paper to search the most suitable papers because of the fact that many researchers just cite the paper to give background study of the topic etc. In this way there is no strong relationship between the both papers. Therefore, there is a need for a system that can discover semantic relatedness between these documents and could recommend a few most relevant cited-by articles. The semantic relatedness could mean different for different people. In this paper, we consider the cited-by paper more relevant to cited paper if the cited paper has been referred within the cited-by paper's text more than a threshold value. We have empirically proved that Citation Frequency could be another measure to find semantically related set of articles. We have tested our system for a dataset of online journal such as: Journal of Universal Computer Science (J. UCS), we found that if cited-by paper refers the cited paper more than five times, both are semantically related and shows a strong relationship.