SES-RREF: The Machine Learning Approach to Credible Metrics of Scholastic Evidence via Recursive Referencing

Citation network analysis of scholarly articles and journals has already been explored in-depth and the subtlety of the differences between citations and references has also been recognized. Despite this recognition, citation network is mainly used for judging the contribution of an author or a journal for the scientific community. Analyzing citations of an article or of the journal to which it belongs, follows a bottom-up approach and provides a varying degree of information. These include the pattern of spread and the influence it has in academics, per se. However, analysis of references provides a top down approach. The present paper introduces the concept of reference network analysis with the objective of measuring the extent of scholarly inculcation of knowledge and effort while pursuing specific research work. Such reference networks can examine how variegated a research is (diversity) and intensity of the concepts studied (depth) by a researcher. We prove that both these aspects play crucial roles in generating recognition by not relying on citations explicitly. The paper uses these features to devise article-level and author-level metrics,like Scholastic Evidence Score and Trust Score. Using two different case studies of highly reputed scholars, we further demonstrate that Trust score of reputed and reliable authors do not fluctuate noticeably with time. On a broader spectrum, the durability of citation might reflect the depth of a scientific contribution. Our contribution imparts multi-dimensional approach to scholarly influence and creates avenues for future explorations in journal credibility study.

Keywords: Convex Optimization, Depth, Diversity, Recursive References (RREF), Scholastic Evidence Score (SES).