Analyzing the Common Wisdom of Binarization Doctrine in Internationality Classification of Journals: A Machine Learning Approach

Evaluating and identifying “Internationality” of peer reviewed journals is a hotly debated topic. The problem broadly focuses on whether a journal is international or not, indicating a strong tilt toward binary classification doctrine. The manuscript investigates the doctrine, for the first time. The authors have validated their study further by using minimum error rate classifier, investigated theoretical lower and upper bounds of classification error in the context of internationality. The novel approach has rich ramifications in Scientometrics. Further, we propose a new principle of classification that results in greater accuracy fortifying the assertion.

Keywords: Bayesian Error rate function, Binary classification, Journal Internationality, Non-Local Influence Quotient (NLIQ), Other Citation Quotient (OCQ), Source- Normalized Impact per Paper (SNIP), supervised learning, Unified Granular Neural Network (UGNN), Web scraping.