NLP Model for Postgraduate Research Examiner Recommendation Based on Scholarly Activity

Fairness and equity is important in postgraduate research assessment to deter malpractice, mitigate victimization of students, and protect the integrity of the process. Most universities in Africa rely on a human based system to appoint internal and external assessors for postgraduate students’ thesis. This approach is fraught with nepotism and devoid of fairness. In this study, a natural language processing model is proposed for the appointment of internal assessors for postgraduate thesis using the Federal University of Technology Akure, Nigeria as a case study. Our approach utilizes the cosine similarity function to compare students’ abstract with published abstracts of qualified examiners in SCOPUS database. Results obtained in this study showed marked difference between existing approach and the proposed method. Difference in cosine similarity scores was found to be in the range 0.0961 and 0.3345. Specifically, the algorithm showed examiners with better similarity to the students’ work. The approach proposed in this work can significantly improve the assignment of assessors for postgraduate students using a transparent, less bias, and more reliable method.