Semantic Similarity versus Co-authorship Networks: A Detailed Comparison

In: info:eu-repo/grantAgreement/EC/H2020/644187/EU/Realising an Applied Gaming Eco-system/RAGE

Abstract

Whether interested in personal work, in learning about trending topics, or in finding the structure of a specific domain, individuals' work of staying up-to-date has become more and more difficult due to the increasing information overflow. ln our previous work our focus has been to create a semantic annotation model accompanied by dedicated views to explore the semantic similarities between scientific articles. This paper focuses on applying our approach on a dataset of 519 project proposal abstracts, with the intention to bring value to the current indexation methodologies that rely primarily on co- citations and keyword matching. Our experiment uses various Social Network Analysis metrics to compare the rankings generated by two complementary models relying on semantic similarity and co-authorship networks. The two models are statistically different based on representative project associations, are significantly correlated in terms of project rankings by eccentricity and closeness centrality, and the semantic similarity network is denser. ; This study is part of the RAGE project. The RAGE project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 644187. This publication reflects only the author's view. The European Commission is not responsible for any use that may be made of the information it contains.