Portal de Periódicos da CAPES

Knowledge-based framework for estimating the relevance of scientific articles

2020; Elsevier BV; Volume: 161; Linguagem: Inglês

10.1016/j.eswa.2020.113692

1873-6793

Alberto Fernández-Isabel, Adrián Alonso Barriuso, Javier Cabezas, Isaac Martín de Diego, J.F. J. Viseu Pinheiro,

Wikis in Education and Collaboration

The volume of published papers provided by the scientific community has increased over the last years in a drastic way. This fact has led to having a considerable growth of the topics covered by different publications. Despite topics under discussion on these publications were usually regarded as cutting edge subjects when released in conferences and journals, the restless evolution of science may have faded their relative importance away over the years. This issue undoubtedly poses big challenges to those researchers interested in gathering information to enrich their own background. Consequently, the development of a system able to automatically organize and provide relevance to scientific papers should play a crucial role to address the aforementioned problem. In this paper, the Webelance framework is presented. It makes use of a lexicon and Machine Learning techniques to accomplish these tasks. It has been built by using specific metrics for the scientific domain to measure the relative importance of papers. Several experiments using more than 50,000 articles focused on the medicine domain have been addressed to illustrate the viability of the proposal. The obtained results both confirm the usability of the system and its good performance.