Live Shine - Uma ferramenta para suporte à avaliação de impacto de eventos científicos em computação

Abstract

A common concern among researchers is that the results of their research are published in venues of impact in the scientific community. In general, the impact indices are obtained through metrics based on the number of citations that your articles receive. Institutions such as the SCImago and Thomson Reuters provide precise impact indices for major international journals. While this is sufficient for most areas, in Computer Science conferences and other scientific events are also important as publishing venues. However, currently, there is no solution that is universally accepted to obtain accurate indices on conferences, because the tools most commonly used for this purpose have differences between the indices generated for the same conference and year. In this dissertation, we propose a tool called Live SHINE, whose goal is to generate high-precision impact indices of Computer Science Conferences from data provided by Google Scholar. Our tool uses a method based on machine learning techniques that automatically filters the metadata provided by Google Scholar, and considers in the calculation of impact indices only citation data from articles that truly belong to the conference. Our experiments show that our method is effective, achieving an average F1 metric above 0.9 for 30 analyzed conferences. In addition, we also developed a new distributed and collaborative strategy of collecting citations, in which the queries sent to Google Scholar to retrieve the updated values of article citations are triggered by the user interface, avoiding problems such as network overload, delay in update citations and frequent blocking by Google Scholar. Thus, this strategy makes the community of users collaborate to keep updated the data citations for the benefit of all.