Utilizando clusterização para identificar padrões em publicações de artigo na América Latina

Abstract

With the technology advancement and easy access to information, the number of publications scientific articles have grown exponentially. However, due to the massive amount of information, identifying relevant patterns has become a challenge. This work proposes an approach clustering using the MiniBatch K-Means method appplied in data from articles published in the last 20 years in Latin America, as well as an exploration of its results. The methodology applied resulted in the creation of 50 distinct groups. Each of these groups is composed of articles related to specific topics, some examples is the existence of a group containing articles only related to "gene expression and metabolism in proteins and cells"and another group related to "polynomial- time optimization and combinatorial complexity". The results obtained by clustering method demonstrated algorithm’s ability to identify and efficiently organize scientific articles based on their abstract and keywords.