In the academy, there is an untold amount of papers, and yet more new ones are presented in different fields of science. It is common in academia influence young students, beginners in the field of research, choose the source articles better known (by reputed universities or researchers) or more cited. However, recently published papers did not have enough time to consolidate itself in the scientific community, thereby having few citations, therefore hindering the search for an interesting paper. To prevent such influence and work around the problem mentioned, in this paper, we discuss a method for classification of articles using their abstracts applied in algorithms to Natural Language Processing, seeking to provide to the beginning student, an interesting read addressing the topic searched. [ABSTRACT FROM AUTHOR]