Método para la determinación de similaridad y distancia entre investigadores a partir de algoritmos de clasificación.

The Cotopaxi Technical University is supporting and promoting the scientific research, resulting an increase of articles, books, projects, papers and other documents that need to be stored. For this reason the Research Direction has approved the implementation of a scientific platform called Ecuciencia, which aims to recompile and visualize the scientific and technological production based on Scientometric indicators. To reach this demanded requirements, the project was divided in phases, the collection of user data, comparison and classification among researchers. Starting from the real characteristics of the project, the use of computational intelligence tools was proposed, in order to generate the graphic representation of similarity and distance between researchers, which serves to make studies related to the scientific productivity of the university. So it has developed methods by applying classification algorithms like K - means, MeanShift, SpectralClustering, AgglomerativeClustering and data mining, which perform the analysis of an extensive dataset, to obtain as a result matrices of similarity and distance according to the number of publications of each user. The programming language Python was fundamental to develop the technological proposal, due to its simplicity and facility to use automatic learning libraries like Sklearn, the same one that contains modules of a lot of classification algorithms. To agilitate the development of the implemented module, the KDD methodology was used (Knowledge Discovery in Databases), which is oriented to the development of projects related to data mining. This process was chosen because it works through iterative life cycle through stages which has facilitated the advancement of technological proposal methodically. Through the implementation of classification algorithms in the Ecuciencia’s system, the representation of the similarity and distance of researchers according to their scientific production was achieved, in graphics that allow users to view information without difficulty.

• Hovedforfatter: Falconí Punguil, Diego Geovanny (author)
• Andre forfattere: Gualpa Mendoza, Jennifer Nataly (author)
• Format: bachelorThesis
• Sprog: spa
• Udgivet: 2019
• Fag: Sistemas; Minería de datos; Similitud; Algoritmos
• Online adgang: http://repositorio.utc.edu.ec/handle/27000/5698