Desarrollo de sistema basado en minería de datos para la búsqueda y visualización de redes de investigadores con filiación en instituciones ecuatorianas y sus areas académicas.

Abstract

The quality of scientific research is based on collaboration among experts. In Ecuador, an increase in scientific production has been observed, which has generated the need for an efficient computer tool to form networks, search for collaborators, and analyze research topics. In response to this, a web application has been developed that utilizes data mining models for searching and visualizing networks of researchers affiliated with Ecuadorian institutions and their academic areas. The project was implemented following Scrum and CRISP-DM methodologies. To carry out the entire data mining process, a series of Python scripts were developed. The data was extracted from Scopus through the set of APIs offered by Elsevier. The generated models were TF-IDF, which were employed to obtain the best results in the searches performed within the application. The architecture of the application consists of three layers: the Neo4j graph-oriented database as the data layer, Flask as the backend that houses the business logic, and Angular as the frontend, utilizing important libraries such as Bootstrap for design and D3.js for creating dynamic and interactive graphs. Ultimately, as the final product, the application offers three types of search: author search, relevant authors search by topic, and relevant articles search by topic.