AnaliTexGra: A novel visual application for academic collaboration prediction based on standard machine learning techniques and text mining

Large universities comprising multiple faculties and research institutes often face challenges in fostering collaboration among academics, primarily due to a lack of awareness regarding the research interests and publications of colleagues across different departments. Consequently, a system that facilitates the discovery of related or complementary research lines and scholarly outputs is needed, enabling more informed decisions for future interdisciplinary projects and enhancing overall academic productivity.This work presents a visual application for link prediction and text mining, designed with a strong focus on usability and deployed in a production-ready environment. It is intended to be easily accessible to the academic community, with the primary goal of fostering collaboration among researchers and supporting greater research productivity, rather than pushing the boundaries of the state of the art.This document presents the design, development, and functionality of a Web application that performs descriptive queries and predicts collaborations between academics. The approach can be generalized to other types of social networks. Various prediction models were generated and tested using statistical link heuristics with six numbers of common neighbors (NCN) and eight Non-NCN methods to train eight Supervised Classification models. The graph design, application design, feature engineering, model validation, and the results obtained are presented. We have successfully developed and deployed a password-protected application designed exclusively for use by researchers at our university. The system fulfills all objectives specified in the initial requirements, including the ability to load both custom datasets and the application's default database. It enables users to consult information on articles and authors through both graphical visualizations and textual queries. Unlike existing platforms, this application enables the identification of isolated research groups that may otherwise remain disconnected, providing university leadership, such as department heads, directors, or rectors, with actionable insights to foster integration and collaboration among scholars.