Android comment classification using BERT

• Autores: Mansilla Ancco, Susana Rosa Elizabeth, Pérez Treviños, Marcelo Antony
• Formato: artículo
• Fecha de Publicación: 2024
• Institución: Universidad La Salle
• Repositorio: Revistas - Universidad La Salle
• Lenguaje: español
• OAI Identifier: oai:ojs.revistas.ulasalle.edu.pe:article/120
• Enlace del recurso: https://revistas.ulasalle.edu.pe/innosoft/article/view/120; https://doi.org/10.48168/innosoft.s15.a120; https://purl.org/42411/s15/a120; https://n2t.net/ark:/42411/s15/a120
• Nivel de acceso: acceso abierto
• Materia: Topic classification; text classification; natural language processing; BERT; clasificación de tópicos; clasificación de texto; procesamiento de lenguaje natural

This project focuses on developing an NLP-based text analysis tool to evaluate Android app user feedback, specifically collected from F-Droid. The lack of an automated solution to analyze and understand these opinions, classifying them into specific topics, motivates research. The goal is to provide developers, users, and data analysts with a detailed view of user preferences and perceptions. Using data sets in English between 2014 and 2017, the proposal is implemented in Python with the Pandas library. The BERT model is used for classification, with a specific focus on the comparison of different models. The graphical interface is built in Visual Studio, allowing users to enter comments and obtain topic rankings, along with word cloud visualizations.