Título: Uncovering Research Potentials: Research Areas Evolution Analysis in Scientific Articles
Autores: Lucas Cerqueira Figueiredo, Leandro A. Silva
Resumo: This paper presents a document visualization tool that captures the main topics and similarity between scientific articles. The approach is based on recent techniques of Natural Language Processing and Deep Learning, using document embeddings. The tool combines vector representations of words with visualization techniques to condense a collection of articles useful for researchers performing a literature review and make a planning of research. Document Metadata analysis provides a temporal mapping of the evolution of the collections areas. The approach employs SCIBERT to extract representations from abstracts, metrics collected from paper metadata, and dimensionality reduction techniques to provide a useful visualization for the researcher exploring the collection
Palavras-chave: information representation, natural language processing, deep learning, scientometrics
Páginas: 6
Código DOI: 10.21528/CBIC2023-127
Artigo em pdf: CBIC_2023_paper127.pdf
Arquivo BibTeX: CBIC_2023_127.bib