Abstract:

This paper proposes a method for conducting literature reviews using digital tools such as Parsifal, Bibliometrix, and GPT-4 to support research and analysis activities. The proposed approach highlights the potential of GPT-4 in extracting information from scientific articles, accelerating the analysis process, and providing rapid results during the literature review. The method was applied to identify artificial intelligence (AI) techniques for detecting geological features in seismic data. The review process was structured in three main steps: planning, execution, and reporting. During the execution step, GPT-4 combined with a RAG (Retrieval-Augmented Generation) proved to be efficient in extracting information from the articles, answering research questions, accelerating the analysis, and contributing to the systematization of results. As a result, it was possible to map the main AI techniques applied to seismic data for identifying geological features, demonstrating the effectiveness of the proposed method both in executing the review and in organizing the knowledge obtained. This paper contributes with a systematic and innovative sequence of activities for conducting literature reviews, focusing on technological and scientific fields with a growing volume of publications.

Referência:

SILVA, Cristina Maria Ferreira da; HELENO, Aline Fernandes; SALES, Elisa Morande; SAAR, Hugo Ferreira CUSTÓDIO, Gustavo Torres; AOYAGI, Thiago Yuji; SILLES, Felipe Silva; SILVA, Celso Luciano Alves da; GUIRELI NETO, Leonides;  GUERRA, Ney Ferreira de Souza; GANDOLFO, Otavio Coaracy Brasil; GAMBA, Carlos Tadeu de Carvalho; VERISSIMO, Denis Bruno; RUBO, Rafael Andrello. Identifying AI techniques for detecting geological features in seismic data: literature review method with GPT-4. In: EUROPEAN ASSOCIATION FOR GEOSCIENTISTS AND ENGINEERS ANNUAL CONFERENCE – EAGE & EXHIBITION, 86., Toulouse, France. Proceedings… 5 p.

Documento com acesso restrito. Logar na BiblioInfo, Biblioteca GITEB/IPT, para acessar o trabalho em PDF:

https://escriba.ipt.br/pdf_restrito/179667.pdf