INFORMATION RETRIEVAL ON THE WEB: an experience with the ChatGPT artificial intelligence language model

• Autor(a) principal: Silva, Patrícia Nascimento
• Data de Publicação: 2023
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Informação em Pauta
• Texto Completo: http://www.periodicos.ufc.br/informacaoempauta/article/view/83566
• topic: Information Retrieval; Artificial Intelligence; ChatGPT; Virtual Assistant; Open Data; Recuperação de Informação; Inteligência Artificial; ChatGPT; Assistente Virtual; Dados Abertos; Recuperação de Informação; Inteligência Artificial; ChatGPT

Information retrieval is a recent term, created in 1951, but its objective has been desired for millennia in society. With the evolution of information technology, in the 1970s, and especially the creation of the Web, in 1989, the term had its meaning enhanced in new scenarios. The classic information retrieval models obtained contributions from computing techniques to optimize results and dialogue with users, as proposed by the ChatGPT artificial intelligence language model, launched in 2022. The objective of this report was to analyze information retrieval on the Web with ChatGPT. For this purpose, a domain was selected: open data, which is originally diverse, and exploratory searches were carried out to identify the information and digital objects recovered, in February 2023. The model used in ChatGPT demonstrated a superior capacity to web search engines, by retrieving information and resources from different sources and presenting them in an organized and cohesive text. However, the lack of reference to the sources used and clarity on ethical issues are problems seen in the free version. It is expected that, with adjustments, ChatGPT will recover several resources on the Web, achieving interoperability with other systems, contributing to the exchange of data in different domains.