Logo do repositório
 
A carregar...
Miniatura
Publicação

Padrões agroecológicos históricos em Portugal Continental: uma abordagem com Inteligência Artificial e SIG

Utilize este identificador para referenciar este registo.
Nome:Descrição:Tamanho:Formato: 
Carvalho_Diogo_TM_2024.pdf3.42 MBAdobe PDF Ver/Abrir

Resumo(s)

Esta dissertação explora como a Inteligência Artificial (IA) e os Sistemas de Informação Geográfica (SIG) podem contribuir para a análise de padrões agroecológicos históricos em Portugal Continental na década de 1950. De forma inovadora, combina Reconhecimento Ótico de Caracteres (OCR) e modelos de linguagem avançados, como o ChatGPT, para melhorar a extração e análise de dados históricos. Utilizando técnicas de NLP, text mining e aprendizagem automática, analisa 163 inquéritos agrícolas para compreender especificidades regionais e criar mapas agrícolas detalhados. A estrutura da dissertação abrange cinco capítulos. O Capítulo I apresenta o enquadramento teórico, destacando a relevância do estudo no contexto das alterações climáticas e da necessidade de práticas agrícolas mais sustentáveis. No Capítulo II, examina-se o papel da IA e dos SIG no apoio às humanidades digitais, com foco nas aplicações de OCR, Spatial Data Mining e ChatGPT para a análise de textos históricos. O Capítulo III detalha os dados e a metodologia, incluindo a digitalização, tratamento cartográfico e análise textual. O Capítulo IV apresenta os resultados, destacando a produção de mapas agrícolas e a identificação de padrões agroecológicos regionais. Por fim, no Capítulo V, conclui-se com as implicações desta abordagem interdisciplinar para a agricultura sustentável. A principal inovação reside na aplicação pioneira, em Portugal, da integração de OCR e ChatGPT para análise de fontes históricas. Esta combinação permite uma extração e interpretação de dados mais eficiente e precisa, contribuindo para o avanço da investigação histórica e oferecendo novas perspetivas para o desenvolvimento agrícola. Além de reforçar o uso de tecnologias modernas em contextos históricos, esta abordagem tem potencial impacto internacional, promovendo novas formas de estudar e interpretar o passado agrícola para orientar práticas futuras.
This dissertation explores how Artificial Intelligence (AI) and Geographic Information Systems (GIS) can contribute to analyzing historical agroecological patterns in mainland Portugal during the 1950s. Innovatively, it combines Optical Character Recognition (OCR) and advanced language models, such as ChatGPT, to enhance the extraction and analysis of historical data. Using techniques such as Natural Language Processing (NLP), text mining, and machine learning, the study analyzes 163 agricultural surveys to understand regional specificities and create detailed agricultural maps. The dissertation is structured into five chapters. Chapter I introduces the theoretical framework, emphasizing the study's relevance in the context of climate change and the need for more sustainable agricultural practices. Chapter II examines the role of AI and GIS in supporting digital humanities, focusing on the applications of OCR, Spatial Data Mining, and ChatGPT for analyzing historical texts. Chapter III outlines the data and methodology, including digitization, cartographic processing, and textual analysis. Chapter IV presents the results, highlighting the production of agricultural maps and the identification of regional agroecological patterns. Finally, Chapter V concludes with the implications of this interdisciplinary approach for sustainable agriculture. The primary innovation lies in the pioneering application of OCR and ChatGPT integration for historical data analysis in Portugal. This combination enables more efficient and accurate data extraction and interpretation, advancing historical research and offering new perspectives for agricultural development. Beyond enhancing the use of modern technologies in historical contexts, this approach has potential international significance, promoting novel ways of studying and interpreting agricultural history to guide future practices.

Descrição

Palavras-chave

K-means Reconhecimento Ótico de Caracteres Análise espacial Humanidades digitais Informação histórica

Contexto Educativo

Citação

Unidades organizacionais

Fascículo

Editora

Licença CC