Dr Jajati Mandal J.Mandal2@salford.ac.uk
University Fellow
Dr Jajati Mandal J.Mandal2@salford.ac.uk
University Fellow
Prof Mike Wood M.D.Wood@salford.ac.uk
Associate Dean Research & Innovation
Arsenic (As) contamination in soil and water is a significant environmental issue that poses severe risks to human health. Despite extensive research, addressing this challenge remains complex due to its diverse source and behavior in different environmental matrices. This study introduces an innovative approach utilizing Retrieval-Augmented Generation (RAG) to enhance accessibility and utility of scientific knowledge related to As contamination. By leveraging advanced natural language processing (NLP) technologies an attempt has been made to develop a generative pretrained transformer (GPT) based tool to assist researchers, policymakers and practitioners. The study involved collecting 157 scientific articles in PDF format and extracting their textual content and saving the text as individual files. The extracted text was converted into numerical representations using sentence models (M1 and M2) followed by saving as NumPy arrays and JavaScript object notation (JSON) files for easy mapping. A similarity search index was constructed using the embeddings for efficient similarity search. OpenAI’s models (O1 and O2) were used to generate contextually relevant responses to various queries. The responses were assessed using Bilingual Evaluation Understudy (BLEU) and Recall-Oriented Understudy for Gisting Evaluation (ROUGE) evaluation metrics. The RAG2_O2, model utilizing the M2 embeddings and O2 query model, demonstrating superior performance in terms of response generation.
Mandal, J., & Wood, M. (2024, October). A Retrieval-Augmented Generation Based Tool to address Arsenic Contamination in Agricultural System. Presented at 9th International Congress Arsenic in the Environment (As2024), Bhubaneswar, India
Presentation Conference Type | Keynote |
---|---|
Conference Name | 9th International Congress Arsenic in the Environment (As2024) |
Conference Location | Bhubaneswar, India |
Start Date | Oct 20, 2024 |
End Date | Oct 24, 2024 |
Deposit Date | Nov 12, 2024 |
Related Public URLs | https://as2024.kiit.ac.in/ |
About USIR
Administrator e-mail: library-research@salford.ac.uk
This application uses the following open-source libraries:
Apache License Version 2.0 (http://www.apache.org/licenses/)
Apache License Version 2.0 (http://www.apache.org/licenses/)
SIL OFL 1.1 (http://scripts.sil.org/OFL)
MIT License (http://opensource.org/licenses/mit-license.html)
CC BY 3.0 ( http://creativecommons.org/licenses/by/3.0/)
Powered by Worktribe © 2025
Advanced Search