Automated Document Analysis for Physical Flow Account of Plastic Material
Articles and reports: 11-522-X202500100005Description: The Physical Flow Account for Plastic Material (PFAPM) aims to enhance environmental-economic analysis by tracking plastic material flows within the Canadian economy. To help streamline this complex process, the project leveraged advanced natural language processing (NLP) such as large language models (LLM) techniques to automate sector classification and summarize the impact of COVID-19 from company reports. By integrating machine learning models and retrieval-augmented generation (RAG) methods, the manual workload was significantly reduced, improving data analysis efficiency, and leading to higher quality insights.
Issue Number: 2025001Author(s): Ogunnoiki, Oladayo; Istrate, AlexandreMain Product:Statistics Canada International Symposium Series: Proceedings
