Advancing Hydrologic Knowledge Access through an AI-Powered Conversational Bot for CIROH DocuHub
Abel Andrés Ramírez Molina, Jiaqi Gong, Austin Schraivogel, Shenglin Li, Arpita Patel, Fabricio Joel Gutiérrez Juárez, Margulan Baizhakyp
Problems Identified (2)
Fragmented Hydrologic Knowledge: Hydrologic knowledge resources are distributed across multiple heterogeneous platforms, making timely and accurate information access difficult for researchers.
Fragmented Hydrologic Knowledge: Hydrologic knowledge resources are distributed across multiple heterogeneous platforms, making timely and accurate information access difficult for researchers.
Proposed Solutions (5)
Domain-Specific RAG Chatbot: The paper presents an AI-powered conversational bot for CIROH DocuHub using a Retrieval-Augmented Generation pipeline grounded in hydrologic documentation.
Two-Stage Semantic Retrieval: The system first searches document summaries for relevant high-level pages and then retrieves associated text chunks and neighbors to preserve local context.
Embedded Chat Interface: A floating chat interface embedded in DocuHub supports curated example queries and open-ended user exploration.
Domain-Specific RAG Chatbot: The paper presents an AI-powered conversational bot for CIROH DocuHub using a Retrieval-Augmented Generation pipeline grounded in hydrologic documentation.
Two-Stage Semantic Retrieval: The system first searches document summaries for relevant high-level pages and then retrieves associated text chunks and neighbors to preserve local context.
Results (3)
Improved Grounded Answers:
Scalable Hydrologic Knowledge Framework:
Expanded Coverage Ongoing Work:
Research Domain
Hydrologic knowledge access and environmental data management