The problem statement:
The documents contain scanned PDFs with tables, references, and other content. Queries may pertain to either references or tables. Some queries, such as “Total quantity of ______ purchased in 2017,” require calculations rather than retrieving direct answers from the table.
Need to create a RAG chatbot for these docs
This is an active area of research in the AI realm. It’s not an easy problem, and it has no fits-all-cases solutions yet.
1 Like