The community member is looking for a PDF search system that can highlight the exact location of a specific number in a document, such as the FY23 Revenue in a 10K report. Another community member suggests checking out llamaIndex and the fuzzy llamapack, which may provide the desired functionality. However, one community member has tried secinsights.ai and found that it does not reliably highlight or extract data from tables. The community members also discuss the potential of using LlamaParse and fuzzycitation to work with table data more effectively.
Has anyone been able to create a PDF search system that highlights the exact location of a number in a document? For example, if I search for FY23 Revenue in a 10K, it would take me straight to that number and show a bounding box around that specific number in a table.
fuzzy llamapack looks interesting. Thanks! Do you know if there are any successful examples online of this being implemented on table data specifically?
If you are working with table data, I would recommend you to try LlamaParse and then use fuzzycitation, This combination can make parsing + retrieving much better IMO
