Business Problem
Many workflows require comparing multiple PDF reports to identify common and variable text, so that future reports can be generated from standardized templates. This can take months of effort and duration, prolonging new client onboarding or consuming valuable FTE’s engaged in tedious error-prone activities
AI/ML Solution
Human-in-the-middle workflow application to monitor AI processing of documents including PDF text extraction, cross-book paragraph clustering, and multi-text differencing comparison.
Text is extracted from the PDF homing in on the section of interest, (e.g. “Notes to Financials”), excluding extraneous text, (e.g. “headers, footers, page numbers, tables”) via sophisticated rules .
Across the corpus, similar paragraphs are grouped together representing similar concepts and wording .
Within a cluster of multiple related ideas, fragments are identified representing common or variable strings.
