Extracting searchable information from a digitized document

摘要

Data extraction and automatic validation from digitized documents in non-editable formats is disclosed. Paper documents are digitized or converted into formats suitable for storage on computers or other digital devices. The digitized documents are classified into one of a plurality of document types and based on the document type, document processing rules are selected for analyzing the digitized documents to enable data extraction and automatic validation. The positions and values of the data fields in the digitized documents are obtained using machine learning techniques. The data field values are automatically validated and assigned confidence scores. Data fields with low confidence scores are flagged for manual review.