Reg Document Extraction

I Trained AI skills with 20 Documents(resumes) under both Highly Structured and Unstructured categories, for this accuracy rate for Highly Structured is 40% and for Unstructure it is 47%. Is there any other ways to extract document for my OCR activity