Transformative Advances in PDF Information Management: Leveraging Transformer Models for Contextual Query-Answering

Abstract

In the information-driven world of today, efficiently extracting and comprehending data from a variety of document formats, especially scanned images and PDFs, is a major challenge. Standard methods often fall short of extracting contextually aware insights from these documents. In order to provide contextual query-answering, this paper introduces a transformative approach to PDF information management by combining OCR technology with a transformer model, namely GPT-2. The system uses OCR to turn scanned documents and images into text, and the GPT-2 is used to contextually translate and understand the text. By using this innovative technique, query-based information retrieval becomes much more accurate and efficient, giving users precise, context-aware responses to their questions.The proposed method offers improved data accessibility and optimised information management, setting a new benchmark in document analysis across different sectors. This work contributes to the advancement of document interaction and understanding by combining OCR technology with sophisticated transformer models, paving the way for future developments in natural processing and document processing. Key Words: Machine Learning, Optical Character Recognition (OCR), Generative Pre-trained Transformer 2 (GPT-2), Natural Language Processing (NLP), FastAPI.