The exponential growth of digital information has created significant challenges in extracting meaningful
knowledge from large volumes of unstructured textual data. Organizations, researchers, government
agencies, healthcare institutions, and enterprises continuously generate documents including research
articles, reports, emails, legal records, medical notes, news articles, and technical documentation.
Manual analysis of these documents is time-consuming, labor-intensive, and often impractical for
real-time decision-making. Recent advances in Generative Artificial Intelligence and Large Language
Models (LLMs), particularly Generative Pre-trained Transformers (GPT), have significantly improved
contextual language understanding, enabling intelligent extraction of relevant information and
generation of coherent summaries. Unlike traditional extractive summarization techniques, GPT-based
contextual summarization captures semantic relationships, contextual dependencies, and document intent
while producing concise and human-readable summaries. This paper proposes OCE (GPT Contextual
Text Extraction and Summarization), an intelligent framework that combines contextual information
extraction, semantic representation learning, transformer-based language modeling, and abstractive
summarization for efficient document understanding. The proposed framework performs document
preprocessing, contextual embedding generation, key information extraction, semantic ranking, GPTbased
summarization, and quality evaluation. By utilizing contextual understanding instead of simple
keyword matching, OCE produces accurate summaries while preserving important semantic information
contained within lengthy documents. Experimental evaluation was conducted using benchmark datasets
comprising research articles, news reports, legal documents, healthcare records, and technical
documents. Comparative analysis demonstrates that the proposed OCE framework achieves superior
contextual understanding, summarization quality, semantic relevance, and computational efficiency
compared with conventional extractive and transformer-based summarization methods. The proposed
architecture provides an intelligent document understanding solution suitable for large-scale knowledge
management, information retrieval, and decision support applications.
