This paper introduces a novel dual graph representation strategy for semantic extraction in Natural Language Processing (NLP). Semantic extraction, a critical component of NLP, seeks to derive meaningful representations from text. Traditional methods often fall short of capturing the nuanced interplay between semantic relationships and syntactic structures. The proposed approach integrates two complementary graph types: Word Co-Occurrence Graphs, which emphasize semantic relationships based on word proximity, and Word Sequence Graphs, which preserve syntactic flow by considering word order. This integration enables a comprehensive understanding of textual data and addresses complex NLP tasks such as semantic analysis, text summarization, and language modeling. Experimental results demonstrate the method’s effectiveness, achieving superior precision, recall, and F1 scores compared to traditional approaches. The dual graph strategy offers a scalable and versatile solution for enhancing NLP applications by providing a balanced and resource-efficient text representation.