The internet contains vast amounts of text-based information across various domains, such as commercial documents, medical records, scientific research, engineering tests, and events affecting urban and natural environments. Extracting knowledge from these texts requires a deep understanding of natural language nuances and accurately representing content while preserving essential information. This process enables effective knowledge extraction, inference, and discovery. This paper proposes a critical study of state-of-the-art contributions exploring the complexities and emerging trends in representing, querying, and analysing content extracted from textual data. This study’s hypothesis states that graph-based representations can be particularly effective when annotated with sophisticated querying and analytics techniques. This hypothesis is discussed through the lenses of contributions in linguistics, natural language processing, graph theory, databases, and artificial intelligence.