Symbl's topic modeling capabilities go beyond traditional keyword-based analysis, providing advanced graph-based modeling that analyzes the internal conversation structure of how concepts are interrelated in a discussion. Unlike traditional models, Symbl's approach is designed for spoken conversations and can detect context switches and extract relevant topics in real-time. The model uses a combination of natural language processing and deep learning to identify key concepts, assign contextual scores, and provide a framework for users to calibrate the relationship among concepts. Symbl's advanced topic modeling capabilities include features such as parent topics, scope, flat and hierarchical topics, abstract topics, and custom topics, allowing for customization and biasing of the analysis. This enables more sophisticated systems that can handle unstructured conversation data and provide a better understanding of the conversation's goals and key concepts.