Topic Segmentation of Semi-Structured and Unstructured Conversational Datasets using Language Models