Create a framework for mining insights from unstructured text data. The system should identify sentiment patterns, extract key entities, and detect anomalies. All text in it should be processed with appropriate tokenization, normalization, and feature extraction techniques.