Text this: Noise reduction in frequent terms sets clustering