Improving Data Quality and Data Mining Using Multiple, Noisy Labelers