A Case Based Deep Neural Network Interpretability Framework and Its User Study