Explainable Recommendation via Interpretable Feature Mapping and Evaluation of Explainability