On conducting better validation studies of automatic metrics in natural language generation evaluation

Re-assign community