Update single.py

single.py

@@ -98,9 +98,9 @@ def body():
             **Legend**
 
             - **AOPC Comprehensiveness** (aopc_compr) measures *comprehensiveness*, i.e., if the
-            explanation captures
-            - **AOPC Sufficiency** (aopc_suff) measures *sufficiency*, i.e., 
-            - **Leave-On-Out TAU Correlation** (taucorr_loo) measures
+            explanation captures all the tokens needed to make the prediction. It is computed as the drop in the model probability if the relevant tokens of the explanations are removed. The higher the comprehensiveness, the more faithful is the explanation.
+            - **AOPC Sufficiency** (aopc_suff) measures *sufficiency*, i.e., if the relevant tokens in the explanation are sufficient to make the prediction. It is computed as the drop in the model probability if only the relevant tokens of the explanations are considered. The lower the sufficiency, the more faithful is the explanation since there is less change in the model prediction.
+            - **Leave-On-Out TAU Correlation** (taucorr_loo) measures the Kendall rank correlation coefficient τ between the explanation and leave-one-out importances. The latter are computed by omittig individual input tokens and measuring the variation on the model prediction. The closer the τ correlation is to 1, the more faithful is the explanation. 
             
             See the paper for details.
             """