changes v5

Author: ArnaudTadic

index.md

@@ -153,7 +153,13 @@ Next, we built a linguistic fingerprint for each domain.
 
 For every LIWC feature, we computed a delta between hostile and non-hostile posts:
 
-signature(feature) = mean(feature | hostile) − mean(feature | non-hostile)
+<div align="center">
+
+$$
+\text{signature}(\text{feature}) = \operatorname{mean}(\text{feature}\mid \text{hostile}) - \operatorname{mean}(\text{feature}\mid \text{non-hostile})
+$$
+
+</div>
 
 This produces one “signature vector” for Politics and one for Sports.  
 We then compared the two vectors with **Pearson** (shape similarity) and **Spearman** (rank similarity) correlations.