Update README.md

README.md

@@ -1,3 +1,27 @@
+Dataset for training an entailment classifier to recognize approval/disapproval of politicians. 
+Documents are Tweets from Kawintiranon (2022), the MTSD dataset, as well as Tweets and sentences taken weekly newsletters for select politicians from the 115th, 116th, and 117th congress.
+Documents are triple coded -- once from the original compilers of the dataset, once from GPT-4, and a third time to adjudicate discrepancies between the two.
+
+Twitter handles from politicians in the dataset have been replaced by their name. Be aware that "rt @realdonaldtrump You're a liar!" has been replaced with "rt trump You're a liar!",
+And means the author is retweeting Trump calling someone else a liar, not the author calling Trump a liar.
+
+Stance:  
+  -1: Against  
+   0: Neutral  
+   1: Support  
+ 
+Label:  
+  0: Entail  
+  1: Not Entail  
+
+In addition to 1,000 documents about targets in the training set, the test set contains documents about 6 politicians not included in the training or validation data:
+  - Ted Cruz
+  - Hakeem Jeffries
+  - Madison Cawthorn
+  - Alexandria Ocasio-Cortez
+  - Mitt Romney
+  - Kyrsten Sinema
+
 ---
 license: mit
 configs: