Update README.md

README.md

@@ -1,3 +1,46 @@
 ---
-license: cc-by-nc-sa-3.0
+annotations_creators:
+- found
+language:
+- cs
+language_creators:
+- found
+license:
+- cc-by-nc-sa-3.0
+multilinguality:
+- monolingual
+pretty_name: Czech Facebook comments
+size_categories:
+- 10K<n<100K
+source_datasets:
+- original
+tags: []
+task_categories:
+- text-classification
+task_ids:
+- sentiment-classification
 ---
+
+
+# Dataset Card for Czech Facebook comments
+
+
+## Dataset Description
+
+The dataset contains user comments from Facebook. Each comment contains text, sentiment (positive/negative/neutral).
+The dataset has in total (train+validation+test) 6,600 reviews. The data is balanced.
+
+
+## Dataset Features
+
+Each sample contains:
+- `comment_id`: unique string identifier of the comment.
+- `sentiment_str`: string representation of the rating - "pozitivní" / "neutrální" / "negativní"
+- `sentiment_int`: integer representation of the rating (1=positive, 0=neutral, -1=negative)
+- `comment`: the string of the comment
+
+
+## Dataset Source
+
+The data is a processed adaptation of [Facebook CZ Corpus](https://liks.fav.zcu.cz/sentiment/).
+This adaptation is label-balanced.