Update README.md

README.md

@@ -5,4 +5,18 @@ language:
 - ru
 size_categories:
 - 1K<n<10K
----
+---
+
+# Dataset Description
+
+Spellcheck Benchmark is a collection of datasets dedicated to spelling correction problem for Russian language. 
+
+**GitHub** - <link>
+
+**Paper** - <link>
+
+# Dataset Summary
+
+Spellcheck Benchmark includes four datasets, each of which consists of pairs of sentences in Russian language. Each pair embodies sentence, which may contain spelling errors, and its corresponding correction. Datasets were gathered from various sources and domains including social networks, internet blogs, github commits, medical anamnesis, literature, news and jurisprudence texts. 
+
+All datasets were passed through two-stage manual labeling pipeline.