Update README.md

README.md

@@ -16,8 +16,4 @@ Furthermore:
 * non-breaking spaces: every \xa0 (both in articles and in summaries) was replaced by a space
 * bi-gram_overlap_fraction between summary and original text <= 0.63, meaning that all summaries in the dataset are on the abstractive side
 
-Then:
-* 5k random rows were selected and kept in the dataset. All other rows were removed.
-* rows in which the article (column "text") contained a space followed by a point, another space and then a lowercase letter (i.e. a string matching this regex: " \\. [a-z]") were removed, because a qualitative analysis had shown that these articles often had full stops in between sentences
-
-This lead to a dataset with 4930 rows.
+Then, 5k random rows were selected and kept in the dataset. All other rows were removed.