--- a
+++ b/src/preprocessing/split_sentences.py
@@ -0,0 +1,46 @@
+# coding: utf-8
+
+# Base Dependencies
+# -----------------
+from tqdm import tqdm
+
+# Local Dependencies
+# ------------------
+from models.document import Document
+from utils import doc_id_n2c2, files_n2c2
+
+
+def split_sentences(dataset: str):
+    if dataset == "n2c2":
+        split_sentences_n2c2()
+    elif dataset == "ddi":
+        print("Splitting ddi dataset in sentences not necessary. All done \n")
+    else:
+        raise ValueError("unsupported dataset '{}'".format(dataset))
+
+
+def split_sentences_n2c2():
+    """
+    Splits the n2c2 records into sentences and cleans the text. The split documents
+    are stored in the same folder as the original text files in json format.
+    """
+
+    print("\nSplitting n2c2 dataset in sentences: ")
+
+    dataset = files_n2c2()
+
+    for split, files in dataset.items():
+        print("\n", split, ": ")
+
+        for basefile in tqdm(files):
+            # read text file
+            with open(basefile + ".txt", "r", encoding="utf-8") as fin:
+                text = fin.read()
+
+            # create document with sentence split
+            document = Document.from_txt_n2c2(doc_id_n2c2(basefile), text)
+
+            # assert len(text) == len(document.to_txt())
+
+            with open(basefile + ".json", "w+", encoding="utf-8") as fout:
+                fout.write(document.toJSON())