diff --git a/src/dictionaries/morphosyntactic/ner_file.py b/src/dictionaries/morphosyntactic/ner_file.py index ef051f5dbed2548ac264fbce1efce07c18b288c6..e2c323adf2c33c9fee7811be0bc10a97a66b2669 100644 --- a/src/dictionaries/morphosyntactic/ner_file.py +++ b/src/dictionaries/morphosyntactic/ner_file.py @@ -1,6 +1,7 @@ """Module responsible for Morphosyntactic dict that uses a tsv file with NER tags.""" import random +import string from collections import defaultdict from typing import List, Optional, Type, Dict @@ -115,7 +116,10 @@ class NERFileMorphosyntacticDictionary(MorphosyntacticDictionary): word = self._dictionary[entry_type][morpho_tag][lemma] elif morpho_tag == "ign": # unknown form - lemma = "ABC" + letters = string.ascii_lowercase + size = random.randint(3, 5) + lemma = "".join(random.sample( + list(letters), size)).upper() word = lemma else: morpho_tag = random.choice(