Add NER map option to convert CLI (#4763)

Instead of a hard-coded NER tag simplification function that was only
intended for NorNE, map NER tags in CoNLL-U converter using a dict
provided as JSON as a command-line option.

Map NER entity types or new tag or to "" for 'O', e.g.:

```
{"PER": "PERSON", "BAD": ""}

=>

B-PER -> B-PERSON
B-BAD -> O
```

spacy/cli/convert.py

@@ -39,6 +39,7 @@ FILE_TYPES_STDOUT = ("json", "jsonl")
     converter=("Converter: {}".format(tuple(CONVERTERS.keys())), "option", "c", str),
     lang=("Language (if tokenizer required)", "option", "l", str),
     morphology=("Enable appending morphology to tags", "flag", "m", bool),
+    ner_map_path=("NER tag mapping (as JSON-encoded dict of entity types)", "option", "N", Path),
 )
 def convert(
     input_file,
@@ -49,6 +50,7 @@ def convert(
     model=None,
     morphology=False,
     converter="auto",
+    ner_map_path=None,
     lang=None,
 ):
     """
@@ -94,6 +96,9 @@ def convert(
             )
     if converter not in CONVERTERS:
         msg.fail("Can't find converter for {}".format(converter), exits=1)
+    ner_map = None
+    if ner_map_path is not None:
+        ner_map = srsly.read_json(ner_map_path)
     # Use converter function to convert data
     func = CONVERTERS[converter]
     data = func(
@@ -104,6 +109,7 @@ def convert(
         lang=lang,
         model=model,
         no_print=no_print,
+        ner_map=ner_map,
     )
     if output_dir != "-":
         # Export data to a file

spacy/cli/converters/conllu2json.py

@@ -7,7 +7,8 @@ from spacy.gold import Example
 from ...gold import iob_to_biluo
 
 
-def conllu2json(input_data, n_sents=10, use_morphology=False, lang=None, **_):
+def conllu2json(input_data, n_sents=10, use_morphology=False, lang=None,
+                ner_map=None, **_):
     """
     Convert conllu files into JSON format for use with train cli.
     use_morphology parameter enables appending morphology to tags, which is
@@ -33,7 +34,8 @@ def conllu2json(input_data, n_sents=10, use_morphology=False, lang=None, **_):
             checked_for_ner = True
         raw += example.text
         sentences.append(generate_sentence(example.token_annotation,
-                has_ner_tags, MISC_NER_PATTERN))
+                has_ner_tags, MISC_NER_PATTERN,
+                ner_map=ner_map))
         # Real-sized documents could be extracted using the comments on the
         # conllu document
         if len(sentences) % n_sents == 0:
@@ -111,8 +113,12 @@ def read_conllx(input_data, use_morphology=False, n=0):
                 break
 
 
-def simplify_tags(iob, tag_pattern):
+def extract_tags(iob, tag_pattern, ner_map=None):
     """
+    Extract tag from MISC column according to `tag_pattern` and map to final
+    entity type with `ner_map` if mapping present.
+
+    For NorNE:
     Simplify tags obtained from the dataset in order to follow Wikipedia
     scheme (PER, LOC, ORG, MISC). 'PER', 'LOC' and 'ORG' keep their tags, while
     'GPE_LOC' is simplified to 'LOC', 'GPE_ORG' to 'ORG' and all remaining tags to
@@ -126,22 +132,24 @@ def simplify_tags(iob, tag_pattern):
             prefix = tag_match.group(2)
             suffix = tag_match.group(3)
             if prefix and suffix:
-                if suffix == "GPE_LOC":
-                    suffix = "LOC"
-                elif suffix == "GPE_ORG":
-                    suffix = "ORG"
-                elif suffix != "PER" and suffix != "LOC" and suffix != "ORG":
-                    suffix = "MISC"
+                new_tag = prefix + "-" + suffix
+                if ner_map:
+                    suffix = ner_map.get(suffix, suffix)
+                    if suffix == "":
+                        new_tag = "O"
+                    else:
                         new_tag = prefix + "-" + suffix
         new_iob.append(new_tag)
     return new_iob
 
 
-def generate_sentence(token_annotation, has_ner_tags, tag_pattern):
+def generate_sentence(token_annotation, has_ner_tags, tag_pattern,
+                      ner_map=None):
     sentence = {}
     tokens = []
     if has_ner_tags:
-        iob = simplify_tags(token_annotation.entities, tag_pattern)
+        iob = extract_tags(token_annotation.entities, tag_pattern,
+                            ner_map=ner_map)
         biluo = iob_to_biluo(iob)
     for i, id in enumerate(token_annotation.ids):
         token = {}

spacy/tests/test_cli.py

@@ -9,7 +9,7 @@ from spacy.cli.pretrain import make_docs
 
 
 def test_cli_converters_conllu2json():
-    # https://raw.githubusercontent.com/ohenrik/nb_news_ud_sm/master/original_data/no-ud-dev-ner.conllu
+    # from NorNE: https://github.com/ltgoslo/norne/blob/3d23274965f513f23aa48455b28b1878dad23c05/ud/nob/no_bokmaal-ud-dev.conllu
     lines = [
         "1\tDommer\tdommer\tNOUN\t_\tDefinite=Ind|Gender=Masc|Number=Sing\t2\tappos\t_\tO",
         "2\tFinn\tFinn\tPROPN\t_\tGender=Masc\t4\tnsubj\t_\tB-PER",
@@ -32,17 +32,16 @@ def test_cli_converters_conllu2json():
     assert [t["ner"] for t in tokens] == ["O", "B-PER", "L-PER", "O"]
 
 
-def test_cli_converters_conllu2json():
-    # https://raw.githubusercontent.com/ohenrik/nb_news_ud_sm/master/original_data/no-ud-dev-ner.conllu
+def test_cli_converters_conllu2json_name_ner_map():
     lines = [
         "1\tDommer\tdommer\tNOUN\t_\tDefinite=Ind|Gender=Masc|Number=Sing\t2\tappos\t_\tname=O",
         "2\tFinn\tFinn\tPROPN\t_\tGender=Masc\t4\tnsubj\t_\tSpaceAfter=No|name=B-PER",
         "3\tEilertsen\tEilertsen\tPROPN\t_\t_\t2\tname\t_\tname=I-PER",
         "4\tavstår\tavstå\tVERB\t_\tMood=Ind|Tense=Pres|VerbForm=Fin\t0\troot\t_\tSpaceAfter=No|name=O",
-        "5\t.\t$.\tPUNCT\t_\t_\t4\tpunct\t_\tname=O",
+        "5\t.\t$.\tPUNCT\t_\t_\t4\tpunct\t_\tname=B-BAD",
     ]
     input_data = "\n".join(lines)
-    converted = conllu2json(input_data, n_sents=1)
+    converted = conllu2json(input_data, n_sents=1, ner_map={"PER": "PERSON", "BAD": ""})
     assert len(converted) == 1
     assert converted[0]["id"] == 0
     assert len(converted[0]["paragraphs"]) == 1
@@ -55,7 +54,7 @@ def test_cli_converters_conllu2json():
     assert [t["tag"] for t in tokens] == ["NOUN", "PROPN", "PROPN", "VERB", "PUNCT"]
     assert [t["head"] for t in tokens] == [1, 2, -1, 0, -1]
     assert [t["dep"] for t in tokens] == ["appos", "nsubj", "name", "ROOT", "punct"]
-    assert [t["ner"] for t in tokens] == ["O", "B-PER", "L-PER", "O", "O"]
+    assert [t["ner"] for t in tokens] == ["O", "B-PERSON", "L-PERSON", "O", "O"]
 
 
 def test_cli_converters_iob2json():