Description
ner_conll_xlm_roberta_base is a Named Entity Recognition (or NER) model, meaning it annotates text to find features like the names of people, places, and organizations. It was trained on the CoNLL 2003 text corpus. This NER model does not read words directly but instead reads word embeddings, which represent words as points such that more semantically similar words are closer together.ner_conll_xlm_roberta_base model is trained withxlm_roberta_base word embeddings, so be sure to use the same embeddings in the pipeline.
Predicted Entities
PER, LOC, ORG, MISC
Live Demo Open in Colab Download Copy S3 URI
How to use
document_assembler = DocumentAssembler() \
.setInputCol('text') \
.setOutputCol('document')
tokenizer = Tokenizer() \
.setInputCols(['document']) \
.setOutputCol('token')
embeddings = XlmRoBertaEmbeddings\
.pretrained('xlm_roberta_base', 'xx')\
.setInputCols(["token", "document"])\
.setOutputCol("embeddings")
ner_model = NerDLModel.pretrained('ner_conll_xlm_roberta_base', 'en') \
.setInputCols(['document', 'token', 'embeddings']) \
.setOutputCol('ner')
ner_converter = NerConverter() \
.setInputCols(['document', 'token', 'ner']) \
.setOutputCol('entities')
pipeline = Pipeline(stages=[
document_assembler,
tokenizer,
embeddings,
ner_model,
ner_converter
])
example = spark.createDataFrame([['My name is John!']]).toDF("text")
result = pipeline.fit(example).transform(example)
val document_assembler = DocumentAssembler()
.setInputCol("text")
.setOutputCol("document")
val tokenizer = Tokenizer()
.setInputCols("document")
.setOutputCol("token")
val embeddings = XlmRoBertaEmbeddings.pretrained("xlm_roberta_base", "xx")
.setInputCols("document", "token")
.setOutputCol("embeddings")
val ner_model = NerDLModel.pretrained("ner_conll_xlm_roberta_base", "en")
.setInputCols("document"', "token", "embeddings")
.setOutputCol("ner")
val ner_converter = NerConverter()
.setInputCols("document", "token", "ner")
.setOutputCol("entities")
val pipeline = new Pipeline().setStages(Array(document_assembler, tokenizer, embeddings, ner_model, ner_converter))
val example = Seq.empty["My name is John!"].toDS.toDF("text")
val result = pipeline.fit(example).transform(example)
import nlu
text = ["My name is John!"]
ner_df = nlu.load('en.ner.ner_conll_xlm_roberta_base').predict(text, output_level='token')
Model Information
| Model Name: | ner_conll_xlm_roberta_base |
| Type: | ner |
| Compatibility: | Spark NLP 3.2.0+ |
| License: | Open Source |
| Edition: | Official |
| Input Labels: | [sentence, token, embeddings] |
| Output Labels: | [ner] |
| Language: | en |
Data Source
https://www.clips.uantwerpen.be/conll2003/ner/
Benchmarking
Test:
precision recall f1-score support
B-LOC 0.93 0.92 0.92 1668
I-ORG 0.87 0.91 0.89 835
I-MISC 0.57 0.77 0.66 216
I-LOC 0.89 0.86 0.88 257
I-PER 0.97 0.98 0.98 1156
B-MISC 0.76 0.85 0.80 702
B-ORG 0.89 0.90 0.90 1661
B-PER 0.96 0.95 0.95 1617
micro avg 0.90 0.92 0.91 8112
macro avg 0.86 0.89 0.87 8112
weighted avg 0.90 0.92 0.91 8112
processed 46435 tokens with 5648 phrases; found: 5712 phrases; correct: 4988.
accuracy: 90.08%; (non-O)
accuracy: 97.45%; precision: 87.32%; recall: 88.31%; FB1: 87.82
LOC: precision: 88.37%; recall: 91.13%; FB1: 89.73 1720
MISC: precision: 77.43%; recall: 78.21%; FB1: 77.82 709
ORG: precision: 86.28%; recall: 84.41%; FB1: 85.33 1625
PER: precision: 91.50%; recall: 93.82%; FB1: 92.64 1658
Dev:
precision recall f1-score support
B-LOC 0.95 0.95 0.95 1837
I-ORG 0.91 0.90 0.91 751
I-MISC 0.88 0.85 0.86 346
I-LOC 0.90 0.91 0.91 257
I-PER 0.97 0.98 0.98 1307
B-MISC 0.91 0.88 0.90 922
B-ORG 0.91 0.89 0.90 1341
B-PER 0.93 0.96 0.95 1842
micro avg 0.93 0.93 0.93 8603
macro avg 0.92 0.92 0.92 8603
weighted avg 0.93 0.93 0.93 8603
processed 51362 tokens with 5942 phrases; found: 5947 phrases; correct: 5483.
accuracy: 93.12%; (non-O)
accuracy: 98.55%; precision: 92.20%; recall: 92.28%; FB1: 92.24
LOC: precision: 94.48%; recall: 95.05%; FB1: 94.76 1848
MISC: precision: 89.72%; recall: 86.12%; FB1: 87.88 885
ORG: precision: 90.01%; recall: 87.99%; FB1: 88.99 1311
PER: precision: 92.64%; recall: 95.71%; FB1: 94.15 1903