Description
ner_conll_xlnet_base_cased 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_xlnet_base_cased model is trained withxlnet_base_cased 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 = XlnetEmbeddings\
.pretrained('xlnet_base_cased', 'en')\
.setInputCols(["token", "document"])\
.setOutputCol("embeddings")
ner_model = NerDLModel.pretrained('ner_conll_xlnet_base_cased', '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 = XlnetEmbeddings.pretrained("xlnet_base_cased", "en")
.setInputCols("document", "token")
.setOutputCol("embeddings")
val ner_model = NerDLModel.pretrained("ner_conll_xlnet_base_cased", "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_xlnet_base_cased').predict(text, output_level='token')
Model Information
| Model Name: | ner_conll_xlnet_base_cased |
| Type: | ner |
| Compatibility: | Spark NLP 3.2.2+ |
| 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 0.87 0.89 0.88 1668
I-ORG 0.81 0.82 0.81 835
I-MISC 0.51 0.61 0.55 216
I-LOC 0.71 0.82 0.76 257
I-PER 0.93 0.99 0.96 1156
B-MISC 0.76 0.76 0.76 702
B-ORG 0.86 0.80 0.83 1661
B-PER 0.90 0.94 0.92 1617
micro avg 0.85 0.86 0.86 8112
macro avg 0.79 0.83 0.81 8112
weighted avg 0.85 0.86 0.86 8112
processed 46435 tokens with 5648 phrases; found: 5630 phrases; correct: 4782.
accuracy: 86.49%; (non-O)
accuracy: 96.74%; precision: 84.94%; recall: 84.67%; FB1: 84.80
LOC: precision: 86.43%; recall: 88.19%; FB1: 87.30 1702
MISC: precision: 72.02%; recall: 72.22%; FB1: 72.12 704
ORG: precision: 84.40%; recall: 77.84%; FB1: 80.99 1532
PER: precision: 89.30%; recall: 93.44%; FB1: 91.33 1692
Dev:
precision recall f1-score support
B-LOC 0.91 0.93 0.92 1837
I-ORG 0.89 0.80 0.84 751
I-MISC 0.82 0.68 0.74 346
I-LOC 0.76 0.87 0.81 257
I-PER 0.95 0.97 0.96 1307
B-MISC 0.86 0.81 0.84 922
B-ORG 0.92 0.83 0.87 1341
B-PER 0.92 0.96 0.94 1842
micro avg 0.90 0.89 0.90 8603
macro avg 0.88 0.86 0.87 8603
weighted avg 0.90 0.89 0.90 8603
processed 51362 tokens with 5942 phrases; found: 5893 phrases; correct: 5254.
accuracy: 89.14%; (non-O)
accuracy: 97.83%; precision: 89.16%; recall: 88.42%; FB1: 88.79
LOC: precision: 89.97%; recall: 92.27%; FB1: 91.10 1884
MISC: precision: 82.49%; recall: 78.20%; FB1: 80.29 874
ORG: precision: 88.99%; recall: 80.76%; FB1: 84.68 1217
PER: precision: 91.50%; recall: 95.28%; FB1: 93.35 1918