Description
This Legal NER model extracts ORG, INST, LAW, COURT, PER, LOC, MISC, ALIAS, and TICKER entities from the US SEC EDGAR documents.
Predicted Entities
ALIAS, COURT, INST, LAW, LOC, MISC, ORG, PER, TICKER
How to use
document_assembler = nlp.DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
sentence_detector = nlp.SentenceDetectorDLModel.pretrained("sentence_detector_dl", "en")\
.setInputCols(["document"])\
.setOutputCol("sentence")
tokenizer = nlp.Tokenizer()\
.setInputCols(["sentence"])\
.setOutputCol("token")
embeddings = nlp.RoBertaEmbeddings.pretrained("roberta_embeddings_legal_roberta_base","en") \
.setInputCols(["sentence", "token"]) \
.setOutputCol("embeddings")\
.setMaxSentenceLength(512)\
.setCaseSensitive(True)
ner_model = legal.NerModel.pretrained("legner_sec_edgar", "en", "legal/models")\
.setInputCols(["sentence", "token", "embeddings"])\
.setOutputCol("ner")
ner_converter = nlp.NerConverter()\
.setInputCols(["sentence", "token", "ner"])\
.setOutputCol("ner_chunk")
nlpPipeline = nlp.Pipeline(stages=[
document_assembler,
sentence_detector,
tokenizer,
embeddings,
ner_model,
ner_converter])
empty_data = spark.createDataFrame([[""]]).toDF("text")
model = nlpPipeline.fit(empty_data)
text = ["""In our opinion, the accompanying consolidated balance sheets and the related consolidated statements of operations, of changes in stockholders' equity, and of cash flows present fairly, in all material respects, the financial position of SunGard Capital Corp. II and its subsidiaries ( SCC II ) at December 31, 2010, and 2009, and the results of their operations and their cash flows for each of the three years in the period ended December 31, 2010, in conformity with accounting principles generally accepted in the United States of America."""]
result = model.transform(spark.createDataFrame([text]).toDF("text"))
Results
+----------------------------------------+---------+
|chunk |ner_label|
+----------------------------------------+---------+
|SunGard Capital Corp. II |ORG |
|SCC II |ALIAS |
|accounting principles generally accepted|LAW |
|United States of America |LOC |
+----------------------------------------+---------+
Model Information
| Model Name: | legner_sec_edgar |
| Compatibility: | Legal NLP 1.0.0+ |
| License: | Licensed |
| Edition: | Official |
| Input Labels: | [sentence, token, embeddings] |
| Output Labels: | [ner] |
| Language: | en |
| Size: | 16.3 MB |
References
In-house annotations
Benchmarking
label precision recall f1-score support
ALIAS 0.86 0.74 0.79 84
COURT 0.86 1.00 0.92 6
INST 0.94 0.76 0.84 76
LAW 0.91 0.93 0.92 166
LOC 0.89 0.88 0.88 140
MISC 0.90 0.83 0.86 226
ORG 0.89 0.93 0.91 430
PER 0.92 0.92 0.92 66
TICKER 1.00 0.86 0.92 7
micro-avg 0.90 0.88 0.89 1201
macro-avg 0.91 0.87 0.89 1201
weighted-avg 0.90 0.88 0.89 1201