Edit on GitHub
  • John Snow Labs
  • Apr 19, 2023

Legal NER for NDA (Return of Confidential Information Clauses)

  • en
  • legal
  • licensed
  • ner
  • nda

Description

This is a NER model, aimed to be run only after detecting the RETURN_OF_CONF_INFO clause with a proper classifier (use legmulticlf_mnda_sections_paragraph_other model for that purpose). It will extract the following entities: ARCHIVAL_PURPOSE, and LEGAL_PURPOSE.

Predicted Entities

ARCHIVAL_PURPOSE, LEGAL_PURPOSE

Download Copy S3 URI

How to use

document_assembler = nlp.DocumentAssembler()\
        .setInputCol("text")\
        .setOutputCol("document")
        
sentence_detector = nlp.SentenceDetector()\
        .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_nda_return_of_conf_info", "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 = ["""Notwithstanding the foregoing, the Recipient and its Representatives may retain copies of the Confidential Information to the extent that such retention is required to demonstrate compliance with applicable law or governmental rule or regulation, to the extent included in any board or executive documents relating to the proposed business relationship, and in its archives for backup purposes subject to the confidentiality provisions of this Agreement."""]

result = model.transform(spark.createDataFrame([text]).toDF("text"))

Results

+--------------+----------------+
|chunk         |ner_label       |
+--------------+----------------+
|applicable law|LEGAL_PURPOSE   |
|governmental  |LEGAL_PURPOSE   |
|regulation    |LEGAL_PURPOSE   |
|archives      |ARCHIVAL_PURPOSE|
|backup        |ARCHIVAL_PURPOSE|
+--------------+----------------+

Model Information

Model Name: legner_nda_return_of_conf_info
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 on the Non-disclosure Agreements

Benchmarking

label             precision  recall  f1-score  support 
ARCHIVAL_PURPOSE  0.94       1.00    0.97      16      
LEGAL_PURPOSE     0.78       0.85    0.81      33      
micro-avg         0.83       0.90    0.86      49      
macro-avg         0.86       0.92    0.89      49      
weighted-avg      0.83       0.90    0.86      49