Description
This models is a version of legmulticlf_mnda_sections_other (sentence, medium) but expecting a bigger-than-sentence context, ideally between 2 and 4-5 sentences, or a small paragraph, to provide with more context.
It should be run on sentences of the NDA clauses, and will retrieve a series of 1..N labels for each of them. The possible clause types detected my this model in NDA / MNDA aggrements are:
- Parties to the Agreement - Names of the Parties Clause
- Identification of What Information Is Confidential - Definition of Confidential Information Clause
- Use of Confidential Information: Permitted Use Clause and Obligations of the Recipient
- Time Frame of the Agreement - Termination Clause
- Return of Confidential Information Clause
- Remedies for Breaches of Agreement - Remedies Clause
- Non-Solicitation Clause
- Dispute Resolution Clause
- Exceptions Clause
- Non-competition clause
- Other: Nothing of the above (synonym to
[])-
Predicted Entities
APPLIC_LAW, ASSIGNMENT, DEF_OF_CONF_INFO, DISPUTE_RESOL, EXCEPTIONS, NAMES_OF_PARTIES, NON_COMP, NON_SOLIC, PREAMBLE, REMEDIES, REQ_DISCL, RETURN_OF_CONF_INFO, TERMINATION, USE_OF_CONF_INFO, OTHER
How to use
document_assembler = (
nlp.DocumentAssembler()
.setInputCol("text")
.setOutputCol("document")
)
sentence_detector = (
nlp.SentenceDetector()
.setInputCols("document")
.setOutputCol("sentence")
.setExplodeSentences(True)
.setCustomBounds(['\n'])
)
embeddings = (
nlp.UniversalSentenceEncoder.pretrained()
.setInputCols("document")
.setOutputCol("sentence_embeddings")
)
paragraph_classifier = (
nlp.MultiClassifierDLModel.pretrained("legmulticlf_mnda_sections_paragraph_other", "en", "legal/models")
.setInputCols(["sentence_embeddings"])
.setOutputCol("class")
)
sentence_pipeline = nlp.Pipeline(stages=[document_assembler, sentence_detector, embeddings, paragraph_classifier])
prediction_pipeline = nlp.Pipeline(stages=[document_assembler, embeddings, paragraph_classifier])
text = """RECITALS
WHEREAS, Corvus Gold Nevada Inc., a Nevada corporation (“Corvus”) and AngloGold Xxxxxxx (U.S.A.) Exploration, a Delaware corporation (“Viewer”) entered into that certain Confidentiality Agreement with an effective date of December 4, 2017 (“CA”); and
WHEREAS, Corvus and Viewer desire to amend the terms of the CA pursuant to the terms of this Amendment; and
WHEREAS, any terms not defined herein shall have the meanings set forth in the CA, as amended from time to time
EXECUTION VERSION
VITAL IMAGES, INC.
TOSHIBA MEDICAL SYSTEMS CORPORATION
Confidentiality Agreement
This Confidentiality Agreement (this “Agreement”) dated as of January 28, 2011, between VITAL IMAGES, INC., a Minnesota corporation (“Vital Images” or the “Company”), and TOSHIBA MEDICAL SYSTEMS CORPORATION, a Japanese corporation (“TMSC” or the “Receiving Company”).
W I T N E S S E T H:
WHEREAS, the Parties wish to consider a strategic business transaction (the “Transaction”) and, in connection therewith, desire to set forth certain agreements regarding such consideration and the sharing of confidential and proprietary information by Vital Images with TMSC;
"""
df = spark.createDataFrame([[""]]).toDF("text")
sentence_model = sentence_pipeline.fit(df)
prediction_model = prediction_pipeline.fit(df)
sentence_lp = nlp.LightPipeline(sentence_model)
prediction_lp = nlp.LightPipeline(prediction_model)
res = sentence_lp.fullAnnotate(text)
sentences = [x.result for x in res[0]['sentence']]
for i, s in enumerate(sentences):
prev_sentence = "" if i==0 else sentences[i-1]
next_sentence = "" if i>=len(sentences)-1 else sentences[i+1]
chunk = " ".join([prev_sentence, s, next_sentence]).strip()
print(f"{prediction_lp.annotate(chunk)['class']}: {chunk}")
Results
['PREAMBLE']: WHEREAS, Corvus Gold Nevada Inc., a Nevada corporation (“Corvus”) and AngloGold Xxxxxxx (U.S.A.) Exploration, a Delaware corporation (“Viewer”) entered into that certain Confidentiality Agreement with an effective date of December 4, 2017 (“CA”); and WHEREAS, Corvus and Viewer desire to amend the terms of the CA pursuant to the terms of this Amendment;
['DEF_OF_CONF_INFO']: and WHEREAS, Corvus and Viewer desire to amend the terms of the CA pursuant to the terms of this Amendment; and
['OTHER', 'PREAMBLE']: WHEREAS, Corvus and Viewer desire to amend the terms of the CA pursuant to the terms of this Amendment; and WHEREAS, any terms not defined herein shall have the meanings set forth in the CA, as amended from time to time
Model Information
| Model Name: | legmulticlf_mnda_sections_paragraph_other |
| Compatibility: | Legal NLP 1.0.0+ |
| License: | Licensed |
| Edition: | Official |
| Input Labels: | [sentence_embeddings] |
| Output Labels: | [class] |
| Language: | en |
| Size: | 13.4 MB |
References
In-house MNDA
Benchmarking
label precision recall f1-score support
APPLIC_LAW 0.86 0.84 0.85 57
ASSIGNMENT 0.87 0.83 0.85 41
DEF_OF_CONF_INFO 0.86 0.76 0.81 67
DISPUTE_RESOL 0.84 0.69 0.76 70
EXCEPTIONS 0.84 0.79 0.82 109
NAMES_OF_PARTIES 0.90 0.76 0.83 50
NON_COMP 0.79 0.67 0.72 33
NON_SOLIC 0.81 0.82 0.81 82
OTHER 0.91 0.89 0.90 838
PREAMBLE 0.86 0.78 0.81 76
REMEDIES 0.91 0.84 0.87 87
REQ_DISCL 0.91 0.77 0.84 83
RETURN_OF_CONF_INFO 0.78 0.79 0.79 78
TERMINATION 0.74 0.67 0.70 42
USE_OF_CONF_INFO 0.77 0.84 0.80 200
micro-avg 0.87 0.83 0.85 1913
macro-avg 0.84 0.78 0.81 1913
weighted-avg 0.87 0.83 0.85 1913
samples-avg 0.82 0.84 0.83 1913