Description
This is a NER model, aimed to be run only after detecting the REQ_DISCL clause with a proper classifier (use legmulticlf_mnda_sections_paragraph_other model for that purpose). It will extract the following entities: DISCLOSURE_BASIS, REQ_DISCLOSURE_CONFID, REQ_DISCLOSURE_COOPERATION, REQ_DISCLOSURE_LEGAL, REQ_DISCLOSURE_NOTICE, REQ_DISCLOSURE_PARTY, REQ_DISCLOSURE_REMEDY, and REQ_OBLIGATION_ACTION.
Predicted Entities
DISCLOSURE_BASIS, REQ_DISCLOSURE_CONFID, REQ_DISCLOSURE_COOPERATION, REQ_DISCLOSURE_LEGAL, REQ_DISCLOSURE_NOTICE, REQ_DISCLOSURE_PARTY, REQ_DISCLOSURE_REMEDY, REQ_OBLIGATION_ACTION
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_req_discl", "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 = ["""If the Discloser waives the Recipient’s compliance with the agreement or fails to obtain a protective order or other appropriate remedies, the Recipient will furnish only that portion of the Confidential Information that is legally required to be disclosed and will use its best efforts to obtain confidential treatment for such Confidential Information."""]
result = model.transform(spark.createDataFrame([text]).toDF("text"))
Results
+----------------------+--------------------------+
|chunk |ner_label |
+----------------------+--------------------------+
|Discloser |REQ_DISCLOSURE_PARTY |
|obtain |REQ_OBLIGATION_ACTION |
|protective order |REQ_DISCLOSURE_REMEDY |
|appropriate remedies |REQ_DISCLOSURE_REMEDY |
|furnish |REQ_OBLIGATION_ACTION |
|legally required |REQ_DISCLOSURE_LEGAL |
|best efforts |REQ_DISCLOSURE_COOPERATION|
|obtain |REQ_OBLIGATION_ACTION |
|confidential treatment|REQ_DISCLOSURE_CONFID |
+----------------------+--------------------------+
Model Information
| Model Name: | legner_nda_req_discl |
| 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
DISCLOSURE_BASIS 0.77 0.70 0.73 57
REQ_DISCLOSURE_CONFID 0.96 0.93 0.95 29
REQ_DISCLOSURE_COOPERATION 1.00 0.94 0.97 17
REQ_DISCLOSURE_LEGAL 0.93 0.77 0.84 35
REQ_DISCLOSURE_NOTICE 0.89 0.89 0.89 19
REQ_DISCLOSURE_PARTY 1.00 0.89 0.94 38
REQ_DISCLOSURE_REMEDY 1.00 1.00 1.00 52
REQ_OBLIGATION_ACTION 0.95 0.86 0.90 121
macro-avg 0.94 0.86 0.90 368
macro-avg 0.94 0.87 0.90 368
weighted-avg 0.93 0.86 0.90 368