Description
This model should be run on Force Majeure clauses. Use a Text Classifier to identify those clauses in your document, then run this NER on them - it will extract keywords related to Force Majeure exemptions.
Predicted Entities
O, FORCE_MAJEURE
How to use
documentAssembler = nlp.DocumentAssembler()
.setInputCol(“text”)
.setOutputCol(“document”)
sentenceDetector = nlp.SentenceDetectorDLModel.pretrained(“sentence_detector_dl”,”xx”)
.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”)
ner_model = legal.NerModel.pretrained(‘legner_force_majeure’,’en’,’legal/models’)
.setInputCols([“sentence”, “token”, “embeddings”])
.setOutputCol(“ner”)
ner_converter = nlp.NerConverter()
.setInputCols([“sentence”,”token”,”ner”])
.setOutputCol(“ner_chunk”)
nlpPipeline = nlp.Pipeline(stages=[ documentAssembler, sentenceDetector, tokenizer, embeddings, ner_model, ner_converter])
empty_data = spark.createDataFrame([[””]]).toDF(“text”)
model = nlpPipeline.fit(empty_data)
text = [”"”Force Majeure. In no event shall the Trustee be responsible or liable for any failure or delay in the performance of its obligations hereunder arising out of or caused by, directly or indirectly, forces beyond its control, including, without limitation, strikes, work stoppages, accidents, acts of war or terrorism, civil or military disturbances, nuclear or natural catastrophes or acts of God, and interruptions, loss or malfunctions of utilities, communications or computer (software and hardware) services; it being understood that the Trustee shall use reasonable efforts which are consistent with accepted practices in the banking industry to resume performance as soon as practicable under the circumstances.”””]
res = model.transform(spark.createDataFrame([text]).toDF(“text”))
+--------------+---------------+
| token| ner_label|
+--------------+---------------+
...
| ,| O|
| directly| O|
| or| O|
| indirectly| O|
| ,| O|
| forces| O|
| beyond| O|
| its| O|
| control| O|
| ,| O|
| including| O|
| ,| O|
| without| O|
| limitation| O|
| ,| O|
| strikes|B-FORCE_MAJEURE|
| ,| O|
| work|B-FORCE_MAJEURE|
| stoppages|I-FORCE_MAJEURE|
| ,| O|
| accidents|B-FORCE_MAJEURE|
| ,| O|
| acts|B-FORCE_MAJEURE|
| of|I-FORCE_MAJEURE|
| war|I-FORCE_MAJEURE|
| or| O|
| terrorism|B-FORCE_MAJEURE|
| ,| O|
| civil|B-FORCE_MAJEURE|
| or| O|
| military|B-FORCE_MAJEURE|
| disturbances|I-FORCE_MAJEURE|
| ,| O|
| nuclear|B-FORCE_MAJEURE|
| or| O|
| natural|B-FORCE_MAJEURE|
| catastrophes|I-FORCE_MAJEURE|
| or| O|
| acts|B-FORCE_MAJEURE|
| of|I-FORCE_MAJEURE|
| God|I-FORCE_MAJEURE|
| ,| O|
| and| O|
| interruptions|B-FORCE_MAJEURE|
| ,| O|
| loss|B-FORCE_MAJEURE|
| or| O|
| malfunctions|B-FORCE_MAJEURE|
| of|I-FORCE_MAJEURE|
| utilities|I-FORCE_MAJEURE|
| ,| O|
|communications|B-FORCE_MAJEURE|
...
+--------------+---------------+
Model Information
| Model Name: | legner_force_majeure |
| Compatibility: | Legal NLP 1.0.0+ |
| License: | Licensed |
| Edition: | Official |
| Input Labels: | [sentence, token, embeddings] |
| Output Labels: | [ner] |
| Language: | en |
| Size: | 16.5 MB |
References
In-house annotations on CUAD dataset
Benchmarking
label tp fp fn prec rec f1
I-FORCE_MAJEURE 91 36 37 0.71653545 0.7109375 0.7137255
B-FORCE_MAJEURE 140 32 17 0.81395346 0.89171976 0.85106385
Macro-average 231 68 54 0.7652445 0.80132866 0.782871
Micro-average 231 68 54 0.77257526 0.8105263 0.7910959