Edit on GitHub
  • John Snow Labs
  • Nov 09, 2022

Extract Information from Termination Clauses (sm)

  • termination
  • en
  • licensed

Description

IMPORTANT: Don’t run this model on the whole legal agreement. Instead:

  • Split by paragraphs. You can use notebook 1 in Finance or Legal as inspiration;
  • Use the legclf_termination_clause Text Classifier to select only these paragraphs;

This is a NER model which extracts information from Termination Clauses, like the subject (Who? Which party?) the action (verb) the object (What?) and the Indirect Object (to Whom?).

Predicted Entities

TERMINATION_SUBJECT, TERMINATION_ACTION, TERMINATION_OBJECT, TERMINATION_INDIRECT_OBJECT

Copy S3 URI

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_termination','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])

text = "(b) Either Party may terminate this Agreement"
data = spark.createDataFrame([[test]]).toDF("text")
model = nlpPipeline.fit(data)

Results

+-----------+---------------------+
|      token|            ner_label|
+-----------+---------------------+
|          (|                    O|
|          b|                    O|
|          )|                    O|
|     Either|B-TERMINATION_SUBJECT|
|      Party|I-TERMINATION_SUBJECT|
|        may| B-TERMINATION_ACTION|
|  terminate| I-TERMINATION_ACTION|
|       this|                    O|
|  Agreement|                    O|
+-----------+---------------------+

Model Information

Model Name: legner_termination
Compatibility: Legal NLP 1.0.0+
License: Licensed
Edition: Official
Input Labels: [sentence, token, embeddings]
Output Labels: [ner]
Language: en
Size: 16.4 MB

References

In-house annotations of CUAD dataset.

Benchmarking

label	 tp	 fp	 fn	 prec	 rec	 f1
I-TERMINATION_INDIRECT_OBJECT	 6	 2	 0	 0.75	 1.0	 0.85714287
B-TERMINATION_INDIRECT_OBJECT	 5	 3	 2	 0.625	 0.71428573	 0.6666667
B-TERMINATION_OBJECT	 48	 13	 25	 0.78688526	 0.65753424	 0.7164179
I-TERMINATION_ACTION	 84	 11	 12	 0.8842105	 0.875	 0.8795811
I-TERMINATION_OBJECT	 337	 75	 145	 0.81796116	 0.6991701	 0.75391495
B-TERMINATION_SUBJECT	 43	 5	 1	 0.8958333	 0.97727275	 0.9347826
I-TERMINATION_SUBJECT	 38	 6	 0	 0.8636364	 1.0	 0.9268293
B-TERMINATION_ACTION	 42	 4	 1	 0.9130435	 0.9767442	 0.94382024
Macro-average	                 -        -        -    0.8170713     0.86250085  0.83917177
Micro-average	                 -        -        -     0.83518004   0.76425856  0.7981469