Description
This is a Multiclass classification model which classifies arguments in legal discourse. These are the following classes: subsumption, definition, conclusion, other.
Predicted Entities
subsumption, definition, conclusion, other
How to use
documentAssembler = nlp.DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
tokenizer = nlp.Tokenizer()\
.setInputCols(["document"])\
.setOutputCol("token")
embeddings = nlp.RoBertaEmbeddings.pretrained("roberta_embeddings_legal_roberta_base", "en")\
.setInputCols(["document", "token"])\
.setOutputCol("embeddings")\
.setMaxSentenceLength(512)
embeddingsSentence = nlp.SentenceEmbeddings()\
.setInputCols(["document", "embeddings"])\
.setOutputCol("sentence_embeddings")\
.setPoolingStrategy("AVERAGE")
docClassifier = legal.ClassifierDLModel.pretrained("legclf_argument_mining","en", "legal/models")\
.setInputCols(["sentence_embeddings"])\
.setOutputCol("category")
nlpPipeline = nlp.Pipeline(stages=[
documentAssembler,
tokenizer,
embeddings,
embeddingsSentence,
docClassifier
])
df = spark.createDataFrame([["There is therefore no doubt – and the Government do not contest – that the measures concerned in the present case ( the children 's continued placement in foster homes and the restrictions imposed on contact between the applicants and their children ) amounts to an “ interference ” with the applicants ' rights to respect for their family life ."]]).toDF("text")
model = nlpPipeline.fit(df)
result = model.transform(df)
result.select("text", "category.result").show()
Results
+--------------------+-------------+
| text| result|
+--------------------+-------------+
|There is therefor...|[subsumption]|
+--------------------+-------------+
Model Information
| Model Name: | legclf_argument_mining |
| Compatibility: | Legal NLP 1.0.0+ |
| License: | Licensed |
| Edition: | Official |
| Input Labels: | [sentence_embeddings] |
| Output Labels: | [class] |
| Language: | en |
| Size: | 22.2 MB |
References
Train dataset available here
Benchmarking
label precision recall f1-score support
conclusion 0.93 0.79 0.85 52
definition 0.87 0.81 0.84 58
other 0.88 0.88 0.88 57
subsumption 0.64 0.79 0.71 52
accuracy - - 0.82 219
macro-avg 0.83 0.82 0.82 219
weighted-avg 0.83 0.82 0.82 219