Description
This is a financial model to detect LIABILITY and CONTRA_LIABILITY mentions in texts.
- CONTRA_LIABILITY: Negative liability account that offsets the liability account (e.g. paying a debt)
- LIABILITY: Current or Long-Term Liability (not from stockholders)
Please note this model requires some tokenization configuration to extract the currency (see python snippet below).
Predicted Entities
LIABILITY, CONTRA_LIABILITY
How to use
document_assembler = nlp.DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
sentence_detector = nlp.SentenceDetectorDLModel.pretrained("sentence_detector_dl","xx")\
.setInputCols(["document"])\
.setOutputCol("sentence")
tokenizer = nlp.Tokenizer()\
.setInputCols(["sentence"])\
.setOutputCol("token")\
.setContextChars(['.', ',', ';', ':', '!', '?', '*', '-', '(', ')', '”', '’', '$','€'])
embeddings = nlp.BertEmbeddings.pretrained("bert_embeddings_sec_bert_base", "en") \
.setInputCols("sentence", "token") \
.setOutputCol("embeddings")\
.setMaxSentenceLength(512)
ner_model = finance.NerModel.pretrained("finner_contraliability", "en", "finance/models")\
.setInputCols(["sentence", "token", "embeddings"])\
.setOutputCol("ner")
ner_converter = nlp.NerConverter()\
.setInputCols(["sentence", "token", "ner"])\
.setOutputCol("ner_chunk")
pipeline = nlp.Pipeline(stages=[
document_assembler,
sentence_detector,
tokenizer,
embeddings,
ner_model,
ner_converter
])
data = spark.createDataFrame([["""Reducing total debt continues to be a top priority , and we remain on track with our target of reducing overall debt levels by $ 15 billion by the end of 2025 ."""]]).toDF("text")
model = pipeline.fit(data)
result = model.transform(data)
result.select(F.explode(F.arrays_zip('ner_chunk.result', 'ner_chunk.metadata')).alias("cols")) \
.select(F.expr("cols['0']").alias("text"),
F.expr("cols['1']['entity']").alias("label")).show(200, truncate = False)
Results
+---------+------------------+----------+
| token| ner_label|confidence|
+---------+------------------+----------+
| Reducing| O| 0.9997|
| total| B-LIABILITY| 0.7884|
| debt| I-LIABILITY| 0.8479|
|continues| O| 1.0|
| to| O| 1.0|
| be| O| 1.0|
| a| O| 1.0|
| top| O| 1.0|
| priority| O| 1.0|
| ,| O| 1.0|
| and| O| 1.0|
| we| O| 1.0|
| remain| O| 1.0|
| on| O| 1.0|
| track| O| 1.0|
| with| O| 1.0|
| our| O| 1.0|
| target| O| 1.0|
| of| O| 1.0|
| reducing| O| 0.9993|
| overall| O| 0.9969|
| debt|B-CONTRA_LIABILITY| 0.5686|
| levels|I-CONTRA_LIABILITY| 0.6611|
| by| O| 0.9996|
| $| O| 1.0|
| 15| O| 1.0|
| billion| O| 1.0|
| by| O| 1.0|
| the| O| 1.0|
| end| O| 1.0|
| of| O| 1.0|
| 2025| O| 1.0|
| .| O| 1.0|
+---------+------------------+----------+
Model Information
| Model Name: | finner_contraliability |
| Compatibility: | Finance NLP 1.0.0+ |
| License: | Licensed |
| Edition: | Official |
| Input Labels: | [sentence, token, embeddings] |
| Output Labels: | [ner] |
| Language: | en |
| Size: | 16.2 MB |
References
In-house annotations on Earning Calls and 10-K Filings combined.
Benchmarking
label precision recall f1-score support
B-CONTRA_LIABILITY 0.7660 0.7200 0.7423 50
B-LIABILITY 0.8947 0.8990 0.8969 208
I-CONTRA_LIABILITY 0.7838 0.6304 0.6988 46
I-LIABILITY 0.8780 0.8929 0.8854 411
accuracy - - 0.9805 8299
macro-avg 0.8626 0.8267 0.8429 8299
weighted-avg 0.9803 0.9805 0.9803 8299