Description
This model is a pre-trained NLP model to analyze sentiment of financial text. It is built by further training the DistilRoBerta language model in the finance domain, using a financial corpus and thereby fine-tuning it for financial sentiment classification. Financial PhraseBank by Malo et al. (2014) and in-house JSL documents and annotations have been used for fine-tuning.
Predicted Entities
positive, negative, neutral
How to use
documentAssembler = nlp.DocumentAssembler() \
.setInputCol("text") \
.setOutputCol("document")
tokenizer = nlp.Tokenizer() \
.setInputCols("document") \
.setOutputCol("token")
classifier = nlp.RoBertaForSequenceClassification.pretrained("finclf_distilroberta_sentiment_analysis","en", "finance/models") \
.setInputCols(["document", "token"]) \
.setOutputCol("class")
nlpPipeline = nlp.Pipeline(
stages = [
documentAssembler,
tokenizer,
classifier])
# couple of simple examples
example = spark.createDataFrame([["Stocks rallied and the British pound gained."]]).toDF("text")
result = nlpPipeline.fit(example).transform(example)
# result is a DataFrame
result.select("text", "class.result").show()
Results
+--------------------+----------+
| text| result|
+--------------------+----------+
|Stocks rallied an...|[positive]|
+--------------------+----------+
Model Information
| Model Name: | finclf_distilroberta_sentiment_analysis |
| Compatibility: | Finance NLP 1.0.0+ |
| License: | Licensed |
| Edition: | Official |
| Input Labels: | [document, token] |
| Output Labels: | [class] |
| Language: | en |
| Size: | 309.1 MB |
| Case sensitive: | true |
| Max sentence length: | 256 |
References
In-house financial documents and Financial PhraseBank by Malo et al. (2014)
Benchmarking
label precision recall f1-score support
positive 0.77 0.88 0.81 253
negative 0.86 0.85 0.88 133
neutral 0.93 0.86 0.90 584
accuracy - - 0.86 970
macro-avg 0.85 0.86 0.85 970
weighted-avg 0.87 0.86 0.87 970