An interactive application that takes either text or article URL as input and outputs sentiment of every sentence in the article. The sentiment is obtained using a fine-tuned BERT model on financial news articles. The performance of the model is compared (side-by-side) in the table with other out-of-thebox models (in this case textblob).
- Sentence: Input sentence
- prediction: Predicted sentiment for sentence using custom model.
- sentiment_score: Sentiment Score using custom/fine-tuned BERT model.
- textblob_prediction: Sentiment using out of the box textblob sentiment analyzer.