Sentiment Classification is a common business use case for any company that sells products or services. Companies utilize the sentiment information of the consumers to improve their products/services and even build new products based on the feedback received. It helps in making rational decisions.
We will build a Sentiment Classification model using BERT. We will need a labeled dataset for model training and evaluation. If the data is not labeled, we have to label it manually.
- Data Collection
- Data Labelling
- Convert target variables into the numeric form
- Text Preprocessing
- tokenization(subword tokenization handled by WordPiece tokenizer)
- text encoding
- text embedding
- Model training
- Model Evaluation
- Model Serving