The aim of this project is to create a custom dataset for sentiment analysis. Use the data to fine-tune a BERT model and deploy your NLP model as an API.
• This repository consists of files required for end to end implementation of Sentiment Analysis of Google Play App Reviews Natural Language Processing Web App created with ___FastApi.
- Clone the repository : https://github.com/ni3choudhary/Sentiment-Analysis-Google-Play-App-Reviews.git
- Inside the project root directory, Create Python Virtual Environment and activate it using below commands
$ python3 -m venv envActivate Virtual Environment
$ .env/bin/activate
OR
$ .\env\Scripts\activateInstall Libraries using below command
$ pip install -r requirements.txt- Run jupyter notebooks to get the necessary files if you try to build it from scratch.
Use the files as following to create the project from scratch or create your own project in an adapted way.
1. scrape_app_information.ipynb to scrape top 15 app information from the productivity category using google-play-scraper package.
2. scrape_app_reviews.ipynb to scrape more than 15k user reviews from those 15 productivity apps using google-play-scraper package.
3. sentiment_analysis_with_bert_and_hugging_face_using_torch.ipynb to fine-tune BERT for sentiment analysis and save the best model for deployment purpose. Here, you'll do the required text preprocessing (special tokens, padding, and attention masks) and build a Sentiment Classifier using Transformers.
4. sentiment_analyzer/assets/ please put saved model file named best_model_state.bin into this directory.
5. sentiment_analyzer/classifier/sentiment_classifier.py to create a classifier that uses the BERT model.
6. sentiment_analyzer/classifier/model.py to create an interface to abstract the inference logic. It exposes a single predict() method with all the text processing required to build a sentiment analysis model.
- Now Inside sentiment_analyzer directory run api.py on terminal to start local server.
$ uvicorn api:app --reload
• Please do ⭐ the repository, if it helped you in anyway.