- I warn you, all steps of the project is for adding emojis like 🙂 and 😤 to movie comments (as sentiments).
I made this project in a day (as weekend project) to warm up with Django more and use a Bert Model / Transformers on it.
Used concepts in the projects:
- Django web app,
- Flask Resp API
- Web Scraping
- Sentiment Analysis (with pretrained model)
- I took the dataset from kaggle and used pretrained bert model on it. And checked if this pretrained model works well or not. (it is good enough) You can find it here.
- After this I needed header photos of the movies in tha dataset. (They are not included in the dataset.) So I made a little web scraping part in my project to take images. Actually I didnt take images, I only took their image urls. here
- And I put the preprocessing part of the sentiment analysis and pretrained bert transformers model in Flask Rest API. here
- And lastly I made a simple Django app to show intermediate steps and the interface. here
How I planned this project before I started.
Sample screenshots from the app:
Dynamic url logic is little wrong in the project. Turkish characters in the url like /movies/türkçe-özel/ is not supported by Django. I realized it in the end but didnt fixed. I didnt dockerized the app, because bert model required anaconda, I didnt try it to save time.