LSTM based model to classify sentiment of movie reviews (IMDB).
-
Use the LSTM to build the model. Optimize the model by tuning the dropout, achieving almost 96% accuracy.
-
The mode in this dashboard is used to evaluate the sentiment of input text or movie review. The output is probability, which is interpreted as follows : positive if more than 0.5, negative if less than 0.5 .
-
The movie reviews are fetched from IMDB website through IMDBPY API. The dashboard layout was built with Dash / Plotly.
-
Finally, the code was dockerized and deployed to Google Cloud Run (https://sutingatchicago-mypbgqf64a-uc.a.run.app/)(Note, it may take some time to load the web app.).
-
The web app is easily adapted to other sentiment analysis such as reviews in twitter, products reviews in E-commence, etc.
-
Contact me if have any question: sutingatchicago@gmail.com