A sentiment analysis application with an inbuilt bot that scrapes the comment section of a twitter post for generating interactive customer review insights.
Analysing twitter comments for a company's post is an indispensable resource of honest customer review. The application makes the process interesting and efficient. This app, Review Beta scrapes the comment section of the twitter post the user wishes to explore. It is inbuilt with a twitter bot, that essentially scrapes the comment section of the required twitter post url, and sends it to a sentiment analyser for sentiment analysis, and delivers beautiful interactive graphs.
- Python
- Selenium
- PyTorch
- Dash
- Flask
- bot.py : The twitter bot used to scrape a continuous loading twitter page for the comment section of the given post url.
- training.ipynb : Trained a bi-directional recurrent neural network model on 1.6 million labelled tweets.
- get_sentimet.py : Perform sentiment analysis on the scraped tweets by the pretrained model.
- clean_collected_data : Clean the collected data obtained by the bot.
- app.py : A dash server embedded into a flask server to deploy the interactive dashboard.
- Home page: Paste your desired twitter post url over here.
- Demo: Copy a URL
- Demo : Paste it to the form and click on search button.
- Demo: Insight 1.1: Sentiment analysis of tweets posted by date time sequence.
- Insight 1.2
- Insight 2: Graph depicting reply comment likes, replies, retweets and the tweet for representing popularity of the comment feedbacks.
- Insight 3: Sentiment score of top 5 comments for a popular opinion analysis.
- Work further to train the model on different languages.
- Feature engineering for better deduced insights.
- Optimise the bot for quicker and more efficient scraping process.
- Analysis of giphys or images posted for added semantic understanding.