Exploratory Data Analysis and Preprocessing.The development phase of the project is divided in 5 phases: This section of the dashboard is for displaying the live, state wise statistics of COVID-19 cases in India.įor the Model Training and Validation the fastai approach was used along with Keras(Tensorflow 2.0) and Pytorch. Also the real time tweets are displayed using the publish functionality of twitter. The analysed tweets are displayed on the page with the tweet text and the tweet sentiment. In order to create the Twitter Live Feed Analysis Dashboard the website scrapes tweets in real-time every 5 minutes, performs sentiment analysis of these tweets and displays a 3-D interactive, live, donut plot depicting the sentiment counts of the tweets. This dashboard updates itself every 24 hours. The analysed results are then displayed on the website using various graphs as in the case of public sentiment analysis dashboard. 2,500 tweets are scraped everyday using this API, and their sentiment is then extracted using the developed language model. A new data set was created using these tweets to create a public sentiment dashboard on the final web application which depicted categories like sentiment triggers, overall sentiment of the tweets using robust graphs like the funnel chart, word nexus plots, bigram frequency, box plots and Network Visualizations etc.įurther creating a real-time sentiment analysis dashboard, COVID-19 tweets in India, the Tweepy API is used to scrape tweets from twitter on a real-time basis. To create a public sentiment analysis dashboard, tweets were scraped on the days when the Government of India made major decisions like the “lockdown”, “lockdown 2.0”, “unlock1.0”, etc based on the hashtags used in the tweets, during a period of 3 months using the Tweepy API and GetOldTweets3 package. Website UI Click the Image to Watch a Demo of the CoVis Dashboard.

Project Presentation Click Image to watch Project Presentation RoBERTa-CNN - For Sentiment Triggers Extraction.Two deep learning models were used,namely: Transfer learning approach was adopted in building the models in the project. In this project the data for training was obtained from IEEE Datasets and tweets scrapped from Twitter over a period of 3 months using the Tweepy API. Tweet is a post on the social media platform Twitter with a maximum of 140 characters. This project is based on analyzing the sentiment of tweets which help in understanding the pulse of the nation towards the pandemic. Hello there! This is the Github repository for CoVis - The Visualisation Dashboard of COVID-19 Twitter Sentiment Analysis. SBSPS-Challenge-2700-Twitter-Sentiment-Analysis-Extraction-for-COVID-19 Twitter Sentiment Analysis & Extraction for COVID-19

Nexus challenge 2.0 end time