![]()
This blog is a short write-up about the sentiment analysis workflow I developed during the first Covid lockdown. You can find the dashboard website I created here and a blog I had written in my institute here.
Basically it's an Azure and c# solution. The main components involve Azure streaming analytics, Azure Event Hubs, and Azure Blob Store.
The code for the sample app can be found here. This sample app analyzes real-time Twitter feeds using sentiment140 (http://www.sentiment140.com/) to determine whether the tweeted text has a positive, negative, or neutral reaction to the current situation. The keywords used for picking the tweets are “lifting lockdown” and “lifting restrictions” in the same order of the words.
The feed from Twitter API is fed to Azure Event hub for handling huge stream of data, and then this stream of data is fed to azure streaming analytics which runs a tumbling window aggregation function for window width of 1hr, to aggregate the sentiment scores. This aggregated data is then stored in Azure blob store and visualized on this website.
The Tweeter reader is the main component which reads the tweets based on keywords and sends them to eventhub to start the sentiment analysis workflow.
The stream analytics query :
References
- https://docs.microsoft.com/en-us/azure/stream-analytics/stream-analytics-twitter-sentiment-analysis-trends