Sentiment analysis is a popular problem in machine learning. People are constantly trying to understand the sentiment of a product or movie review. Currently, for sentiment analysis, we extract the text from a client/browser, pass it on to a server that runs a machine learning model to predict sentiment of the text, and the server then sends the result back to the client.
This is perfectly fine if we don't care about the latency in the system. However, there are many applications, such as stock trading, customer support conversations where it might be helpful to predict sentiment of the text with low latency. One obvious bottleneck in reducing latency is the server call.
If sentiment analysis could be achieved on the browser/client itself, we can get rid of the server call and can predict the sentiment in real...
