Fake news spread through social media has become a serious problem. In this article we will understand how sentiment classification can be done with ‘fact-check’ on forwarded messages and integrate the solution as a chatbot in Telegram.

We will go through the architecture that I have built for this solution and try to address the veracity and correctness of forwarded messages in social media.

The corpus is small and extracted from verified facts on PIB official Twitter page.

AI chatbot: As the name suggests, it is an AI software program that imitates human conversation through text or voice interactions. The reason why chatbots are becoming more prominent is because they can work 24x7, save time & money, connect business to customers & employees, home automation etc.

Limiting the scope, we concentrate on Sentiment analysis instead of Conversational bot. I will address the problem to detect if the text message is a fake or fact rather than intent classification.

Chatbot approach: Choose if the approach is proactive (i.e. suggestions to the users request) or reactive (just responding to users request). For the fact-check use-case we go with the reactive approach, as we just need to respond telling if the message is a fact or not.

Following is the architecture:

#artificial-intelligence #chatbot #machine-learning #data-science