In today’s digital world although the internet has made all our lives easier, there is a much bigger concern associated with it. It’s really easy for us to go into the web to give a quick glance over today’s news from sources of all sorts from reputed news websites to forwarded news on WhatsApp groups but there is no guarantee about the authenticity of the news. This could lead to an fast propagating chain of misleading information spreading from one point of origin to an infinite number of service points. Some of the most common sources of news around us are: news blogs, social media and online newspapers. Such a huge pool of information source has made it quite difficult for us to distinguish fake news. The first step for this would be to gather a reliable dataset and two open-source datasets from kaggle are used namely “Fake News detection” and “Real_or_Fake”. And then the process, annotation, and most importantly the validation process are described and also several exploratory data analysis are presented on the attempt to identify the differences based on linguistic features between fake and genuine news content. And then, we carry forward a lot of learning experiments to gradually build a reliable and accurate fake news detection model. Using the collected datasets, we also carry forward a very thorough data pre-processing to help determine linguistic features that set apart fake news content.