Unmask

Analysing hatred or extremism spread by a user on Twitter profile

Do you know how much hatred a user spreads on Twitter?

How can one find the extent of hatred and extremism spread by a User on the Twitter platform?

The answer to these questions is our project, Unmask. It shows how much a user is spreading hatred in society using Twitter as a platform.

Motivation and Problem Statement:

We often see that social media is used as a battlefield by two or more political parties during the time of elections to spread hatred about the ideas of the opposition parties and different communities, which may affect the results in the election. Influencing on social media, particularly Twitter, is a common essential part of the political campaign. Twitter is a prime example of a social media platform used by political leaders, celebrities, and organizations to communicate their thoughts about politics and social issues. Nowadays, celebrities or media influencers can make a controversy just by posting a tweet, to which a lot of Twitter users flood the comments sections with their different opinions like some might agree with the tweet, some might show hatred towards that twitter handler due to this tweet, others might have some other opinion.

So, by analysing the sentiment of each of these thousands of comments for a post, we can determine how a tweet is affecting the emotions of the common users and how much impact it has upon society. So, in our project Unmask we are defining the sentiments generated by a tweet. Also, we will be thoroughly analyzing a profile to determine the tone of this handle’s whole profile, connectivity between the followers, providing the heatmap of the location of followers of that account, and any bots involved with the profile. Thus giving you an unmasked shape of the desired Twitter handle in real-time.

Our Approach :

When a user of the Unmask, our project, suggests a Twitter profile to analyze, we will collect the data required, analyse them, and give away the results to the user.

Data we collected for the analysis:

Profile's tweets

Comments of the tweets

User handles of the followers

First, we will be considering the tweets and analyze their sentiment and polarity. This will give us an idea of how polarised, extreme, and biased the selected Twitter account is. Then we will read the overall emotion of the common users based on the comments of the tweets. We will know if the words are sad, angry, happy or otherwise. This will say if the account is spreading any negativity among its readers. After the analysis of tweets and replies, we now read the reports of followers. We read the data of their bio's and give an overall scan of the follower's bias to positivity and negativity. We then make a network of the followers, and by the denseness of the graph, we can say how well connected the followers. We will also be telling if there are bots in the followers. We also generate the word cloud of the most frequently used words by the selected Twitter profile to infer what kind of tweets this handle usually posts and what kind of issues this handle emphasizes. From all these results, one would surely know the tone/mood of the account.

Tools we used :

Tweepy :

Python has a very active developer community that creates many libraries which extend the
language and push the limits by making it easier to use those services.
One of those libraries is Tweepy. Tweepy is open-sourced, hosted on GitHub, and enables Python to communicate with the Twitter platform and use its API.

text2emotion :

‘text2emotion’ is the python package that helps in the extraction of the emotions from the content.
It processes any textual message and recognizes the emotions embedded in it.
It is compatible with five different emotion categories as Happy, Angry, Sad, Surprise, and Fear.

ML techniques for bot detection, plotly and wordcloud modules are also used.