Sabeeh Ali Akbar

Academic

Career

Verbatim text

ExRaDe(Expose Racism Detection) 
Racism detector is a web application that helps the social media users to check their tweets before 
posting it on the twitter either it contains racism or not. More importantly, it is for Roman Urdu. A 
deep learning model LSTM is used for tweet classification either its racial or non-racial. The purpose 
of this application is to facilitate the users to avoid the posts that spread racism of any kind and in 
this way we can move towards a better future. We collected the dataset from Kaggle and then 
annotated it according to our problem. It was labelled by three of our team members so, it can 
have labels that are opinion based and have a subjective type which means someone can say its 
racial while it non-racial on the basis of other person’s opinion. We are working on an extension 
which will automatically count the tweets on the twitter that contains racism posted by the user on 
the twitter platform.  
Features include: 
- A front end which takes an input tweet from the user. 
- Preprocess the tweet taken as input. A model at backend will check the tweet and then will 
determine its class and will tell class of tweet to the user. Since it’s a binary classification therefore, 
there will be only two classes 0 and 1 where 0 stands for non-racial and 1 stands for racial class.  
The application will classify only the tweets in Roman Urdu as the model is trained on that specific 
dataset. Contextual based classification will also be performed in which the model will keep the 
track of history of the text taken as input.  
 
 
 
 
Technology Used: 
Python3, Flask, Keras, TensorFlow, 
 Pandas, HTML, CSS, Jupyter-notebook 
Supervisor Name: 
Mr. Jawad Hassan 
Group Members:   
Sabeeh Ali Akbar (i16-0017)

AI enrichment