Offinsive tweets classification

School Assignment

Task Description

The task to be completed within this project is to develop and evaluate a system that processes

a single tweet and decides whether or not the tweet contains offensive language. This task

could be tackled as a binary text classification problem. To this end, a dataset of

tweets annotated for offensiveness is provided in the repo. The files “train.tsv” and “dev.tsv” all have the same format.

The second column (text) contains the text of a tweet, the first column (label) contains an

offensiveness label:

  • (NOT) Not Offensive - This post does not contain offense or profanity.

  • (OFF) Offensive - This post contains offensive language or a targeted (veiled or direct)


The table below shows a few examples of tweets that

have been annotated as offensive.


1 @USER Liberals are all Kookoo !!!

2 @USER @USER Most stupid tweet yet…and yes…its a libtard

3 @USER I bet the first ones she calls when she is being robbed is those

very same police . People are freaking stupid …

4 @USER @USER Go home you’re drunk!!! @USER #MAGA #Trump2020 URL

Baseline Classifier

Naive bayes and svm classifiers were used as models. link to the repo.

Abdallah Bashir
Abdallah Bashir
Research Intern at Allen AI