satire-news-detector

Made for an Assignment for the Artificial Intelligence course (CS F407) at BITS Pilani, Hyderabad.
This system takes in an input (news headline) and says whether the headline sounds satirical (from news sources that mock real life news) or is a genuine news headline.

Files and Folders

barplots/: Bar plots of useful data from real and fake news datasets (scraped from online sources)
data/: .csv files of data
clean_data/: .csv files of cleaned data
frontend/ : Contains the User-Interface for the website, developed in Python Flask.
imports.py: functions and library imports
dataclean.py: program to clean the raw data from data/
domains.py: program to find most common domains referenced in both datasets
combined.csv: Combined clean data of both datasets. Further work wil be carried out using this data.
nlp.py: NLP on the clean, combined data
Text_Outputs.txt: Intermediate outputs the NLP process, for reference.
model.py : Working model which uses a Naive Bayes Classifier at its core.
optimal.py : GridSearch parameters on two different models for determining the best model and its parameters.

How to run

Clone the repo
Run frontend/main.py

Libraries

Make sure you have the following libraries installed:

sklearn  
matplotlib  
seaborn  
pandas  
numpy

Group

Ekanshi Agrawal - @ekanshi258
Kushagra Srivastava - @z3r0dmg
Kunal Verma - @stumblef00l