HackBU2018-Fake-News-Detector

Fake News Detector

With the 2018 midterm elections approaching and fake news continuing to be a prominent issue in society, we set out to develop a method of automatically detecting and labeling fake news. In about 20 seconds, our program will scrape an article's contents, pull its ICANN WHOIS information, and run several text analysis algorithms to determine features such as readability. Then, these features are fed into a neural network with Tensorflow, which then labels the articles as "real news" or "fake news", as well as reporting the confidence scale of its findings.

Built With

Submitted to

HackBU 2018
- Winner First Place sponsored by GTS
- Winner Best Newbie Hack

Created by

I was the software lead and developed the structure of the program. I determined which features to use for training the neural network, such as text readability score and location where the website is registered. I programmed the text analysis, ICANN WHOIS, Google search, and a page scraper. I also helped build the training and testing datasets for the neural network. Lastly, I was responsible for integration and QA between my algorithms and the neural network.

Colin Fiutak
I worked on the frontend, building the website and displaying the result of the algorithm after taking the user's input. To accomplish this, I used Flash to create an html page that ran through python. I also did some work on scraping text from the articles.

koneill7 O'Neill
I helped create the datasets of fake news for our testing, and helped Gabe develop the machine learning method to train our data sets (using the tensorflow and pandas libraries).

hbrown10
I worked on the machine learning code. I used sample code from the tensorflow website and modified it to fit our needs.

GSteinberg Steinberg