Inspiration

In the 21st century, the modern media has become a sea of rhetoric and misinformation. Sym-Metrics aims to increase the accessability of unbiased & factual data through its use of cutting edge machine learning and natural language processing technology.

What it does

The Sym-Metrics app acts as an all in one news solution for the inquisitive individual. We utilize state of the art analytics to provide users with a collection of the most unbiased and relevant news articles.

How we built it

The entire application is built on RESTful api that allows communication between multiple iOS front end clients and a Python-Flask back end. Smart caching and data filtering are utilized to keep connections snappy, and allow for a seamless user experience.

The back end gathers information through a custom Beautiful-Soup web scraper which pulls from a myriad of diverse news outlets.

The scraped articles are analyzed using IBM's Alchemy API, and are transformed into 16 dimensional feature vectors. These feature vectors are then piped through our custom built neural network, trained using Amazon Web Services, to generate a final bias score.

Accomplishments that we're proud of

1) Custom Unsupervised Kernel Clustering Algorithm to assist in generating bias ratings 2) Home grown MLP Neural Net for classification 3) Full Swift 3 front end, complete with animations and caching system 4) Robust, custom built, highly generalizable web scraping algorithm

Share this project:

Updates