In every major democracy the news firms in the region control which way the election goes and which politician wins and loses. Since these journalism firms play a huge part in how people form their opinions.
That undermines the purpose of democracy since then the control of power is being run from the people who run these journalism firms, not the regular population. This also means people lose power from their ability to vote for the candidate they believe in.
However, while looking at this problem I realized that I could solve this problem using machine learning. I could create a machine learning program that could detect the political bias in a piece of text and then plot that data.
Then readers of any article can read the article of their choice while identifying the bias in the article. This has the possibility to radically shift the way democracies around the world operate.
It uses deep learning and specifically the word2vec feature to assign a random mathematical number to every word which helps with classification. Then it vectorization each word and plots the vectorization to see the polarity of each word. Using vectors rather than sentiment analysis allows for the detection of contextual bias rather than just word for word bias.
This tool has been been validated by employees at the Globe and Mail and numerous startups in the region.
What inspired you (or your team)?
From a young age, I was exposed to my parents reading articles regarding political affairs from many trusted sources and I’ve seen the radical bias in these sources and how these biases impacted the way, my parents thought and the political stances they firmly took.
I realized the importance of unbiased journalism and I tried my best to look for unbiased news sources or tools that would open my parent’s eyes up to a world of different perspectives and show them how little they were exposed to, but I was disappointed to see almost no high-quality options out there.
Witnessing that I realized I had a skill many other people my age don’t have, the ability to program machine learning algorithms, so I decided to try my very best to develop a program that would allow people like my parents to identify the contextual bias in any article they’re reading.
I couldn’t find anyone else passionate about solving this specific problem. So I decided I’ll try to develop this by myself. It took me over 100 days of failing to get myself my first actual progress.
I had the vision to build a simple but impactful solution to a problem I’ve seen throughout my life.