Fake news spread faster than COVID-19 itself and creates panic in public. We encounter many fake tweets in our family whatsapp group of FB news feed. I wanted a method to verify if that tweet is genuine or not before sharing it further
What it does
Given tweet screenshot, FakePosts.com identifies :
- If Tweeter Handle in screenshot exists in real or not.
- If Tweeter Handle exits, either he posted that tweet o
- It will link screenshot with its original tweet
- System will showcase list of fake tweets currently circulating
- System will keep a record of all different image urls that contain same image/tweet. It will help us to link same tweet screenshot on difference sources.
How I built it
I combined OCR with Twitter API which verifies tweet's authenticity. Chrome extension for FakePosts.com enables user to "Check Tweet's Authenticity" on just one click on that image.
Challenges I ran into
Training of OCR was biggest challenge. e.g It confuses capital 'i' I and small 'L' l as they both look same. Handeling such OCR issues was biggest challenge
Accomplishments that I'm proud of
More than 500+ tweets are authenticated using FakePosts.com in less last 4 days and many were identified as FAKE. I consider it an achievement if this service is helping to stop fake news to spread further.
What I learned
Do not blame society on problems. Play your part and Be a one who adds something to correct that problem.
What's next for FakePosts.com
Phase 1 is ready and now I'm looking forward to next phases in coming days -- PHASE 2 -- Once we have enough data, we will mark accounts frequently used for speading fake tweets. -- PHASE 3 --
- Third phase includes to introduce "Credebility Score 1-10"
- Credibility Score will be given on factual authenticity of text using citation and references that endorses fact.