-
-
Example Run of Program, shows accuracy for the first two articles generated for Ukranian Bombing
-
csv file of all data extracted from websites
Inspiration
From everything that’s been going on this past year ranging from the Russian-Ukranian war to the drama shown in the presidential election, we both have seen just how much of an impact media sources can have in influencing our views about what goes on in society today. What we’ve realized is that the way somebody is in terms of their personal opinions can highly impact the way news is delivered. There can be cases for example in which opinions of highly negative or highly positive stance on an issue can cause information to be extremely over exaggerated causing the issue of false news to arise.
What it does
Looking at how fake news is so prone to developing inspired us to be able to design and develop a proof-of-concept browser extension that is able to read the first 9-12 articles on a news event and determine which ones are biased/inaccurate and which ones aren’t. We utilized a sentimental analysis algorithm in order to gauge the accuracy of each article through comparing sequences of keywords/phrases and calculating the accuracy based on the overall score of the article in terms of positive, neutral, and negative text.
How we built it
For our approach, we were able to integrate python and html together as well as utilize a wide range of libraries such as pandas, numpy, sklearn, and spacy to be able to put this project together. Additionally, we utilized google cloud to be able to extract data from a list of websites that were generated from searching the keyword/news event up.
Challenges we ran into
ERRORS. SO. MANY. ERRORS. Given that this was our first time participating in a hackathon, and also that we were new to machine learning, we learned and implemented much of the information in our code within our 24 hours. There were so many errors that we had to overcome, but we learned a lot from it and are so glad we could participate. Even though the errors were annoying, they gave us a chance to debug and understand what we were doing.
Accomplishments that we're proud of
WE FINISHED! That was the biggest accomplishment. We learned so much from this experience and we were impressed with how much we could get done in 24 hours. Going from not having much knowledge in Sentiment Analysis and constantly changing our approach to the problem to now having a thorough understanding of how to tackle a machine learning based browser extension is the greatest achievement we could've asked for.
What we learned
We learned a lot about Sentiment Analysis Machine Learning and creating Google Extensions. We both can now recreate code like this. The biggest skill we learned was Googling -- we learned how to sift through information online and find the most helpful tools to put together a successful project. Most importantly, we learned that hackathons are so much fun and we're ready to do another!
What's next for FakeSnake
FakeSnake has the potential to be a part of the big companies -- Honey, Grammarly, Loom, Google Translate, and more. With Fake News on the rise, it's become increasingly important for people to have a tool that can do the hard work for them so that they can get the unbiased, reasonable, and statistically accurate reality of the world fast.
Log in or sign up for Devpost to join the conversation.