COVID Bias Checker

Inspiration

Arguably, the biggest problem in America and most other western countries is that the media from both the left and the right often lies about facts in order to get their ratings up. Donald Trump likes to criticize them for this by calling them fake news. This problem is especially important during times of crisis, such as this. The COVID-19 pandemic is wreaking havoc across the world and part of the reason it has gotten so serious is because people are trusting media sources that are lying about the virus. For example, CNN lied a lot in May about the number of corona virus cases. They exaggerate the death toll by thousands and this caused unnecessary panic among citizens which made things a million times worse. We realized the gravity of this problem and decided that we wanted to help.

What It Does

Our website's main app page allows the user to input in the URL of the article that they want to read. Our website then applies two main algorithms to it. It uses both a natural language processing model algorithm that scans the article and gives out a numerical bias rating, and it shortens the URL that the user inputted into just the domain name and runs it through the mediabiasfactcheck.com API to determine the political bias of that news source. After it has these two pieces of information, the website gives the user the bias rating of the article. The user can then use the information that they have received to become more weary of how fake or real the COVID-19 article they are reading is. By being more weary, they can avoid unnecessary panic and stay safe. Apart from the main page, the web app also has two static about us and submission pages.

How We Built It

We built this website using Python, Flask, Bootstrap, Javascript, CSS, HTML Jinja, and a variety of Natural Language Processing and linear regression algorithms.

Challenges We Ran Into

We had to combine a few NLP and classification algorithms which took a lot of time for processing and made us kind of impatient. Downloading all of the modules and setting up the virtual environment also took forever. Lastly, once we were all done the website and were ready to deploy to heroku, we encountered a major problem. We kept exceeding the slug size limit. After hours of debugging, we found out that all we had to do to solve this issue was remove the python mkl modules. That was really funny.

Accomplishments That We Are Proud Of

We are proud that we could complete such an advanced and complicating project in such a small amount of time. It was definitely a very confusing project even though both of us had worked with NLP before. Nonetheless, this was a huge feet of achievement for both of us and we could not be more proud.

What We Learned

We learned a lot during this hackathon. We learned a lot about natural language processing this hackathon. We had only done topic modelling before but learning how you can use AI for as complicated as things as sentimental analysis was definitely surprising. It was also our first times using Flask to host an actual website. We had only used it to make rest apis before and we are proud of having learned this new skill.

What's Next For COVID Bias Checker

We are going to refine the app more by getting expert opinions. Once the app is completely finished, we will spread the word about it in the hopes of making it go viral so that people can stay aware and safe.

Built With

• anaconda
• flask
• heroku
• mediabiasfactcheck.com
• natural-language-processing
• python