Inspiration
With the spread of misinformation and polarized narratives, the inspiration for PARser, a Chrome extension that dissects news articles for accuracy and political bias emerged with the goal to restore truth and foster informed citizenship. This tool aims to cut through the noise, equipping users with the means to discern fact from fiction and navigate the complex news landscape with clarity and confidence.
What it does
When reading any news article on the web, a user can simply highlight the text of the article and click the extension icon from the menu bar. The extension will use cutting-edge technology to analyze the degree of misinformation and political bias within the article, and it will report back and explain its results to the user.
How we built it
The front end of our Chrome extension was built with HTML, CSS, and JavaScript, while our back-end server was implemented with Flask and Python. Our solution's innovation is provided in the text processing module by using the state-of-the-art Large Language Model, GPT, accessible through Open AI's API. The system was deployed to GCP using Compute Engine Virtual Machines.
Challenges we ran into
We ran into challenges every step of the way, but our most testing experience was with developing the front end. Getting selected text in a browser may seem like a simple task, but alas, it. is. NOT. Furthermore, we initially tried to have our extension open a popup window instead of a new tab, but Google Chrome's infrastructure makes that extremely difficult. Fortunately, opening up a new tab pushed the limits of our extension because it allowed us to report more information and more compelling visuals to the user due to the lack of size constraints.
Accomplishments that we're proud of
- Making a visually compelling front-end.
- Utilizing cutting-edge OpenAI's API in an inventive way.
What we learned
- How to develop Chrome extensions.
- HTML, CSS, and JavaScript.
- How to use GPT through OpenAI's API.
- Develop a simple yet efficient API with Flask and Python.
- Deploy a functional system to Google Cloud Platform using a Compute Engine Virtual Machine.
What's next for PARser: Political Articles Reviewer
We would love to add the feature of recommending more reliable and unbiased news articles on the same topic that the user is currently reading. We firmly believe that open-source research is the key to democratizing AI, and politics is no exception. Therefore, we would like to explore how open-source models, such as Llama-2, can perform in this task.
Log in or sign up for Devpost to join the conversation.