VGlass

• 

  Articles Found

• 

  Main Page

Inspiration

In high school, we spent dozens of hours writing research papers. We hunched over our shoulders gathering sources, analyzing data points, and drawing conclusions meticulously for many assignments. Motivated to find a more efficient process to approach these Meta-Analyses in AP Research and Psychology, we wondered if Generative AI could assist us when writing them.

What it does

VGlass compiles papers and allows users to choose which articles appear in an AI-generated Meta-Analysis. The articles are then processed in a Large-Language-Model, which proceeds to write in detail elements such as a research abstract, data analysis, potential gaps in others' research, and a conclusion based on the data points. Gathering these sources toward the central language processing model is a step forward in providing accessibility to users and providing data analysis of papers.

How we built it

We used React.js with Vite for the frontend and Python/Flask for the backend.

Challenges we ran into

Our main problem was getting past CORS but we fixed it by hosting a backend server from Heroku. Another problem that came up was sending API calls to the server.

Accomplishments that we're proud of

We're proud of the intuitive UI.

What we learned

We learned a lot about using Flask which caused many difficulties in connecting the frontend and backend.

What's next for VGlass

The AI model still needs to be refined. We are currently using Gemini but theoretically, we could train our own model on existing meta-analyses. In addition, we could use a technological API like Aryn to enhance the production of visuals like graphs in the PDFs. We would also be able to expand our scope and find more articles.

Built With

• flask
• python
• react
• thread