Inspiration
One of our group members faced difficulties conducting research. The first step to research is literature review where you have to find relevant research papers on your topic and write reviews about them. Most of the time the articles are not even relevant to the topic of research and they are too long. Annoyed by this our member was going to give up. To help, we created Scholar Scribe.
What it does
Scholar Scribe is an application that uses AI to summarize 100 research papers. Users enter their research topic into Scholar Scribe's search bar, and using serpAPI it finds 100 relevant research papers. Next, BeautifulSoap and tika scrape through these research papers to get the information, which is then processed through gpt-3.5-turbo to display a relevancy score and descriptive summary of each research paper to the user. Overall, for the user, Scholar Scribe is an easy way to find relevant articles for any research topic, preventing struggle and saving time.
How we built it
Scholar Scribe was built with powerful technologies. For the UI we used Svelte, a javascript framework, and Tailwind-CSS, a CSS framework, alongside HTML. Through these technologies, we were able to create a minimalistic and easy-to-use UI. The core functionality of the application uses Flask, a Python server, Serp API, to find the research papers, and Beautiful Soup along with Tika Apache, to scrape the information. Finally, with the help of the Open-AI API, we summarize and rate the research papers.
Challenges we ran into
First off we ran into the challenge with Tailwind CSS. It was not bundling properly which prompted us to use Rollup JS in order to integrate it with Flask. Next, there is a problem with the APIs as we had to pass the data through each and make sure that it would not take too long. We also ran into problems with the server not handling the requests for the summaries properly which made us turn it all into a JSON to use later.
Accomplishments that we're proud of
We're proud of how we were able to pick up several APIs such as SerpAPI and the Open AI API quite quickly. We were able to extract a majority of research papers successfully using Beautiful Soup and Apache Tika, which was new to us. We were able to create a good-looking front end and were able to integrate it all with routing using Flask. Overall, our project showed a good amount of complexity and we're proud of it.
What we learned
We learned how to use modern technologies and complex APIs that are used in the real world. We learned how to use Serp API and Open AI API to get and summarize the research papers. We also used technologies like Svelte, Tailwind CSS, and Flask and learned how to integrate all of them to create a full-stack application.
What's next for Scholar Scribe
Next for Scholar Scribe, we are planning to deploy so it is accessible to everyone in the world. We also plan on finding alternatives to the Open AI API possibly training our model to keep the costs low and affordable for everyone to use. We also plan on scaling it so that you can have your papers summarized and improving the system to show more results and allowing the user to filter them.
Log in or sign up for Devpost to join the conversation.