A conversational AI to learn research articles

Research papers are currently too complex and tedious for non-researchers to read. Many of us want to learn about topics, but don’t understand research jargon or want to read a ten-page research paper. Our solution, Ark, is an AI chatbot to help everyone understand research papers. Using natural language processing, Ark analyzes a user’s question and parses the Arxiv research paper for an answer, thus making it simpler to read research papers.

Here’s how Ark works. First, search the Arxiv database on our website for a topic. Next, choose a research paper you like. If you want, read the abstract we provide. Finally, start asking Ark questions. Ark will parse the research paper and answer any questions about the research paper you might have. Ark understands your question and provides a satisfactory answer based on the contents of the article. Questions to ask include “Who was the author?”, “What does X mean?”, and “What is the article about?”. Our chatbot, Ark, can also provide users with any definitions they might need.

Our chatbot uses natural language processing to understand a user’s question and generate an answer based on the article. However, Ark has more innovative technology than just our chatbot. Ark contains the only program found on the web to convert directly from an Arxiv paper’s ID number to the paper’s LaTeX form. Additionally, we used a Selenium script to automate the process of understanding and answering questions users have about the article.

We are currently adding Google Scholar and KhanAcademy support to Ark. Using Google Scholar, Ark will suggest similar articles based on title, author, and abstract. KhanAcademy will offer tools to teach the user topics they may need to understand the research paper. We currently have the backend developed and are currently working on frontend integration. And since no other chatbots for research papers exist, we plan to perfect Ark and release her to the public.

Share this project:

Updates