Student-Developed AI Tool Fast-Tracks Literature Review

A new tool created by a team of student researchers at the University of California, Davis, could reduce the time spent on literature review from months to minutes.

ResearchQuest.ai was developed by Akash Bonagiri, a Ph.D. student in computer science, and Gerard Anderias, an undergraduate student in computer science, who were paired together as part of the UC Davis College of Engineering E-SEARCH program. The program connects undergraduate students with graduate students for a research project. 

Before they could get started on any idea that they were entertaining, the literature review — a review of all the academic papers that could relate to the project’s subject, even tangentially — loomed large. It’s a time-consuming process with no shortcuts.

“It generally takes me several months to a year to curate a lot of papers that are relevant to my research and to really get an understanding of what is happening in the field,” said Bonagiri.

Recognizing this hardship, Bonagiri and Anderias decided to tackle it as their research project and develop a potential solution to this time-consuming, tedious process.

Their idea: ResearchQuest.ai.

Your Research Sidekick

ResearchQuest.ai is an end-to-end AI agent (meaning it can autonomously manage a complete task from start to finish with minimal human input) that compiles academic papers related to a query. It identifies the sections within the papers that seem important to that query and creates summaries for the user. 

It also creates a table comparing the different papers, showing a bird’s-eye view of what the different papers are talking about, enabling users to see which papers are more aligned with their research focus. 

Not only does ResearchQuest.ai speed up the entire process by compiling and sorting papers related to the topic, but it also allows users to “interact” with the papers via a chat feature.

“Say I want to find out what the main information from the abstract is,” Anderias said. “ResearchQuest.ai will take information from the abstract and give context to the large language model to be able to answer and give output to the user.”

Bonagiri points out that when the answers are produced, ResearchQuest.ai cites sources to show where the information came from, a feature they added to mitigate potential hallucinations.  

Turning Theory into Tools

Bonagiri and Anderias are currently working to have the tool available via a website this October. It will be free to use, and users will need to create an account to save their work progress and revisit their work.

In its early stages, ResearchQuest.ai pulled papers from Open Review only. Bonagiri and Anderias want to open it to Google Scholar and all websites that host academic papers, widening the pool of papers to select from.

The team is also integrating more citation generation and management so that users can easily export the citations. Before making it available to the public, they will need to deploy it onto a server and implement strategies for traffic management.

Bonagiri says he uses ResearchQuest.ai for his Ph.D., particularly to stay on top of the current research in his chosen field of AI and machine learning. Using a tool like ResearchQuest.ai could be a game-changer for all researchers to expedite this tedious step in the research process.

“The entire process and pipeline of the literature review can be a cakewalk with ResearchQuest.ai,” Bonagiri said. “If you want to catch up on things superfast and focus on the interesting part, the implementation part, this is the tool you need to use.” 

Bonagiri and Anderias are excited for their peers and professors to start using ResearchQuest.ai. They both agree that one of the most exciting parts of this project and the E-SEARCH experience has been creating a tangible product that people can use and seeing their research in action.

“A paper mostly works on the theoretical side of things,” Anderias said. “Having a product people can actually use is super interesting because you get a lot of user feedback, and it becomes a lot more rewarding.”