VeriFind: A React Web App for Searching and Validating Academic Research

[ghaziakmalf]

Project Name

VeriFind

Description

Project VeriFind is a web application that enables users to search academic literature and validate research documents with ease. The application provides advanced search functionality with customizable result counts, real-time search with loading states, paginated results, and a responsive search interface. Users can upload PDF documents for validation, view a general summary of the document analysis, and receive page-by-page suggestions for improvement. The application also includes a warning system for large result sets and a statement validation feature. VeriFind is built using React, FastAPI, LangChain, and Groq.

Build

Yes

Train

No

Analyze

No

Challenge Topic / Topic Category

• Customized Information Extraction
• Education
• Enhancing Accessibility in Healthcare Through AI
• Mitigating Natural Disasters in a Changing Climate
• Reducing Food Waste Through AI-Powered Innovation
• other - World Issues
• other - Open-Source
• other - Scientific Computing
• other

Project Repository URL

https://github.com/ScientiFolks/VeriFind

Deployed Endpoint URL

No response

Project Video File (not folder) Link (ensure viewer access)

https://drive.google.com/file/d/1vXfirl_e9DAfsEtA_Tg0Ri1nSqLhTRBs/view?usp=sharing

Team Members

@ghaziakmalf, @Mehmed13, @akmaldika, @Breezy-DR, @MuhLibri

[naumnaum]

Your project is an impressive tool that addresses a key challenge in academic research—efficiently searching and validating academic literature. The attention to code quality, documentation, and presentation is evident and commendable. Excellent work!
Beautiful frontend though!

The "Find" feature is particularly noteworthy. Have you considered implementing it as a browser extension? It could be incredibly useful for researchers and scientific workers who frequently navigate academic resources online.

Regarding the "Analyze" feature, I'd like to share an insight from working with retrieval-augmented generation for scientific papers. A common challenge is the model generating hallucinated answers to user queries when the retrieved data doesn't contain the relevant information. A potential enhancement could involve implementing a validation mechanism to ensure that the model's answers are strictly based on the provided data. If the information isn’t available, the model should refrain from answering. This would significantly enhance the reliability of your solution.

Overall, this is a well-thought-out project with strong potential for adoption in the academic and research community. Great job!