ResearcHaven

A funding pipeline that catches duplicate research proposals with NLP.

Overview

Built for Smart India Hackathon, ResearcHaven centralizes how research proposals are submitted across funding agencies — and runs an NLP plagiarism checker to catch duplicate or overlapping submissions before they ever reach a reviewer.

The problem

Funding agencies receive overlapping and duplicate proposals with no shared view, wasting reviewer time and risking double-funding. Manual screening cannot scale, and it misses the subtle overlap that matters most.

Architecture

Client: Progressive web app (Submit and track proposals)
(50+ REST APIs) Backend: Node / Express API (Auth, drafts, checks, forwarding)
Services: Plagiarism checker (Python: AHP + NLTK / YAKE, three strictness levels); MongoDB (Proposals encrypted at rest); Ethereum (Record of funded proposals)
(accepted proposals) Out: Funding authority (Receives the proposals it should review)

Every proposal is checked for overlap before a reviewer sees it; accepted ones go on to the right funding authority.

Approach

NLP plagiarism detection

A custom checker combines the Analytic Hierarchy Process (AHP) with the NLTK and YAKE libraries to assign weights to each section of a proposal, then scores similarity at three levels of strictness.

A 50+ API backend

A Node/Express service exposes over fifty endpoints spanning authentication, draft management, plagiarism checks, and forwarding accepted proposals to the right funding authority.

Zero-knowledge storage

Proposals are encrypted at rest and rendered without ever being persisted in plaintext, so sensitive research stays confidential through the review pipeline.

A tamper-evident trail

Funded proposals are recorded on Ethereum for an auditable, tamper-evident history; the client ships as an installable progressive web app.

Highlights

AHP + NLTK/YAKE plagiarism scoring at three levels
50+ REST APIs across the proposal lifecycle
Encrypted, zero-knowledge proposal storage
Ethereum record for funded proposals

Stack

React.js, Node.js, Express.js, MongoDB, Python (NLTK/YAKE), Ethereum