Smart Research System — Intelligent Paper Analysis
✦ Project Overview
An AI-powered academic research accelerator that replaces hours of manual literature review with minutes of automated insight. The system queries IEEE Xplore for the top papers on any topic, feeds their abstracts to a Groq LLM, and automatically categorizes findings into three research phases — Foundational, Active Development, and Emerging — then exports a formatted PDF literature review with clickable citation links.
✦ Key Features
- ♥Direct IEEE Xplore API integration to programmatically fetch the top-ranked papers for any research topic and date range — no manual browsing or copy-pasting required.
- ♥Groq LLM-powered abstract synthesis that automatically categorizes papers into three research phases (Foundational, Active Development, Emerging) using strict JSON-constrained prompting for structured, parseable output.
- ♥One-click PDF export via jsPDF-AutoTable that generates a formatted literature review table with clickable citation links — ready to share with supervisors or include in a research proposal.
- ♥LocalStorage-based search history so researchers can instantly revisit and compare past queries and analysis sessions without re-running the pipeline.
- ♥State-based multi-view React UI (Search → Results → Detail) providing an intuitive, smooth navigation experience across the research workflow.
✦ Methodology
A three-phase AI pipeline that transforms a single research query into a structured, export-ready literature map:
Intelligent Paper Retrieval
A POST /api/research request carries the topic, start year, and end year to the FastAPI backend, which calls the IEEE Xplore API and fetches metadata and abstracts for the most relevant papers — removing the bottleneck of manual academic search.
LLM-Based Categorization
All fetched abstracts are assembled into a single structured prompt sent to the Groq LLM. Strict output constraints enforce a JSON response that groups papers into Foundational, Active Development, and Emerging categories along with synthesized summaries for each phase.
UI Rendering & PDF Export
The React frontend displays the categorized results as interactive cards with smooth state-based routing. Users can click 'Export Full Doc' to trigger client-side jsPDF-AutoTable generation, producing a professional literature review table with live citation hyperlinks — all without a server round-trip.