Skip to content

sai161812/Research_Orchestrator

BranchesTags

Repository files navigation

Orchestrix

Multi-Agent Research Intelligence Platform
Discover papers, analyze trends, generate citations, and synthesize insights — all orchestrated by autonomous AI agents.

Features • Architecture • Agents • Getting Started • Tech Stack • Deployment

✨ Features

🔬 Intelligent Research Pipeline

  • Natural language queries — ask anything from broad topics to exact paper titles
  • Intent classification — hybrid rule-based + LLM classifier routes queries through the optimal pipeline
  • Automatic query expansion — generates complementary sub-queries via Groq LLM to maximize paper coverage
  • Entity detection — recognizes researchers, organizations, and concepts, then auto-extracts related sub-topics

📄 Multi-Source Paper Discovery

  • Dual-source fetching — concurrent retrieval from Semantic Scholar and arXiv
  • Exact title matching — fast-path lookup via Semantic Scholar's match endpoint for precise title queries
  • Smart deduplication — DOI + normalized title matching across sources with metadata merging
  • Hybrid ranking — composite scoring based on title match, keyword relevance, citation impact, recency, and source quality

📊 Analytics Dashboard

  • Publication volume trends — year-over-year research output visualization
  • Top contributing authors — most prolific researchers in your result set
  • Keyword & topic frequency — treemap of dominant terms across abstracts
  • Citation impact distribution — bucketed histogram of citation counts
  • Emerging sub-topics — trending terms from papers published in the last 2 years

📝 Citation Generation

  • 5 citation formats — APA, MLA, IEEE, Chicago, and BibTeX generated instantly for every paper
  • One-click copy — copy any citation to clipboard
  • Bulk export — download all citations as .txt or .bib files
  • Session reports — export full research reports as plain text or print-ready PDF

🎓 AI Mentor ("Mentor Me")

  • Learning path generation — LLM-powered prerequisite analysis for any paper
  • Key concepts glossary — plain-English explanations of foundational concepts
  • Step-by-step curriculum — ordered learning path from fundamentals to the paper's advanced topics

🧠 Paper Intelligence

  • AI-powered summaries — short, medium, or detailed structured summaries via Groq
  • Paper comparison — side-by-side LLM-driven comparison of any two papers
  • Multi-paper synthesis — cross-paper thematic analysis identifying common themes, differences, and gaps
  • Smart filtering — filter results by year, citation count, and sort order

💾 Session Persistence

  • Auto-save — every research session is automatically saved to localStorage
  • Session management — load, delete, annotate, and compare past sessions
  • Session comparison — side-by-side metric comparison of any two sessions
  • Execution trace — real-time transparency into every agent decision and API call

🏗 Architecture

Orchestrix follows a multi-agent orchestration pattern where a central Orchestrator coordinates specialized agents through a defined pipeline:

┌─────────────────────────────────────────────────────────┐
│                      USER QUERY                         │
└────────────────────────┬────────────────────────────────┘
                         │
                         ▼
              ┌─────────────────────┐
              │    ORCHESTRATOR     │
              │  Intent Classifier  │
              │  Query Expander     │
              │  Pipeline Router    │
              └────┬───┬───┬───┬───┘
                   │   │   │   │
        ┌──────────┘   │   │   └──────────┐
        ▼              ▼   ▼              ▼
  ┌───────────┐  ┌──────────┐  ┌──────────┐  ┌──────────┐
  │ DISCOVERY │  │ ANALYSIS │  │ CITATION │  │  SUMMARY │
  │   Agent   │  │   Agent  │  │   Agent  │  │   Agent  │
  │           │  │          │  │          │  │          │
  │• S2 API   │  │• Trends  │  │• APA     │  │• Groq    │
  │• arXiv    │  │• Authors │  │• MLA     │  │• Compare │
  │• Match    │  │• Keywords│  │• IEEE    │  │• Synth.  │
  │• Dedup    │  │• Citations│ │• BibTeX  │  │• Mentor  │
  └───────────┘  └──────────┘  └──────────┘  └──────────┘
        │              │             │              │
        └──────────────┴─────────────┴──────────────┘
                         │
                         ▼
              ┌─────────────────────┐
              │   SESSION STORE     │
              │   (localStorage)    │
              └─────────────────────┘

Query Flow

  1. Intent Classification — The Orchestrator classifies the query as title, topic, or entity using a rule-based classifier with LLM fallback via Groq
  2. Query Expansion — For topic queries, the LLM generates two complementary sub-queries to broaden paper coverage
  3. Parallel Discovery — The Discovery Agent fetches papers concurrently from Semantic Scholar and arXiv, with exact-match fast paths for title queries
  4. Deduplication & Ranking — Results are merged, deduplicated by DOI/title, and scored using a hybrid ranking algorithm
  5. Smart Fallback — If results are weak, the Orchestrator automatically pivots to entity-mode and extracts fallback sub-topics
  6. Analysis & Citation — The Analysis Agent computes bibliometric analytics while the Citation Agent generates formatted citations
  7. Session Persistence — All results, analyses, and traces are persisted to localStorage for future retrieval

🤖 Agents

Agent Responsibility Engine
Orchestrator Intent classification, query expansion, pipeline routing, fallback logic Rule-based + Groq LLaMA 3.3 70B
Discovery Agent Multi-source paper retrieval, exact title matching, deduplication Semantic Scholar API, arXiv API
Analysis Agent Publication trends, top authors, keyword frequency, citation distribution, emerging topics Local computation
Citation Agent APA/MLA/IEEE/Chicago/BibTeX generation, TXT/BIB/PDF export Local computation
Summarization Agent Paper summaries (short/medium/detailed), paper comparison, multi-paper synthesis Groq LLaMA 3.3 70B
Learning Agent Prerequisite extraction, learning path generation, concept glossary Groq LLaMA 3.3 70B

🚀 Getting Started

Prerequisites

  • Node.js ≥ 18
  • npm ≥ 9
  • A free Groq API key — get one here

Installation

# Clone the repository
git clone https://github.com/sai161812/Research_Orchestrator.git
cd Research_Orchestrator

# Install dependencies
npm install

# Configure environment
cp .env.example .env

Edit .env and add your Groq API key:

VITE_GROQ_API_KEY=your_groq_api_key_here

Development

npm run dev

The app will be available at http://localhost:5173. The Vite dev server automatically proxies requests to the Semantic Scholar and arXiv APIs, so no additional backend setup is needed.

Production Build

npm run build
npm run preview

🛠 Tech Stack

Layer Technology
Framework React 19 with Vite 8
Styling Tailwind CSS 4
Routing React Router v7
Charts Recharts
LLM Groq Cloud (LLaMA 3.3 70B Versatile)
Paper APIs Semantic Scholar Graph API, arXiv API
State localStorage (zero-dependency persistence)
Deployment Vercel (Serverless Functions for API proxies)

📁 Project Structure

orchestrix/
├── api/                          # Vercel Serverless Functions
│   ├── arxiv.js                  #   arXiv API proxy
│   └── semantic-scholar.js       #   Semantic Scholar API proxy
├── public/                       # Static assets
├── src/
│   ├── agents/                   # Autonomous agent modules
│   │   ├── AnalysisAgent.js      #   Bibliometric analysis engine
│   │   ├── CitationAgent.js      #   Citation formatting & export
│   │   ├── DiscoveryAgent.js     #   Multi-source paper retrieval
│   │   ├── LearningAgent.js      #   AI Mentor learning paths
│   │   └── SummarizationAgent.js #   LLM-powered summarization
│   ├── components/               # React UI components
│   │   ├── charts/               #   Recharts visualizations
│   │   │   ├── AuthorsChart.jsx
│   │   │   ├── CitationChart.jsx
│   │   │   ├── EmergingTopics.jsx
│   │   │   ├── KeywordChart.jsx
│   │   │   └── TrendChart.jsx
│   │   ├── EntityConfirm.jsx     #   Entity disambiguation UI
│   │   ├── FilterBar.jsx         #   Smart result filtering
│   │   ├── Hero.jsx              #   Animated hero section
│   │   ├── MentorPanel.jsx       #   AI Mentor slide-over panel
│   │   ├── Navbar.jsx            #   Navigation bar
│   │   ├── PaperCard.jsx         #   Rich paper result card
│   │   ├── SearchBar.jsx         #   Search input component
│   │   └── TracePanel.jsx        #   Real-time execution trace
│   ├── orchestrator/
│   │   └── Orchestrator.js       # Central orchestration engine
│   ├── pages/
│   │   ├── DashboardPage.jsx     #   Analytics dashboard
│   │   ├── ResearchPage.jsx      #   Main research interface
│   │   └── SessionsPage.jsx      #   Session management
│   ├── store/
│   │   ├── sessionStore.js       #   Session persistence (localStorage)
│   │   └── traceStore.js         #   Execution trace state
│   ├── utils/
│   │   ├── paperUtils.js         #   Dedup, ranking, filtering utilities
│   │   └── schema.js             #   Data model definitions
│   ├── App.jsx                   # Root component with routing
│   ├── App.css                   # Global styles
│   ├── index.css                 # Base styles
│   └── main.jsx                  # Entry point
├── .env.example                  # Environment variable template
├── vercel.json                   # Vercel deployment config
├── vite.config.js                # Vite + proxy configuration
└── package.json

☁️ Deployment

Orchestrix is designed for one-click deployment to Vercel:

  1. Push the repository to GitHub
  2. Import it in Vercel Dashboard
  3. Add the environment variable: VITE_GROQ_API_KEY
  4. Deploy — Vercel auto-detects the Vite framework and configures serverless functions from the api/ directory

The vercel.json rewrites ensure that /api/* routes are handled by the serverless proxy functions, while the Vite dev server proxy handles the same routes locally.

📄 License

This project is open source under the MIT License.

Built with ❤️ for researchers, by researchers.

About

Research orchestrator that retrieves, summarizes, and cites research papers.

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages