Skip to content

Yogs4R/fuenzer-research

BranchesTags

Repository files navigation

Fuenzer Research 🔬

JuaraVibeCoding Hackathon Golang React Gemini AI Vite Cloud Run

Fuenzer Research Banner

🇮🇩 Indonesia | 🌐 English

Riset Akademis, Disintesis AI — Asisten riset ilmiah bertenaga kecerdasan buatan untuk membantu akademisi Indonesia mencari, memetakan indeks SINTA/Garuda, dan menyintesis literatur ilmiah skala nasional & global secara instan.

Live Demo: https://research.fuenzer.web.id

Built for JuaraVibeCoding Season 1 Hackathon by Google.

📋 Table of Contents

🎯 The Problem We Solve

Kriteria Penilaian #1 — Masalah / Problem (30%)

Target Audiens

Fuenzer Research dibangun untuk akademisi, mahasiswa S1/S2/S3, dan peneliti Indonesia yang menghadapi hambatan sistemik dalam proses riset literatur:

Masalah Dampak
Tidak ada tool yang memetakan tier SINTA (1-6) secara otomatis Peneliti harus cek manual satu per satu di situs SINTA untuk menentukan kualitas jurnal
Database Garuda (portal riset nasional) sulit di-search secara programatik Ribuan artikel Indonesia tidak terekspos ke tool riset modern
Tool riset global (Elicit, Semantic Scholar) tidak mengenal ekosistem akademik Indonesia Peneliti Indonesia terpaksa menggunakan banyak platform terpisah
Membaca 10-20 abstrak untuk literature review memakan 2-4 jam Waktu riset terbuang untuk tugas repetitif yang bisa diotomasi AI

Mengapa Ini Penting Sekarang

  • 300.000+ mahasiswa pascasarjana dan 100.000+ dosen di Indonesia membutuhkan akses cepat ke literatur terakreditasi
  • SINTA sebagai standar akreditasi nasional tidak terintegrasi dengan tools riset manapun secara otomatis
  • Era AI memungkinkan sintesis literatur dalam hitungan detik — tetapi belum ada yang melakukannya untuk konteks Indonesia

Potensi Skala (Impact)

🎯 TARGET PASAR

  • 🎓 300K+ Mahasiswa S2/S3 Indonesia
  • 👨‍🏫 100K+ Dosen & Peneliti Aktif
  • 📚 15.456 Jurnal Terakreditasi SINTA
  • 📄 3.6M+ Artikel di Database Garuda
  • 🌍 200M+ Publikasi Global (OpenAlex)

Fuenzer Research menjembatani kesenjangan antara indeks akademis lokal Indonesia dengan tools riset global — sesuatu yang belum dilakukan platform lain.

💡 Our Solution

Kriteria Penilaian #2 — Solusi / Solution (40%)

User Experience (UX)

Fuenzer Research dirancang dengan filosofi "Zero Friction Research" — pengguna bisa mulai riset dalam 3 detik tanpa perlu registrasi:

Aspek UX Implementasi
Intuitif Search bar langsung di hero section, zero sign-up required. Autocomplete "Did you mean?" dari OpenAlex
Delightful Narrative skeleton loader 3 fase ("Searching → Filtering → Synthesizing"), animasi rotating text, glassmorphism cards
Responsif Full mobile support — AI panel full-screen di mobile, breakpoint handling di setiap komponen
Dark Mode Implementasi penuh dengan design token system (Fuenzer Teal, Cloud Canvas, Ink Black, dll)
Bilingual (i18n) Seluruh UI tersedia dalam Bahasa Indonesia dan English — bisa di-toggle secara real-time
Accessible Semantic HTML, contrast ratio yang memadai, keyboard navigation support

Proposisi Nilai (Value Proposition)

Berikut hasil nyata yang Fuenzer Research berikan kepada pengguna:

Tanpa Fuenzer Research Dengan Fuenzer Research
Buka SINTA → cek tier manual → 15 menit per jurnal Pemetaan SINTA otomatis dalam <1 detik
Buka 3-4 platform terpisah (Scholar + SINTA + Garuda + Books) 1 search query → 4 sumber sekaligus
Baca 10 abstrak → 2-3 jam untuk literature review AI synthesis dalam <5 detik
Copy-paste sitasi manual ke berbagai format 5 format sitasi (APA/Harvard/MLA/Chicago/Vancouver) + BibTeX export
Tidak bisa compare multiple papers sekaligus AI Compare: pilih referensi → tanya AI tentang perbandingan

Workflow Lengkap

graph TD
    classDef step fill:#fafafa,stroke:#0d9488,stroke-width:2px,color:#262626;
    classDef accent fill:#0d9488,stroke:#0f766e,stroke-width:2px,color:#fff;

    Step1["🔍 1. CARI<br/>Pengguna memasukkan topik riset dan menyaring berdasarkan Tipe, Cakupan, dan Indeks"]
    Step2["🚀 2. TEMUKAN<br/>Permintaan paralel ke OpenAlex, SINTA, Garuda, dan Google Books dengan pemetaan tier SINTA otomatis"]
    Step3["🧠 3. SINTESIS<br/>Google Gemini 3.1 Flash Lite membaca abstrak untuk menghasilkan ringkasan terstruktur"]
    Step4["💬 4. INTERAKSI<br/>Menyaring, mengurutkan, menandai referensi, mengajukan pertanyaan lanjutan, atau membandingkan makalah"]
    Step5["📂 5. EKSPOR<br/>Mengunduh laporan PDF, ekspor massal ke ZIP, menyalin format sitasi, atau ekspor BibTeX"]

    Step1 --> Step2 --> Step3 --> Step4 --> Step5

    class Step1,Step2,Step3,Step4,Step5 step;
Loading

✨ What Makes Us Unique

Kriteria Penilaian #3 — Keunikan / Uniqueness (30%)

Orisinalitas

Fuenzer Research bukan wrapper ChatGPT dan bukan template standar. Berikut yang membedakan kami:

Aspek Detail
SINTA Auto-Mapping Satu-satunya platform yang secara otomatis memetakan tier SINTA (1-6) ke hasil pencarian literatur. Tidak ada tool riset manapun yang melakukan ini
Garuda SQLite 652K Kami mengkurasi 652.144 artikel dari database Garuda ke SQLite lokal — effort data engineering yang signifikan untuk aksesibilitas riset Indonesia
Custom Design System Bukan Material UI / Chakra / template. Kami mendesain dari nol: Firecrawl-inspired aesthetic dengan token warna, tipografi, dan komponen unik
Dual AI Mode Search Mode (cari literatur baru) + Ask Mode (tanya AI tentang referensi terpilih) dalam satu panel interaktif
Anti-Hallucination Pipeline AI hanya menjawab berdasarkan abstrak yang diberikan, bukan pengetahuan umum. Temperature 0.3 + strict academic prompt

Faktor "Wow" — Penggunaan AI yang Elegan

🤖 AI USAGE YANG TIDAK SEKADAR "CALL API"

  1. Narrative Skeleton Loader
    • 3-phase loading (Searching... → Filtering... → Synthesizing...) membuat pengguna tetap engaged.
  2. Token Economy
    • Hanya kirim Title + Abstract + Year ke Gemini, menghemat 70% token vs mengirim full paper data.
  3. Anti-Prompt Injection
    • System prompt melarang Gemini mengikuti instruksi dari user yang mencoba override.
  4. Graceful Degradation
    • Jika Gemini gagal → tetap return referensi tanpa synthesis (user tidak stuck).
  5. Gibberish Detection (Frontend + Backend)
    • Deteksi keyboard mashing sebelum memanggil AI untuk menghemat kuota & mencegah abuse.
  6. Contextual Q&A
    • User pilih 3 papers → tanya "bandingkan metodologi" → AI menjawab HANYA dari abstrak papers yang dipilih.

Bukan Template — Bukti Orisinalitas

  • Zero UI library template — tidak menggunakan admin dashboard template, landing page template, atau starter kit
  • Custom animations — word-flip-in, marquee tracks, number scramble, fade-in with intersection observer
  • Design dari scratch — design token system terinspirasi Firecrawl.dev dengan adaptasi untuk konteks akademis
  • Data engineering — mengkurasi & membersihkan 3.6M artikel Garuda menjadi 652K yang berkualitas (filter tahun 2024)

🏛️ Full Architecture

System Architecture Diagram

graph TD
    classDef client fill:#0f766e,stroke:#115e59,stroke-width:2px,color:#fff;
    classDef gateway fill:#0d9488,stroke:#0f766e,stroke-width:2px,color:#fff;
    classDef service fill:#14b8a6,stroke:#0d9488,stroke-width:2px,color:#fff;
    classDef database fill:#2dd4bf,stroke:#14b8a6,stroke-width:2px,color:#000;
    classDef ai fill:#8e75b2,stroke:#7c3aed,stroke-width:2px,color:#fff;
    classDef deploy fill:#4285f4,stroke:#1a73e8,stroke-width:2px,color:#fff;

    %% Client Layer
    subgraph Client ["Client Layer (Browser)"]
        SPA["React Router (SPA)"]
        Store["Zustand (State)"]
        Cache["React Query (Cache)"]
        Auth["Firebase Auth & Firestore (Sync)"]
    end

    %% Gateway Layer
    subgraph Gateway ["API Gateway (Go Fiber)"]
        Security["Security Headers (CSP/HSTS)"]
        CORS["CORS (Strict Rules)"]
        Limiter["Rate Limiter (15/min)"]
        Logging["Logging Middleware"]
    end

    %% Data Services
    subgraph Services ["Data Services & Local DBs"]
        OpenAlex["OpenAlex API (200M+ Papers)"]
        GoogleBooks["Google Books API"]
        SINTA["SINTA Mapper (JSON Dict)"]
        Garuda["Garuda SQLite DB (652K+ Docs)"]
    end

    %% AI Engine
    subgraph AIEngine ["AI Engine"]
        Gemini["Google Gemini 3.1 Flash Lite"]
    end

    %% Deployment
    subgraph ProdDeployment ["Deployment"]
        Docker["Docker Multi-Stage Build"]
        CloudRun["Google Cloud Run (HTTPS)"]
    end

    %% Relationships
    SPA -->|HTTPS POST/GET JSON| Gateway
    Gateway -->|Concurrent API Calls| OpenAlex
    Gateway -->|Concurrent API Calls| GoogleBooks
    Gateway -->|Local Mapping| SINTA
    Gateway -->|SQLite FTS Search| Garuda
    Gateway -->|Prompt Synthesis / Q&A| Gemini
    Docker -->|Deploy Container| CloudRun

    class SPA,Store,Cache,Auth client;
    class Security,CORS,Limiter,Logging gateway;
    class OpenAlex,GoogleBooks,SINTA,Garuda service;
    class Gemini ai;
    class Docker,CloudRun deploy;
Loading

Request Flow (Sequence)

sequenceDiagram
    autonumber
    actor User
    participant FE as Frontend (React)
    participant BE as Backend (Go Fiber)
    participant API as External APIs (OpenAlex/Books)
    participant DB as Local Data (SINTA/Garuda DB)
    participant AI as Gemini AI

    User->>FE: Types query & clicks Search
    Note over FE: 1. Gibberish detection (client-side)<br/>2. Keyword extraction & sanitization<br/>3. Autocomplete fetch
    FE->>BE: POST /api/v1/research
    Note over FE: Phase 1: "Searching OpenAlex..."
    Note over BE: Input validation & routing
    par Fetch Global
        BE->>API: Query OpenAlex / Google Books
        API-->>BE: Return papers metadata
    and Fetch Local
        BE->>DB: Query Garuda SQLite / SINTA JSON
        DB-->>BE: Return local articles
    end
    Note over FE: Phase 2 (1.5s): "Filtering & mapping SINTA..."
    BE->>BE: SINTA tier mapping (match publisher -> tier 1-6)
    Note over FE: Phase 3 (3.5s): "Synthesizing with Gemini AI..."
    BE->>AI: Send sanitized abstracts (Title, Abstract, Year)
    AI-->>BE: Return structured summary/synthesis
    BE-->>FE: Return unified JSON response
    Note over FE: 1. Render AI synthesis (Markdown -> HTML)<br/>2. Render reference cards with badges<br/>3. Enable interaction (filter, export, Q&A)
Loading

Kenapa Stack Ini Cepat?

Layer Teknologi Keunggulan Performa
Bundler Vite 8 (ESBuild) 10-100x lebih cepat dari Webpack. HMR <50ms
Routing React Router (SPA) Navigasi antar halaman = 0ms (swap komponen di memori)
State Zustand ~1KB, no boilerplate, selective re-render
Cache React Query (staleTime: 5min) Data di-cache — navigasi bolak-balik tanpa re-fetch
Backend Go Fiber Zero allocation HTTP, response <5ms untuk logic lokal
Concurrency Go Goroutines Handle 100K+ concurrent requests (vs Node.js single-thread)
Cold Start Alpine binary ~50MB Cloud Run cold start <200ms (vs Node.js 500-2000ms)
AI Gemini Flash Lite Model tercepat Google — optimized untuk low-latency inference

⚡ Features

Fitur Deskripsi
🔍 Multi-Source Search Satu query → OpenAlex (200M+ papers) + Google Books + SINTA (~700 journals) + Garuda (652K articles)
🇮🇩 SINTA Auto-Mapping Pemetaan otomatis tier jurnal Indonesia (SINTA 1-6) — fitur unik yang tidak dimiliki platform lain
🤖 AI Synthesis Google Gemini 3.1 Flash Lite menghasilkan ringkasan literatur terstruktur dalam <5 detik
💬 Dual AI Mode Search Mode (cari referensi baru) + Ask Mode (Q&A tentang referensi terpilih)
📂 PDF, BibTeX & CSV Export Ekspor referensi ke PDF, BibTeX (.bib) untuk LaTeX/Zotero, serta Matriks Literatur otomatis ke Excel/CSV
📚 Multi-Library Folder Simpan referensi ke beberapa folder pustaka kustom, disinkronkan via Firebase secara real-time
🔗 Shareable Link & Fork Bagikan workspace sesi riset atau folder pustaka publik via tautan unik, tamu bisa menduplikasi (fork)
🎨 Dark Mode + i18n Tema gelap/terang + bilingual penuh (ID/EN)
🔒 Anti-Hallucination Temperature 0.3, strict prompt, AI hanya menjawab dari data yang diberikan
🛡️ Security-First Rate limiting, CORS strict, CSP headers, DOMPurify, gibberish detection
⚡ Narrative Loader 3-phase animated loading yang membuat pengguna tetap engaged saat menunggu

🛠️ Tech Stack

Layer Technology
Frontend React 19, TypeScript 6, Vite 8
Styling Tailwind CSS 4.3, Custom Design Token System
UI Components Lucide React, Glassmorphism, Custom Animations
State Management Zustand 5 (global state), TanStack React Query 5 (server cache)
Authentication Firebase Auth 12 (Google, Microsoft, Email)
Cloud Storage Firebase Firestore (history & bookmark sync)
PDF & Export jsPDF 4.2, JSZip 3.10, BibTeX generator
Markdown Rendering Marked 18 + DOMPurify 3 (XSS-safe)
Backend API Golang 1.25, Fiber (REST API berkinerja tinggi)
AI Engine Google Gemini 3.1 Flash Lite (temperature 0.3)
Data Sources (Global) OpenAlex API (200M+ publications), Google Books API
Data Sources (Lokal) SINTA JSON (~700 jurnal, tier 1-6), Garuda SQLite (652K+ artikel)
Deployment Docker Multi-Stage, Google Cloud Run, Alpine 3.19
Security CORS strict, Rate Limiting 15/min, CSP Headers, HSTS

Detailed Dependency Versions

Package Version Role
react 19.2.6 UI library
react-dom 19.2.6 DOM rendering
react-router-dom 7.15.1 Client-side SPA routing
typescript 6.0.2 Type safety
vite 8.0.16 Build tool & dev server
@vitejs/plugin-react 6.0.1 React Fast Refresh
tailwindcss 4.3.0 Utility-first CSS
zustand 5.0.13 Lightweight state management
@tanstack/react-query 5.100.14 Server state & caching
firebase 12.14.0 Auth + Firestore
axios 1.16.1 HTTP client
lucide-react 1.16.0 Icon library
marked 18.0.4 Markdown parser
dompurify 3.4.11 XSS sanitization
jspdf 4.2.1 PDF generation
jszip 3.10.1 ZIP packaging
go 1.25 Backend language
gofiber/fiber v2 HTTP framework
gofiber/limiter v2 Rate limiting middleware
gofiber/cors v2 CORS middleware
joho/godotenv latest Environment loader

🚀 Getting Started

Prerequisites

  • Node.js 20+
  • Go 1.22+
  • Google AI Studio API Key (Gemini)
  • Google Books API Key (optional)

Frontend Installation

cd frontend
npm install
npm run dev

Frontend berjalan di http://localhost:5173

Backend Installation

cd backend
cp .env.example .env
# Edit .env → tambahkan GEMINI_API_KEY dan GOOGLE_BOOKS_API_KEY
go mod tidy
go run ./cmd/api

Backend berjalan di http://localhost:8080

Docker (Production)

docker build -t fuenzer-research .
docker run -p 8080:8080 \
  -e GEMINI_API_KEY=your_key \
  -e GOOGLE_BOOKS_API_KEY=your_books_key \
  -e ENV=production \
  fuenzer-research

📁 Project Structure

/fuenzer-research
├── /frontend                    # React SPA (Vite + TypeScript + Tailwind CSS 4)
│   ├── /public                  # Static assets (favicon, OG image, logos)
│   ├── /src
│   │   ├── /assets              # Logo images (SINTA, Garuda, Scopus, etc.)
│   │   ├── /components
│   │   │   ├── /home            # Landing page components (HeroBackground, etc.)
│   │   │   ├── /playground      # Playground-specific (AIAssistantPanel)
│   │   │   └── /shared          # Reusable (Navbar, Footer, JournalCard, CookieConsent, etc.)
│   │   ├── /hooks               # Custom React hooks (useSEO, etc.)
│   │   ├── /lib                 # Firebase config, Firestore helpers
│   │   ├── /locales             # i18n translations (en.ts, id.ts)
│   │   ├── /pages               # Route pages
│   │   │   ├── /auth            # Authentication pages (Login, SignUp, VerifyEmail, ResetPassword)
│   │   │   ├── LandingPage.tsx
│   │   │   ├── PlaygroundPage.tsx
│   │   │   ├── LibraryPage.tsx
│   │   │   ├── CitationsPage.tsx
│   │   │   ├── TermsPage.tsx
│   │   │   ├── PrivacyPage.tsx
│   │   │   └── NotFoundPage.tsx
│   │   ├── /services            # API client (axios → backend)
│   │   ├── /store               # Zustand stores (research, auth, UI)
│   │   ├── /types               # TypeScript interfaces
│   │   ├── /utils               # Helpers (PDF export, keyword extractor)
│   │   ├── App.tsx              # Root component + React Router
│   │   └── main.tsx             # Entry point
│   ├── package.json
│   └── vite.config.ts
├── /backend                     # Go Fiber REST API
│   ├── /cmd/api/main.go         # Entry point — server setup + middleware
│   ├── /internal
│   │   ├── /config              # Environment variables loader
│   │   ├── /handlers            # HTTP route handlers (research, ask, autocomplete)
│   │   ├── /models              # Go structs (request/response types)
│   │   └── /services
│   │       ├── /gemini          # Google Gemini AI SDK integration
│   │       ├── /openalex        # OpenAlex API client (works, sources, autocomplete)
│   │       ├── /googlebooks     # Google Books API client
│   │       ├── /garuda          # Garuda SQLite local database client
│   │       └── /sinta           # SINTA tier dictionary mapper
│   └── /data                    # Static data (sinta_journals_data.json, garuda.db)
├── /docs                        # Architecture docs, design guidelines, progress log
├── Dockerfile                   # Multi-stage production build
├── DESIGN.md                    # Visual design system specification
├── AGENTS.md                    # AI coding assistant configuration
└── README.md                    # This file

🔒 Security

Layer Implementasi
CORS Strict whitelist: hanya localhost:5173 dan research.fuenzer.web.id
Rate Limiting 15 req/menit (global), 3 req/5 menit (Auth) untuk mencegah email bombing
Security Headers HSTS, X-Frame-Options, CSP, X-Content-Type-Options, Referrer-Policy, Permissions-Policy
Input Validation Query 3-200 karakter, scope validation, type validation
Email Security Validasi Regex RFC 5322, sanitasi CRLF (Header Injection Prevention), log sistem tanpa PII
Gibberish Detection Dual-layer: frontend (vowel check + keyboard mash) + backend (pattern matching)
XSS Prevention DOMPurify untuk sanitasi output Markdown dari AI
SQLi Prevention Parameterized queries pada semua eksekusi database SQLite Garuda
Anti-Prompt Injection System prompt dengan instruksi eksplisit untuk menolak override attempts
Secret Management API keys di backend, file .env di-ignore eksplisit di frontend & root .gitignore
Auth Firebase Auth (Google + Microsoft + Email) + anonymous session fallback

📄 License

Dilindungi di bawah lisensi Apache 2.0.

Made with 🧠 + ☕ for JuaraVibeCoding Season 1 by Google

Fuenzer Research — Accelerating Indonesian Academic Discovery with AI

About

AI Scientific Research Assistant. Find scientific journal references and get instant synthesis with Google Gemini. Built for JuaraVibeCoding Hackathon

research.fuenzer.web.id

Topics

golang gemini google-books garuda cloud-run sinta react-vite openalex ai-scientific-search juaravibecoding

Resources

Readme

License

Apache-2.0 license
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases 2

v0.0.2 Latest
Jul 3, 2026
+ 1 release

Packages

 
 
 

Contributors

Languages