feat: Upload multilingual PPMB QA dataset and update the dataset card

README.md

@@ -1,3 +1,137 @@
 ---
 license: mit
+task_categories:
+- question-answering
+language:
+- en
+- id
+- jv
+tags:
+- qa
+- helpdesk
+- chatbot
+- university
+- education
+- ppmb
+- pmb
+- admission
+pretty_name: PPMB QA Dataset
+size_categories:
+- 1K<n<10K
 ---
+### Overview
+
+The **PPMB QA Dataset** is a multilingual question-answering dataset specifically designed for university admission helpdesk and chatbot applications. It focuses on the **New Student Admission (Penerimaan Mahasiswa Baru / PMB)** process at **UPN "Veteran" Jawa Timur** (Universitas Pembangunan Nasional "Veteran" East Java), Indonesia, for the academic year 2025/2026.
+
+---
+
+### Dataset Statistics
+
+| Attribute | Value |
+|-----------|-------|
+| **Total QA Pairs** | 1,059 pairs (353 per language × 3 languages) |
+| **Languages** | English (en), Indonesian (id), Javanese (jv) |
+| **File Format** | JSONL (JSON Lines) |
+| **License** | MIT |
+| **Size Category** | 1K < n < 10K |
+
+---
+
+### Language Distribution
+
+1. **ppmb-qa-en.jsonl** – 353 entries in **English**
+2. **ppmb-qa-id.jsonl** – 353 entries in **Indonesian**
+3. **ppmb-qa-jv.jsonl** – 353 entries in **Javanese**
+
+The Javanese version uses authentic regional language patterns including informal expressions (e.g., "cak" for "bro", "awakmu" for "you", "iso" for "can"), making it suitable for localized chatbot development targeting East Javanese speakers.
+
+---
+
+### Content Domain & Topics
+
+The dataset covers comprehensive information about university admission, including:
+
+#### 1. **Admission Pathways (Jalur Penerimaan)**
+- **SNBP** (Seleksi Nasional Berdasarkan Prestasi) – National Achievement-Based Selection
+- **SNBT** (Seleksi Nasional Berdasarkan Tes) – National Test-Based Selection (UTBK)
+- **Seleksi Mandiri** (Independent Selection):
+  - SM-Rapor (Report Card-based)
+  - SM-UTBK (UTBK Score-based)
+  - SM-Prestasi (Achievement-based)
+  - SM-CBT (Computer-Based Test)
+
+#### 2. **Program-Specific Information**
+- Detailed info for various study programs (Agrotechnology, Communication Studies, Visual Communication Design, Accounting, Medicine, Mechanical Engineering, etc.)
+- Intake capacity (daya tampung) per program
+- Competition ratios (tingkat keketatan)
+- Supporting subjects requirements
+- Portfolio requirements for art programs
+
+#### 3. **Tuition & Financial Aid**
+- **UKT** (Uang Kuliah Tunggal) – Single Tuition Fee across 7 categories
+- **IPI** (Iuran Pengembangan Institusi) – Institutional Development Fee for independent pathway
+- **KIP Kuliah** (Kartu Indonesia Pintar) – Government scholarship for underprivileged students
+
+#### 4. **Special Programs**
+- **Golden Ticket Award** – Recognition program for high-achieving 12th-grade students with national/international achievements
+- Cross-major enrollment policies
+- Partnership/cooperation pathways (TNI, Polri, alumni)
+
+#### 5. **Administrative Information**
+- Registration schedules and deadlines
+- Required documents
+- Contact information (Instagram, email, Telegram groups)
+- PDSS (School Database) requirements
+
+#### 6. **Career Outcomes**
+- Graduate employment statistics
+- Job-education alignment data
+- Time-to-employment metrics
+
+---
+
+### Data Structure
+
+Each entry follows this JSON schema:
+```json
+{
+  "query": "User's question (natural language)",
+  "answer": "Comprehensive answer with structured information, bullet points, and details"
+}
+```
+
+**Characteristics of the Q&A pairs:**
+- **Queries**: Natural, conversational questions as prospective students would ask (including informal expressions, emojis, social media mentions like @ppmb.upnjatim)
+- **Answers**: Detailed, well-structured responses with:
+  - Markdown formatting (headers, bullet points, bold text)
+  - Specific data (dates, numbers, fees)
+  - Official sources and contact information
+  - Contextual explanations
+
+---
+
+### Use Cases
+
+1. **Chatbot Development** – Train conversational AI assistants for university admission inquiries
+2. **Multilingual NLP Research** – Low-resource language (Javanese) question-answering
+3. **Information Retrieval** – Domain-specific retrieval systems for educational institutions
+4. **Machine Translation** – Parallel corpus for EN-ID-JV translation tasks
+5. **Helpdesk Automation** – Automated FAQ systems for university administration
+6. **Fine-tuning LLMs** – Domain adaptation for Indonesian higher education context
+
+---
+
+### Unique Features
+
+- **Trilingual parallel data**: Same questions/answers across three languages enable cross-lingual research
+- **Low-resource language inclusion**: Javanese is an underrepresented language in NLP resources
+- **Real-world informal queries**: Questions capture authentic student communication styles including slang, abbreviations, and social media conventions
+- **Comprehensive domain coverage**: Covers virtually all aspects of university admission process
+- **Up-to-date information**: Based on 2025/2026 academic year admission data
+- **Institutional specificity**: Focused on one university but applicable patterns for similar Indonesian universities
+
+---
+
+### Tags
+
+`qa` · `helpdesk` · `chatbot` · `university` · `education` · `ppmb` · `pmb` · `admission`