hehekev/crypto-slang-dataset

🚀 Crypto Slang & Market Language Dataset

📌 Overview

This dataset is a structured, open-source dataset designed to interpret and standardize crypto-native language. It focuses on translating informal, fast-evolving slang used across platforms (e.g., Twitter/X, Discord, Telegram) into clear, beginner-friendly explanations.

It aims to improve accessibility in crypto by helping new users and non-English speakers better understand conversations, sentiment, and intent within the ecosystem.

AI models often struggle to understand slang like "rekt", "wen moon", or "ngmi". This dataset helps bridge that gap.

Dataset includes harmful/offensive slang for moderation training.

🎯 Motivation

Crypto communication is highly informal and slang-driven, creating a barrier to entry for many users—especially those from under-resourced language backgrounds.

Most existing data is:

Unstructured (tweets, chats, forums) Context-dependent and ambiguous Not optimized for training or educational use

This dataset addresses these gaps by providing clean, labeled, and context-aware data for crypto slang interpretation.

---

🧱 Dataset Structure

Column	Description
text	Original slang or phrase
meaning	Clean explanation
sentiment	positive / negative / neutral
context	usage (trading, meme, community, etc.)
language	language of the phrase
example	real-world usage reflecting natural crypto conversations

---

Data Adaption

This dataset was iteratively refined using Adaptive Data to improve quality, consistency, and usability.

Key Adaptations: Standardized meanings to ensure clarity and simplicity for beginners Enhanced examples to reflect real conversational usage (closer to actual crypto discourse) Improved intent clarity, distinguishing between hype, warnings, and neutral usage Normalized sentiment labeling across all entries Reduced ambiguity in slang terms with multiple meanings

The goal of adaptation was to transform raw, inconsistent slang into a structured, training-ready dataset.

---

🎯 Use Cases

• Train AI models to understand crypto communities
• Sentiment analysis for trading signals
• Web3 chatbots and assistants
• Social media monitoring

---

🚀 Ongoing Improvements

The dataset is actively being expanded and improved:

🔤 Multilingual support (Yoruba, Hausa, Pidgin – in progress) 🔄 Context-dependent meanings (same slang, different intent) ⚖️ Contrastive examples to reduce ambiguity 📈 Expanded coverage of emerging crypto narratives and trends

---

🌍 Why This Matters

Crypto culture evolves fast, but most datasets ignore informal language.

This dataset focuses on:

• Real user behavior
• Community-driven language
• Undocumented slang

---

⚠️ Limitations

• Slang evolves quickly
• Context may vary across communities
• Currently focused on English (multilingual expansion in progress)

---

📊 Example

term: "HODL"
meaning: "Hold onto assets instead of selling"
sentiment: positive
category: trading
example: "I'm just going to HODL through this dip"

---

📜 License

MIT

---

🤝 Contributions

Open to contributions and improvements.

---

Conclusion

This dataset is designed to bridge the gap between crypto-native communication and broader accessibility, making it easier for anyone—regardless of background—to understand and participate in the crypto ecosystem.