TUNiBrain–Korean Morphological Analyzer
The next-gen Korean morphological analyzer
4 followers
The next-gen Korean morphological analyzer
4 followers
TUNiBrain is a hybrid Korean morphological analyzer that combines deep learning and dictionary-based NLP to understand meaning, not just tokens. It detects proper nouns, disambiguates words like “눈” (eye vs snow), and outputs definitions, senses, and translations for every token. Built with CRF, Bi-LSTM, and semantic dictionaries, it powers more accurate Korean chatbots, search, and LLM pipelines.
Hey Product Hunters 👋
If you’ve ever built NLP systems for Korean, you already know the ugly truth:
**tokenization alone is not understanding.**
Most analyzers can split text — but they can’t tell whether
**“눈” means an eye or snow**,
**“배” is a fruit, a ship, or a stomach**,
or whether a word is a **person, place, or brand**.
So we built **TUNiBrain**.
TUNiBrain is not just a tokenizer — it’s a **meaning-aware Korean language engine**.
Under the hood, it combines:
- **CRF-based MeCab parsing**
- **Bi-LSTM proper-noun detection**
- **LESK + TF-IDF word-sense disambiguation**
- **A semantic dictionary (TUNiBank)**
This allows it to produce output like:
> “나는 에배레스트에서 배를 먹고 내려왔다”
→ mistyped Everest(에배레스트) is detected as a proper noun
→ “배” is disambiguated as **pear**, not **stomach**
→ each token is assigned **sense ID, definition, origin, and English equivalent**
TUNiBrain
Compared to MeCab-ko or Khaiii, TUNiBrain was built to win where real products fail:
**proper nouns, ambiguity, and meaning**
TUNiBrain
It delivers:
- 🧠 **Word-sense disambiguation (WSD)**
- 🏷 **Proper noun recognition**
- 🔢 **Number detection**
- 📚 **Definitions, origins, translations, semantic & topic tags**
- 🔧 **Retrainable models & customizable dictionaries** TUNiBrain
### 🔓 Open Source, with a Clear Boundary
TUNiBrain is **open source for research and non-commercial use**.
If you’d like to use it in a **commercial product or service**, please contact us — we offer enterprise licenses and support.
• **Attribution required**: Please cite TUNiB Inc. when using this software in academic papers or research publications.
**Commercial inquiries:👉 📩tunibridge@tunib.ai
We already use TUNiBrain inside **TUNiB’s AI Guardrails, chatbots, and enterprise NLP pipelines**, and we’re excited to see what the community builds with it.
If you’re working with Korean text — try it, break it, and tell us what you need.
Let’s make Korean NLP truly intelligent 🧠🚀
— Team TUNiB