soynlp
Unsupervised Korean Natural Language Processing Toolkits
Decision gist · record as of 2026-08-14
Yes, if you are processing Korean text and need unsupervised word/noun extraction or tokenization without external models. The low install friction and active maintenance make it accessible. However, GPLv3 licensing restricts proprietary use, and the package requires a reasonably large, homogeneous corpus to work well—single documents or mixed-domain text will produce poor results. Not suitable for production systems requiring copyleft-incompatible licensing.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires a corpus of Korean text documents (homogeneous document collection works best); single sentences or heterogeneous document sets produce poor results.
- Low friction install with four well-established scientific dependencies (numpy, scipy, scikit-learn, psutil).
- Package is actively maintained with recent commits and no known vulnerabilities.
License · maintenance · safety
copyleft license (copyleft) — Licensed under GPLv3 (copyleft). Any derivative work or distribution must also be open-source under GPLv3; proprietary use is not permitted.
last release 2019-08-25 (2546 days) · last repo commit 2026-03-10 · 986 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 757,471 downloads/mo, #5,135 on PyPI
Alternatives
Verify before relying
from soynlp.noun import LRNounExtractor_v2
from soynlp.word import WordExtractor
noun_extractor = LRNounExtractor_v2()
nouns = noun_extractor.train_extract(sentences)
word_extractor = WordExtractor(min_frequency=100)
word_extractor.train(sentences)
words = word_extractor.extract()- Whether the package is actively maintained beyond the 2019-08-25 latest release date despite recent repository commits.
- Compatibility with Python versions beyond 3.6 given the classifier listing only 3.6 support.
What it is and what it does
soynlp is an unsupervised Korean NLP toolkit designed to extract linguistic structure from text without pre-trained models or labeled training data. It uses statistical patterns—cohesion scores, branching entropy, and accessor variety—to identify word boundaries, extract nouns, and tokenize sentences. The package includes WordExtractor for discovering new words in a corpus, multiple noun extractors (LRNounExtractor versions 1 and 2, NewsNounExtractor), and tokenizers that split sentences into word sequences based on learned word scores.
The toolkit works by analyzing how characters and substrings co-occur and branch within a corpus, learning left-right (L-R) structure patterns typical of Korean morphology. It is intended for processing document collections from a single domain or genre (news articles, movie reviews, social media) where homogeneous vocabulary and writing style yield better statistical signals. Dependencies are standard scientific Python libraries: numpy, scipy, scikit-learn, and psutil for memory monitoring during training.
Use it for
- Extract new or domain-specific Korean words from a corpus of news articles or social media posts without a pre-trained dictionary.
- Tokenize Korean sentences into word sequences using learned word boundaries from a corpus of similar documents.
- Identify nouns in Korean text by analyzing right-side character patterns (e.g., particles following noun candidates).
- Build a custom Korean tokenizer by combining noun scores with word cohesion scores for domain-specific text.
- Analyze L-R graph structure of Korean words to understand morphological patterns in a specific corpus.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you are processing Korean text and need unsupervised word/noun extraction or tokenization without external models.
The low install friction and active maintenance make it accessible. However, GPLv3 licensing restricts proprietary use, and the package requires a reasonably large, homogeneous corpus to work well—single documents or mixed-domain text will produce poor results. Not suitable for production systems requiring copyleft-incompatible licensing.
Install
soynlp on PyPI
Before you install
Low friction install with four well-established scientific dependencies (numpy, scipy, scikit-learn, psutil). Package is actively maintained with recent commits and no known vulnerabilities.
Requires a corpus of Korean text documents (homogeneous document collection works best); single sentences or heterogeneous document sets produce poor results.
License in practice
Licensed under GPLv3 (copyleft). Any derivative work or distribution must also be open-source under GPLv3; proprietary use is not permitted.
Quickstart
from soynlp.noun import LRNounExtractor_v2
from soynlp.word import WordExtractor
noun_extractor = LRNounExtractor_v2()
nouns = noun_extractor.train_extract(sentences)
word_extractor = WordExtractor(min_frequency=100)
word_extractor.train(sentences)
words = word_extractor.extract()Verify before relying
- Whether the package is actively maintained beyond the 2019-08-25 latest release date despite recent repository commits.
- Compatibility with Python versions beyond 3.6 given the classifier listing only 3.6 support.
Package facts
| License | copyleft license copyleft |
| Python support | Not specified |
| Install friction | Low. Pure-Python wheel |
| Runtime dependencies | 4 packagesnumpypsutilscipyscikit-learn |
| Maintenance | Actively maintained 2,546 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 757,471 / month, #5,135 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | License :: OSI Approved :: GNU General Public License v3 (GPLv3)Operating System :: OS IndependentProgramming Language :: Python :: 3.6 |
Evidence: soynlp-0.0.493-py3-none-any.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “korean word extraction”
- soynlpUnsupervised Korean natural language processing toolkit that extracts…
- pyhwpParses and extracts data from HWP Document Format v5 files, with…
- python-hwpxReads, edits, and creates HWPX documents (Korean word processor…
Give your agent the search over MCP, or paste the wish link into any chat.
More Linguistic packages
Detects and normalizes text encoding from unknown or ambiguous sources, supporting all IANA character sets that Python's core library provides codecs for, with the ability to register custom codecs.
tiktoken is a fast BPE tokenizer that converts text into token sequences compatible with OpenAI models, supporting multiple encoding schemes including o200k_base and model-specific encodings.
Install it if you work with OpenAI APIs or need to understand token boundaries in GPT-family models.
Detects character encoding and language in byte sequences with high accuracy, supporting 99 encodings and returning confidence scores, language tags, and MIME types.
Install it if you need to detect character encoding or language in byte data; the rewrite makes it substantially faster and more accurate than its predecessors.
Converts Unicode text to ASCII by transliterating non-ASCII characters into their closest ASCII equivalents, with no runtime dependencies.
However, if transliteration quality or ongoing maintenance matters, consider unidecode instead despite its GPL-only license.
Lark is a parsing library that builds abstract syntax trees from context-free grammars, supporting multiple parsing algorithms (Earley, LALR(1), CYK) with automatic line and column tracking.
Python bindings to the tree-sitter parsing library, enabling incremental parsing and syntax tree analysis for source code.
See also textblob · jieba3k · nagisa · mecab-ko-dic · konlpy · rake-nltk · pyvi · segtok · mecab-ko · python-mecab-ko