spaCy-PyThaiNLP

This package wraps the PyThaiNLP library to add Thai language support for spaCy.

Features

Support List

• Word segmentation (tokenization)
• Part-of-speech tagging
• Named entity recognition (NER)
• Sentence segmentation
• Dependency parsing
• Word vectors

Table of Contents

• Installation
• Quick Start
• Usage Examples
  • Basic Sentence Segmentation
  • Part-of-Speech Tagging
  • Named Entity Recognition
  • Dependency Parsing
  • Word Vectors
• Configuration
• License

Installation

Prerequisites

• Python 3.9 or higher
• spaCy 3.0 or higher
• PyThaiNLP 3.1.0 or higher

Install via pip

pip install spacy-pythainlp

Quick Start

import spacy
import spacy_pythainlp.core

# Create a blank Thai language model
nlp = spacy.blank("th")

# Add the PyThaiNLP pipeline component
nlp.add_pipe("pythainlp")

# Process text
doc = nlp("ผมเป็นคนไทย แต่มะลิอยากไปโรงเรียนส่วนผมจะไปไหน ผมอยากไปเที่ยว")

# Access sentences
for sent in doc.sents:
    print(sent)
# Output:
# ผมเป็นคนไทย แต่มะลิอยากไปโรงเรียนส่วนผมจะไปไหน
# ผมอยากไปเที่ยว

Usage Examples

Basic Sentence Segmentation

import spacy
import spacy_pythainlp.core

nlp = spacy.blank("th")
nlp.add_pipe("pythainlp")

doc = nlp("ผมเป็นคนไทย แต่มะลิอยากไปโรงเรียนส่วนผมจะไปไหน ผมอยากไปเที่ยว")

# Get sentences
sentences = list(doc.sents)
print(f"Number of sentences: {len(sentences)}")
for i, sent in enumerate(sentences, 1):
    print(f"Sentence {i}: {sent.text}")

Part-of-Speech Tagging

import spacy
import spacy_pythainlp.core

nlp = spacy.blank("th")
nlp.add_pipe("pythainlp", config={"pos": True})

doc = nlp("ผมเป็นคนไทย")

# Print tokens with POS tags
for token in doc:
    print(f"{token.text}: {token.pos_}")

Named Entity Recognition

import spacy
import spacy_pythainlp.core

nlp = spacy.blank("th")
nlp.add_pipe("pythainlp", config={"ner": True})

doc = nlp("วันที่ 15 กันยายน 2564 ทดสอบระบบที่กรุงเทพ")

# Print named entities
for ent in doc.ents:
    print(f"{ent.text}: {ent.label_}")

Dependency Parsing

import spacy
import spacy_pythainlp.core

nlp = spacy.blank("th")
nlp.add_pipe("pythainlp", config={"dependency_parsing": True})

doc = nlp("ผมเป็นคนไทย")

# Print dependency relations
for token in doc:
    print(f"{token.text}: {token.dep_} <- {token.head.text}")

Word Vectors

import spacy
import spacy_pythainlp.core

nlp = spacy.blank("th")
nlp.add_pipe("pythainlp", config={"word_vector": True, "word_vector_model": "thai2fit_wv"})

doc = nlp("แมว สุนัข")

# Access word vectors
for token in doc:
    print(f"{token.text}: vector shape = {token.vector.shape}")
    
# Calculate similarity
token1 = doc[0]  # แมว
token2 = doc[1]  # สุนัข
print(f"Similarity: {token1.similarity(token2)}")

Configuration

You can customize the PyThaiNLP pipeline component by passing a configuration dictionary to nlp.add_pipe():

nlp.add_pipe(
    "pythainlp",
    config={
        "pos_engine": "perceptron",
        "pos": True,
        "pos_corpus": "orchid_ud",
        "sent_engine": "crfcut",
        "sent": True,
        "ner_engine": "thainer",
        "ner": True,
        "tokenize_engine": "newmm",
        "tokenize": False,
        "dependency_parsing": False,
        "dependency_parsing_engine": "esupar",
        "dependency_parsing_model": None,
        "word_vector": True,
        "word_vector_model": "thai2fit_wv"
    }
)

Configuration Options

Parameter	Type	Default	Description
tokenize	bool	False	Enable/disable word tokenization (spaCy uses PyThaiNLP's newmm by default)
tokenize_engine	str	"newmm"	Tokenization engine. See options
sent	bool	True	Enable/disable sentence segmentation
sent_engine	str	"crfcut"	Sentence tokenizer engine. See options
pos	bool	True	Enable/disable part-of-speech tagging
pos_engine	str	"perceptron"	POS tagging engine. See options
pos_corpus	str	"orchid_ud"	Corpus for POS tagging
ner	bool	True	Enable/disable named entity recognition
ner_engine	str	"thainer"	NER engine. See options
dependency_parsing	bool	False	Enable/disable dependency parsing
dependency_parsing_engine	str	"esupar"	Dependency parsing engine. See options
dependency_parsing_model	str	None	Dependency parsing model. See options
word_vector	bool	True	Enable/disable word vectors
word_vector_model	str	"thai2fit_wv"	Word vector model. See options

Important Notes:

• When dependency_parsing is enabled, word segmentation and sentence segmentation are automatically disabled to use the tokenization from the dependency parser.
• All configuration options are optional and have sensible defaults.

Resources

• PyThaiNLP Documentation
• spaCy Documentation
• GitHub Repository
• Issue Tracker

Contributing

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

License

   Copyright 2016-2026 PyThaiNLP Project

   Licensed under the Apache License, Version 2.0 (the "License");
   you may not use this file except in compliance with the License.
   You may obtain a copy of the License at

       http://www.apache.org/licenses/LICENSE-2.0

   Unless required by applicable law or agreed to in writing, software
   distributed under the License is distributed on an "AS IS" BASIS,
   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
   See the License for the specific language governing permissions and
   limitations under the License.