### Install SnowNLP via pip

Source: https://context7.com/isnowfy/snownlp/llms.txt

The standard command to install the SnowNLP library into your Python environment.

```bash
pip install snownlp
```

--------------------------------

### TextRank for Custom Summarization using SnowNLP

Source: https://context7.com/isnowfy/snownlp/llms.txt

Provides direct access to TextRank classes (`TextRank`, `KeywordTextRank`) for fine-grained control over keyword and summary extraction. The example demonstrates preparing a document by segmenting and filtering stopwords, then using TextRank to identify top sentences for summarization.

```python
from snownlp import normal, seg
from snownlp.summary.textrank import TextRank, KeywordTextRank

text = '''
机器学习是人工智能的一个分支。
它使用算法来分析数据并学习模式。
深度学习是机器学习的一个子领域。
神经网络是深度学习的核心技术。
'''

# Prepare document: segment text and filter stopwords
sentences = normal.get_sentences(text)
doc = []
for sent in sentences:
    words = seg.seg(sent)
    words = normal.filter_stop(words)
    doc.append(words)

# TextRank for sentence summarization
rank = TextRank(doc)
rank.solve()
top_sentence_indices = rank.top_index(2)
print("Top sentences:")
for idx in top_sentence_indices:
    print(f"  {sentences[idx]}")

```

--------------------------------

### Perform Sentiment Analysis

Source: https://context7.com/isnowfy/snownlp/llms.txt

Provides examples of calculating sentiment scores for Chinese text and using the sentiment module for custom classification.

```python
from snownlp import SnowNLP
from snownlp import sentiment

s1 = SnowNLP('这个产品非常好用，我很喜欢')
print(s1.sentiments)

score = sentiment.classify('服务态度很好，下次还会再来')
print(score)
```

--------------------------------

### Initialize SnowNLP and Access Features

Source: https://context7.com/isnowfy/snownlp/llms.txt

Demonstrates how to initialize the SnowNLP class with Chinese text and access various NLP properties like word segmentation, POS tagging, and sentiment analysis.

```python
from snownlp import SnowNLP

s = SnowNLP('这个东西真心很赞')
print(s.words)
print(s.tags)
print(s.sentiments)
print(s.pinyin)
print(s.sentences)
```

--------------------------------

### Perform Similarity Search with BM25

Source: https://context7.com/isnowfy/snownlp/llms.txt

This snippet shows how to initialize the BM25 algorithm with a corpus of tokenized documents. It allows for calculating term frequency, IDF scores, and similarity rankings between a query and the document collection.

```python
from snownlp.sim.bm25 import BM25
corpus = [['Python', '编程', '语言', '开发'], ['Java', '编程', '语言', '企业'], ['机器学习', 'Python', '数据', '科学'], ['Web', '开发', 'JavaScript', '前端']]
bm25 = BM25(corpus)
print(bm25.f)
print(bm25.idf)
query = ['Python', '开发']
scores = bm25.simall(query)
print(scores)
score = bm25.sim(query, 0)
print(f"Similarity to doc 0: {score}")
```

--------------------------------

### Execute Part-of-Speech Tagging

Source: https://context7.com/isnowfy/snownlp/llms.txt

Demonstrates how to retrieve POS tags for words using the SnowNLP class or the tag module directly.

```python
from snownlp import SnowNLP
from snownlp import tag

s = SnowNLP('我今天去北京旅游')
print(list(s.tags))

words = ['我', '今天', '去', '北京', '旅游']
print(list(tag.tag(words)))
```

--------------------------------

### SnowNLP Initialization and Core Features

Source: https://context7.com/isnowfy/snownlp/llms.txt

The main SnowNLP class provides a unified interface to access NLP features by initializing with a Unicode string.

```APIDOC
## Initialization

### Description
Initializes the SnowNLP object with a Chinese text string to enable various NLP analysis properties.

### Method
Constructor

### Parameters
#### Request Body
- **text** (string) - Required - The Chinese text to process (must be Unicode).

### Request Example
from snownlp import SnowNLP
s = SnowNLP('这个东西真心很赞')

### Response
- **words** (list) - Segmented words
- **tags** (list) - POS tags
- **sentiments** (float) - Sentiment score (0-1)
- **pinyin** (list) - Pinyin representation
```

--------------------------------

### Training Custom Models with SnowNLP

Source: https://context7.com/isnowfy/snownlp/llms.txt

Enables training custom models for word segmentation, POS tagging, and sentiment analysis using user-provided data. The code demonstrates how to train and save models for each of these components using their respective modules (`seg`, `tag`, `sentiment`).

```python
# Training word segmentation model
from snownlp import seg

# Train on custom data file (space-separated words, one sentence per line)
seg.train('custom_seg_data.txt')
seg.save('custom_seg.marshal')

# Load custom model
seg.load('custom_seg.marshal')

# Training POS tagging model
from snownlp import tag

# Train on custom data (word/tag pairs, space-separated)
tag.train('custom_tag_data.txt')
tag.save('custom_tag.marshal')

# Training sentiment analysis model
from snownlp import sentiment

# Train with separate negative and positive example files
sentiment.train('negative_reviews.txt', 'positive_reviews.txt')
sentiment.save('custom_sentiment.marshal')

# After training, modify the data_path in respective __init__.py files
# to point to your custom models, or use load() to switch models at runtime
```

--------------------------------

### Generate Text Summaries

Source: https://context7.com/isnowfy/snownlp/llms.txt

Shows how to use the TextRank algorithm to extract the most important sentences from a document for summarization.

```python
from snownlp import SnowNLP

text = '自然语言处理是计算机科学领域与人工智能领域中的一个重要方向。'
s = SnowNLP(text)
summary = s.summary(3)
for sent in summary:
    print(sent)
```

--------------------------------

### Perform Word Segmentation

Source: https://context7.com/isnowfy/snownlp/llms.txt

Shows how to segment Chinese text into individual words using the SnowNLP class or the lower-level seg module.

```python
from snownlp import SnowNLP
from snownlp import seg

s = SnowNLP('自然语言处理是计算机科学领域与人工智能领域中的一个重要方向')
print(s.words)

words = seg.seg('我爱北京天安门')
print(words)
```

--------------------------------

### Extract Keywords with TextRank

Source: https://context7.com/isnowfy/snownlp/llms.txt

Demonstrates how to extract the most relevant keywords from a text block, with options to limit the count and merge adjacent terms.

```python
from snownlp import SnowNLP

text = '自然语言处理是计算机科学领域与人工智能领域中的一个重要方向。'
s = SnowNLP(text)
print(s.keywords(5))
print(s.keywords(3, merge=True))
```

--------------------------------

### Training Custom Models

Source: https://context7.com/isnowfy/snownlp/llms.txt

Allows training custom models for segmentation, POS tagging, and sentiment analysis.

```APIDOC
## Training Custom Models

SnowNLP allows training custom models for segmentation, POS tagging, and sentiment analysis using your own data.

### Training Word Segmentation Model

#### Method
N/A (Function calls for training and saving)

#### Endpoint
N/A

#### Parameters
- **custom_seg_data.txt** (file path) - File containing custom segmentation data (space-separated words, one sentence per line).

#### Request Example
```python
from snownlp import seg

seg.train('custom_seg_data.txt')
seg.save('custom_seg.marshal')
```

### Training POS Tagging Model

#### Method
N/A (Function calls for training and saving)

#### Endpoint
N/A

#### Parameters
- **custom_tag_data.txt** (file path) - File containing custom POS tagging data (word/tag pairs, space-separated).

#### Request Example
```python
from snownlp import tag

tag.train('custom_tag_data.txt')
tag.save('custom_tag.marshal')
```

### Training Sentiment Analysis Model

#### Method
N/A (Function calls for training and saving)

#### Endpoint
N/A

#### Parameters
- **negative_reviews.txt** (file path) - File containing negative sentiment examples.
- **positive_reviews.txt** (file path) - File containing positive sentiment examples.

#### Request Example
```python
from snownlp import sentiment

sentiment.train('negative_reviews.txt', 'positive_reviews.txt')
sentiment.save('custom_sentiment.marshal')
```

### Loading Custom Models

After training, you can load custom models at runtime.

#### Request Example
```python
# Example for loading segmentation model
from snownlp import seg
seg.load('custom_seg.marshal')
```
```

--------------------------------

### Text Similarity Calculation with SnowNLP

Source: https://context7.com/isnowfy/snownlp/llms.txt

Computes TF-IDF statistics and BM25 similarity scores between documents. SnowNLP must be initialized with a list of word lists (documents) to enable similarity features. The `tf`, `idf`, and `sim` properties/methods are used for these calculations.

```python
from snownlp import SnowNLP

# Initialize with a corpus of documents (each document is a list of words)
docs = [
    ['这篇', '文章', '讲述', '自然语言处理'],
    ['那篇', '论文', '研究', '机器学习'],
    ['这个', '文档', '关于', '深度学习']
]

s = SnowNLP(docs)

# Get term frequencies for each document
print(s.tf)
# Output: [{'这篇': 1, '文章': 1, ...}, {'那篇': 1, ...}, ...]

# Get IDF scores for terms
print(s.idf)
# Output: {'这篇': 0.405, '文章': 0.405, '那篇': 0.405, ...}

# Calculate similarity of a query to all documents
query = ['文章', '自然语言处理']
similarity_scores = s.sim(query)
print(similarity_scores)
# Output: [1.216, 0, 0]  # First document most similar
```

--------------------------------

### Sentence Tokenization using SnowNLP

Source: https://context7.com/isnowfy/snownlp/llms.txt

Splits Chinese text into individual sentences based on common punctuation marks (。！？；) and line breaks. The `sentences` property of SnowNLP handles this task, and the `normal.get_sentences` function provides an alternative method.

```python
from snownlp import SnowNLP

text = '''今天天气很好。我们去公园玩吧！
你觉得怎么样？应该会很有趣。'''

s = SnowNLP(text)
sentences = s.sentences
print(sentences)
# Output: ['今天天气很好', '我们去公园玩吧', '你觉得怎么样', '应该会很有趣']

# Using normal module directly
from snownlp import normal

sents = normal.get_sentences('这是第一句。这是第二句！这是第三句？')
print(sents)
# Output: ['这是第一句', '这是第二句', '这是第三句']
```

--------------------------------

### TextRank for Custom Summarization

Source: https://context7.com/isnowfy/snownlp/llms.txt

Utilizes TextRank classes for fine-grained control over keyword and summary extraction.

```APIDOC
## TextRank for Custom Summarization

Use the TextRank classes directly for fine-grained control over keyword and summary extraction.

### Method
N/A (Class instantiation and method calls)

### Endpoint
N/A

### Parameters
N/A

### Request Example (Sentence Summarization)
```python
from snownlp import normal, seg
from snownlp.summary.textrank import TextRank, KeywordTextRank

text = '''
机器学习是人工智能的一个分支。
它使用算法来分析数据并学习模式。
深度学习是机器学习的一个子领域。
神经网络是深度学习的核心技术。
'''

# Prepare document: segment text and filter stopwords
sentences = normal.get_sentences(text)
doc = []
for sent in sentences:
    words = seg.seg(sent)
    words = normal.filter_stop(words)
    doc.append(words)

# TextRank for sentence summarization
rank = TextRank(doc)
rank.solve()
top_sentence_indices = rank.top_index(2)
print("Top sentences:")
for idx in top_sentence_indices:
    print(f"  {sentences[idx]}")
```

### Response (Sentence Summarization)
#### Success Response (200)
- **Top sentences** (list of strings) - The most relevant sentences from the text.

#### Response Example
```
Top sentences:
  机器学习是人工智能的一个分支。
  深度学习是机器学习的一个子领域。
```

### Request Example (Keyword Extraction - Conceptual)

*Note: The provided code focuses on sentence summarization. Keyword extraction would typically involve using `KeywordTextRank` similarly.* 

```python
# Conceptual example for keyword extraction
# keyword_rank = KeywordTextRank(doc)
# keywords = keyword_rank.solve(2)
# print(keywords)
```
```

--------------------------------

### Keyword Extraction and Summarization

Source: https://context7.com/isnowfy/snownlp/llms.txt

Extracts keywords and summarizes text using the TextRank algorithm.

```APIDOC
## POST /extract

### Description
Extracts key information from the initialized text.

### Parameters
#### Query Parameters
- **limit** (int) - Optional - Number of items to return.
- **merge** (bool) - Optional - Whether to merge adjacent keywords.

### Response
#### Success Response (200)
- **result** (list) - List of strings representing keywords or summary sentences.

### Response Example
['自然语言', '计算机科学', '领域']
```

--------------------------------

### Traditional to Simplified Chinese Conversion using SnowNLP

Source: https://context7.com/isnowfy/snownlp/llms.txt

Converts Traditional Chinese characters to Simplified Chinese using the SnowNLP library. It utilizes a Trie-based maximum matching algorithm. The `han` property provides a direct conversion, and the `normal.zh2hans` function offers an alternative method.

```python
from snownlp import SnowNLP

# Traditional Chinese text
s = SnowNLP('「繁體字」「繁體中文」的叫法在臺灣亦很常見。')

# Convert to Simplified Chinese
simplified = s.han
print(simplified)
# Output: 「繁体字」「繁体中文」的叫法在台湾亦很常见。

# Using normal module directly
from snownlp import normal

text = '計算機科學與人工智能'
print(normal.zh2hans(text))
# Output: 计算机科学与人工智能
```

--------------------------------

### Extract Keywords using KeywordTextRank

Source: https://context7.com/isnowfy/snownlp/llms.txt

This snippet demonstrates how to use the KeywordTextRank class to extract the top N keywords from a document. It requires a pre-processed document and returns the indices of the most relevant terms.

```python
from snownlp.summary import KeywordTextRank
keyword_rank = KeywordTextRank(doc)
keyword_rank.solve()
top_keywords = keyword_rank.top_index(5)
print(f"Top keywords: {top_keywords}")
```

--------------------------------

### Sentence Tokenization

Source: https://context7.com/isnowfy/snownlp/llms.txt

Splits text into individual sentences based on Chinese punctuation marks and line breaks.

```APIDOC
## Sentence Tokenization (sentences property)

Splits text into individual sentences based on Chinese punctuation marks (。！？；) and line breaks.

### Method
N/A (This is a property access, not an API endpoint)

### Endpoint
N/A

### Parameters
N/A

### Request Example
```python
from snownlp import SnowNLP

text = '''今天天气很好。我们去公园玩吧！
你觉得怎么样？应该会很有趣。'''
s = SnowNLP(text)
sentences = s.sentences
print(sentences)
```

### Response
#### Success Response (200)
- **sentences** (list of strings) - A list of strings, where each string is a sentence.

#### Response Example
```
['今天天气很好', '我们去公园玩吧', '你觉得怎么样', '应该会很有趣']
```

## Using normal module directly

### Method
N/A (This is a function call, not an API endpoint)

### Endpoint
N/A

### Parameters
N/A

### Request Example
```python
from snownlp import normal

sents = normal.get_sentences('这是第一句。这是第二句！这是第三句？')
print(sents)
```

### Response
#### Success Response (200)
- **sents** (list of strings) - A list of segmented sentences.

#### Response Example
```
['这是第一句', '这是第二句', '这是第三句']
```
```

--------------------------------

### Text Similarity (tf, idf, sim)

Source: https://context7.com/isnowfy/snownlp/llms.txt

Computes TF-IDF statistics and BM25 similarity scores between documents.

```APIDOC
## Text Similarity (tf, idf, sim)

Computes TF-IDF statistics and BM25 similarity scores between documents. Initialize SnowNLP with a list of word lists (documents) to use similarity features.

### Method
N/A (These are property accesses and a method call, not API endpoints)

### Endpoint
N/A

### Parameters
N/A

### Request Example (Initialization and TF/IDF)
```python
from snownlp import SnowNLP

docs = [
    ['这篇', '文章', '讲述', '自然语言处理'],
    ['那篇', '论文', '研究', '机器学习'],
    ['这个', '文档', '关于', '深度学习']
]

s = SnowNLP(docs)

# Get term frequencies for each document
print(s.tf)

# Get IDF scores for terms
print(s.idf)
```

### Response (TF/IDF)
#### Success Response (200)
- **s.tf** (list of dictionaries) - Term frequencies for each document.
- **s.idf** (dictionary) - Inverse Document Frequency scores for terms.

#### Response Example (TF/IDF)
```
[{'这篇': 1, '文章': 1, ...}, {'那篇': 1, ...}, ...]
{'这篇': 0.405, '文章': 0.405, '那篇': 0.405, ...}
```

### Request Example (Similarity)
```python
query = ['文章', '自然语言处理']
similarity_scores = s.sim(query)
print(similarity_scores)
```

### Response (Similarity)
#### Success Response (200)
- **similarity_scores** (list of floats) - Similarity scores of the query to each document.

#### Response Example (Similarity)
```
[1.216, 0, 0]
```
```

--------------------------------

### Naive Bayes Text Classification with SnowNLP

Source: https://context7.com/isnowfy/snownlp/llms.txt

Implements a Naive Bayes classifier for custom text classification tasks beyond sentiment analysis. The `Bayes` class allows training on custom data, classifying new text, and saving/loading the trained model.

```python
from snownlp.classification.bayes import Bayes

# Create and train a custom classifier
classifier = Bayes()

# Training data: list of [features, label] pairs
training_data = [
    [['苹果', '手机', '电子产品'], 'tech'],
    [['华为', '芯片', '5G'], 'tech'],
    [['足球', '比赛', '冠军'], 'sports'],
    [['篮球', '运动员', 'NBA'], 'sports'],
    [['股票', '投资', '基金'], 'finance'],
    [['银行', '贷款', '利率'], 'finance'],
]

classifier.train(training_data)

# Classify new text
result, probability = classifier.classify(['手机', '芯片', '技术'])
print(f"Category: {result}, Probability: {probability}")
# Output: Category: tech, Probability: ~0.8

# Save and load the model
classifier.save('my_classifier.marshal')
classifier.load('my_classifier.marshal')
```

--------------------------------

### Sentiment Analysis API

Source: https://context7.com/isnowfy/snownlp/llms.txt

Provides sentiment classification using a Naive Bayes model trained on product reviews.

```APIDOC
## GET /sentiment

### Description
Returns a sentiment score between 0 and 1, where higher values indicate positive sentiment.

### Method
Property Access / Method Call

### Response
#### Success Response (200)
- **score** (float) - Sentiment probability (0.0 to 1.0)

### Response Example
0.98
```

--------------------------------

### Chinese Character to Pinyin Conversion using SnowNLP

Source: https://context7.com/isnowfy/snownlp/llms.txt

Converts Chinese characters to their Pinyin romanization using SnowNLP. The `pinyin` property handles the conversion, supporting both pure Chinese text and mixed text with English characters. It employs a Trie-based maximum matching approach.

```python
from snownlp import SnowNLP

s = SnowNLP('中华人民共和国')
pinyin = s.pinyin
print(pinyin)
# Output: ['zhong', 'hua', 'ren', 'min', 'gong', 'he', 'guo']

# Mixed text with Chinese and non-Chinese
s2 = SnowNLP('Hello世界')
print(s2.pinyin)
# Output: ['Hello', 'shi', 'jie']
```

--------------------------------

### Traditional to Simplified Chinese Conversion

Source: https://context7.com/isnowfy/snownlp/llms.txt

Converts Traditional Chinese characters to Simplified Chinese using a Trie-based maximum matching algorithm.

```APIDOC
## Traditional to Simplified Chinese Conversion (han property)

Converts Traditional Chinese characters to Simplified Chinese using a Trie-based maximum matching algorithm.

### Method
N/A (This is a property access, not an API endpoint)

### Endpoint
N/A

### Parameters
N/A

### Request Example
```python
from snownlp import SnowNLP

s = SnowNLP('「繁體字」「繁體中文」的叫法在臺灣亦很常見。')
simplified = s.han
print(simplified)
```

### Response
#### Success Response (200)
- **simplified** (string) - The converted Simplified Chinese text.

#### Response Example
```
「繁体字」「繁体中文」的叫法在台湾亦很常见。
```

## Using normal module directly

### Method
N/A (This is a function call, not an API endpoint)

### Endpoint
N/A

### Parameters
N/A

### Request Example
```python
from snownlp import normal

text = '計算機科學與人工智能'
print(normal.zh2hans(text))
```

### Response
#### Success Response (200)
- **output** (string) - The converted Simplified Chinese text.

#### Response Example
```
计算机科学与人工智能
```
```

--------------------------------

### Pinyin Conversion

Source: https://context7.com/isnowfy/snownlp/llms.txt

Converts Chinese characters to Pinyin romanization using Trie-based maximum matching.

```APIDOC
## Pinyin Conversion (pinyin property)

Converts Chinese characters to Pinyin romanization using Trie-based maximum matching.

### Method
N/A (This is a property access, not an API endpoint)

### Endpoint
N/A

### Parameters
N/A

### Request Example
```python
from snownlp import SnowNLP

s = SnowNLP('中华人民共和国')
pinyin = s.pinyin
print(pinyin)
```

### Response
#### Success Response (200)
- **pinyin** (list of strings) - A list of Pinyin strings for the Chinese characters.

#### Response Example
```
['zhong', 'hua', 'ren', 'min', 'gong', 'he', 'guo']
```

## Mixed text with Chinese and non-Chinese

### Method
N/A (This is a property access, not an API endpoint)

### Endpoint
N/A

### Parameters
N/A

### Request Example
```python
s2 = SnowNLP('Hello世界')
print(s2.pinyin)
```

### Response
#### Success Response (200)
- **pinyin** (list of strings) - A list of Pinyin strings, preserving non-Chinese parts.

#### Response Example
```
['Hello', 'shi', 'jie']
```
```

--------------------------------

### Naive Bayes Text Classification

Source: https://context7.com/isnowfy/snownlp/llms.txt

Uses the Bayes classifier for custom text classification tasks.

```APIDOC
## Naive Bayes Text Classification

The Bayes classifier can be used directly for custom text classification tasks beyond sentiment analysis.

### Method
N/A (Class instantiation, method calls for training, classifying, saving, and loading)

### Endpoint
N/A

### Parameters
N/A

### Request Example (Training)
```python
from snownlp.classification.bayes import Bayes

classifier = Bayes()

training_data = [
    [['苹果', '手机', '电子产品'], 'tech'],
    [['华为', '芯片', '5G'], 'tech'],
    [['足球', '比赛', '冠军'], 'sports'],
    [['篮球', '运动员', 'NBA'], 'sports'],
    [['股票', '投资', '基金'], 'finance'],
    [['银行', '贷款', '利率'], 'finance'],
]

classifier.train(training_data)
```

### Request Example (Classification)
```python
result, probability = classifier.classify(['手机', '芯片', '技术'])
print(f"Category: {result}, Probability: {probability}")
```

### Response (Classification)
#### Success Response (200)
- **result** (string) - The predicted category.
- **probability** (float) - The probability of the predicted category.

#### Response Example
```
Category: tech, Probability: ~0.8
```

### Request Example (Save and Load)
```python
classifier.save('my_classifier.marshal')
classifier.load('my_classifier.marshal')
```
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.