patternmind 1.1.2

A lightweight text processing library for pattern matching, similarity, classification and clustering

# PatternMind

[![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
[![Python](https://img.shields.io/badge/python-3.6+-blue.svg)](https://www.python.org/downloads/)
[![PyPI](https://img.shields.io/badge/pypi-patternmind-orange.svg)](https://pypi.org/project/patternmind/)

**PatternMind** is a lightweight, dependency-free Python library for text processing and pattern learning. It provides powerful tools for similarity calculation, pattern recognition, text classification, clustering, and segmentation — all without requiring external NLP libraries.

## Features

- **Pattern Learning**: Train models to recognize patterns and generate responses (`train_pattern`, `predict_pattern`)
- **Similarity Calculation**: Measure semantic similarity between texts (`predict_similarity`, `train_similarity`)
- **Text Classification**: Train classifiers and predict labels for new texts (`train_class`, `predict_class`)
- **Clustering**: Automatically group similar sentences (`predict_cluster`)
- **Segmentation**: Split sentences into all possible n-grams (`predict_segment`)

## Installation

```bash
pip install patternmind

Quick Start

Similarity Calculation

import patternmind as pm

# Calculate similarity with default model
result = pm.predict_similarity("I love programming", "I enjoy coding")
print(result["result"])  # 85.5

# Train a custom similarity model
model = pm.train_similarity(
    terms1=["I love Python", "I like Python"],
    terms2=["I enjoy Python", "I adore Python"],
    marks=[0.9, 0.8]
)
print(model["alpha"], model["beta"])  # 0.7, 0.3

# Use trained model
score = pm.predict_similarity("I love Python", "I enjoy Python", model)
print(score["result"])  # 87.3

Pattern Learning

import patternmind as pm

# Train a pattern model
model = pm.train_pattern(
    inputs=["What is the capital of France?", "Capital of Germany?"],
    outputs=["Paris", "Berlin"]
)

# Predict response
result = pm.predict_pattern("What is the capital of Italy?", model)
print(result["result"])  # Rome

# Find best matching pattern
best = pm.find_pattern("Capital of Spain?", [model])
print(best["result"])  # 0 (index of matched model)

Text Classification

import patternmind as pm

# Train classifier
model = pm.train_class(
    inputs=["I love this product", "Terrible service"],
    outputs=["positive", "negative"]
)

# Predict
result = pm.predict_class("Great experience!", model)
print(result["result"])  # positive

Clustering

import patternmind as pm

sentences = [
    "I love programming",
    "Coding is fun",
    "Python is great",
    "I hate bugs"
]

clusters = pm.predict_cluster(sentences, least=0.5)
print(clusters["result"])
# [['I love programming', 'Coding is fun', 'Python is great'], ['I hate bugs']]

Segmentation

import patternmind as pm

segments = pm.predict_segment("I love Python")
print(segments["result"])
# ['I', 'I love', 'I love Python', 'love', 'love Python', 'Python']

API Reference

Core Functions

predict_segment(sentence)

Generates all possible n-gram segments of a sentence.

Parameters:

· sentence (str): Input sentence

Returns:

· dict: {"result": list_of_segments}

predict_similarity(term1, term2, model)

Calculates similarity between two texts using a trained model.

Parameters:

· term1 (str): First text · term2 (str): Second text · model (dict): Model with alpha, beta, and synonyms

Returns:

· dict: {"result": similarity_score}

train_similarity(terms1, terms2, marks)

Trains a similarity model by finding optimal alpha and beta weights.

Parameters:

· terms1 (list): List of first texts · terms2 (list): List of second texts · marks (list): List of similarity scores (0-100)

Returns:

· dict: Model with alpha, beta, synonyms, and loss

train_pattern(inputs, outputs)

Trains a pattern recognition model.

Parameters:

· inputs (list): List of input patterns · outputs (list): List of corresponding outputs

Returns:

· dict: Model with pattern, sames, and loss

predict_pattern(text, model)

Predicts output for new text using a trained pattern model.

Parameters:

· text (str): Input text · model (dict): Trained pattern model

Returns:

· dict: {"result": predicted_output}

find_pattern(query, content, least)

Finds the best matching pattern model for a query.

Parameters:

· query (str): Text to match · content (list): List of pattern models · least (int): Minimum score threshold

Returns:

· dict: {"result": best_model_index}

train_class(inputs, outputs)

Trains a text classifier.

Parameters:

· inputs (list): List of texts · outputs (list): List of corresponding labels

Returns:

· dict: Model with tags, words, and loss

predict_class(query, model)

Predicts label for new text using a trained classifier.

Parameters:

· query (str): Text to classify · model (dict): Trained classifier model

Returns:

· dict: {"result": predicted_label}

predict_cluster(sentences, least, model)

Clusters similar sentences together.

Parameters:

· sentences (list): List of texts to cluster · least (float): Similarity threshold (0-1) · model (dict): Similarity model (optional)

Returns:

· dict: {"result": list_of_clusters}

Advanced Usage

Custom Similarity Training

import patternmind as pm

# Train with custom data
model = pm.train_similarity(
    terms1=[
        "I love programming",
        "Python is awesome",
        "Coding makes me happy"
    ],
    terms2=[
        "I enjoy coding",
        "Python is great",
        "Programming brings joy"
    ],
    marks=[0.9, 0.85, 0.8]
)

# Use trained model
score = pm.predict_similarity(
    "I love coding",
    "Programming is fun",
    model
)
print(score["result"])

Building a Chatbot with Pattern Recognition

import patternmind as pm

# Train patterns
patterns = [
    {"input": "What is your name?", "output": "My name is PatternMind"},
    {"input": "How are you?", "output": "I'm doing great, thanks!"},
    {"input": "What can you do?", "output": "I can process text and learn patterns"}
]

inputs = [p["input"] for p in patterns]
outputs = [p["output"] for p in patterns]

# Train model
model = pm.train_pattern(inputs, outputs)

# Chat loop
while True:
    user_input = input("You: ")
    response = pm.predict_pattern(user_input, model)
    print(f"Bot: {response['result']}")

Text Classification with Custom Data

import patternmind as pm

# Training data
texts = [
    "I love this movie", "Great film", "Amazing acting",
    "Terrible movie", "Waste of time", "Boring film"
]
labels = [
    "positive", "positive", "positive",
    "negative", "negative", "negative"
]

# Train classifier
model = pm.train_class(texts, labels)

# Test
test_texts = ["Fantastic movie!", "Horrible experience"]
for text in test_texts:
    result = pm.predict_class(text, model)
    print(f"{text} -> {result['result']}")

Model Persistence

Models can be saved and loaded using JSON serialization:

import json

# Save model
with open("model.json", "w") as f:
    json.dump(model, f)

# Load model
with open("model.json", "r") as f:
    loaded_model = json.load(f)

License

Apache License 2.0. See LICENSE and NOTICE for details.

Links

· Website: patternmind.ir · GitHub: github.com/patternmind/patternmind · Telegram: @patternmind

Contributing

Contributions are welcome! Please feel free to submit issues and pull requests on GitHub.

---

Made with ❤️ by Ehsan Saeedi