Skip to content

Files

Latest commit

searchsolvedsearchsolved
Add comprehensive documentation and professional repository structure
Jun 7, 2025
cf4b5dc · Jun 7, 2025

History

History
This branch is up to date with searchsolved/search-solved-public-seo:main.

keyword-clustering

README.md

Keyword Clustering Tools

Advanced semantic clustering solutions for organizing and analyzing large keyword datasets. These tools use machine learning and natural language processing to group semantically similar keywords, enabling more effective content strategy and SEO planning.

Tools Overview

🧠 Semantic Clustering Suite

A comprehensive collection of keyword clustering tools powered by SentenceTransformers and various clustering algorithms.

Features

  • Multiple Clustering Methods: Standard clustering and HDBScan for large datasets
  • Semantic Understanding: Uses sentence embeddings for true semantic similarity
  • CLI & Notebook Versions: Choose your preferred interface
  • Visualization: Generate treemaps and sunburst charts for cluster analysis
  • Configurable Parameters: Adjust similarity thresholds, models, and output formats

Available Versions

  • CLI Application: Command-line interface for production workflows
  • HDBScan CLI: Optimized for very large keyword datasets
  • Python Script: Direct integration into existing workflows
  • Jupyter Notebooks: Interactive analysis and experimentation

Clustering Algorithms Supported

  • Standard semantic clustering with similarity thresholds
  • HDBScan for density-based clustering of large datasets
  • Configurable minimum similarity parameters

Quick Start

Basic Usage

cd semantic-clustering/semantic-clustering-cli-app/CLI
python cluster.py keywords.csv --column_name "Keyword" --output_path "output.csv"

Advanced Configuration

python cluster.py keywords.csv \
  --column_name "Keyword" \
  --output_path "clustered_keywords.csv" \
  --chart_type "sunburst" \
  --device "cpu" \
  --model_name "all-MiniLM-L6-v2" \
  --min_similarity 0.80 \
  --remove_dupes True \
  --volume "Volume" \
  --stem True

For Large Datasets (10k+ keywords)

cd semantic-clustering/semantic-clustering-cli-app/CLI-HDBScan
python cluster-hdbscan.py large_keywords.csv

Use Cases

🎯 Content Strategy

  • Group keywords into topical clusters for content planning
  • Identify content gaps and opportunities
  • Plan content pillars and supporting pages

📊 SEO Analysis

  • Analyze competitor keyword strategies
  • Group keywords by search intent
  • Optimize content for semantic keyword groups

🔍 PPC Campaign Organization

  • Create logical ad groups based on semantic similarity
  • Improve Quality Scores through better keyword grouping
  • Reduce keyword cannibalization

📈 Content Optimization

  • Identify related keywords for existing content
  • Optimize for semantic search and entities
  • Improve topical authority

Input Requirements

  • CSV file with keyword data
  • Keyword column containing the terms to cluster
  • Optional volume column for weighted analysis
  • Minimum 50+ keywords for meaningful clustering

Output Formats

  • CSV file with cluster assignments
  • Excel file with pivot tables
  • Interactive visualizations (treemap/sunburst charts)
  • Cluster statistics and similarity scores

Technical Specifications

Models Supported

  • all-MiniLM-L6-v2 (default, balanced speed/accuracy)
  • all-mpnet-base-v2 (higher accuracy, slower)
  • distilbert-base-nli-stsb-mean-tokens
  • Custom SentenceTransformer models

Hardware Requirements

  • CPU: 4+ cores recommended for large datasets
  • RAM: 8GB minimum, 16GB+ for large datasets
  • GPU: Optional CUDA support for faster processing
  • Storage: Varies by dataset size

Dependencies

pip install sentence-transformers pandas numpy plotly scikit-learn hdbscan

Legacy Versions

The repository includes several legacy implementations for reference and backward compatibility:

  • Google Colab versions for cloud processing
  • Search Engine Journal optimized versions
  • Historical clustering approaches

Performance Guidelines

Dataset Size Recommendations

  • < 1,000 keywords: Standard CLI version
  • 1,000 - 10,000 keywords: CLI with optimized settings
  • 10,000+ keywords: HDBScan version
  • 100,000+ keywords: Contact for enterprise solutions

Processing Time Estimates

  • 1,000 keywords: ~2-5 minutes
  • 10,000 keywords: ~10-30 minutes
  • 50,000 keywords: ~1-3 hours (HDBScan)

Support & Documentation

For detailed implementation guides and advanced configurations, visit the tool-specific directories. Each version includes comprehensive documentation and example usage.

Author

Lee Foot - SEO Consultant specializing in semantic search and content optimization.

Part of the Search Solved Public SEO toolkit - Advanced clustering for modern SEO workflows.