​

Similarity Search

Compare and rank similar strings using Levenshtein distance algorithm for fuzzy string matching and text similarity analysis.

​

Overview

The Similarity Search skill provides functionality for:

• Finding closest string matches using Levenshtein distance
• Ranking results by similarity score
• Smart extraction for path-like structures
• Configurable result filtering and limits
• Pattern exclusion for refined matching

​

Connection Requirements

This skill uses an internal helper service and doesn’t require external connections.

​

Basic Usage

// Simple similarity search
const searchRequest = {
  "searchTerm": "example search",
  "items": [
    "example text",
    "sample search",
    "different content",
    "example search match"
  ],
  "maxResults": 3
};

​

Key Features

​

String Matching

• Levenshtein Distance: Calculate edit distance between strings
• Ranked Results: Sort matches by similarity score (lower is better)
• Configurable Limits: Control maximum number of results returned
• Full String Comparison: Compare entire strings for similarity

​

Smart Extraction

• Path Processing: Extract relevant parts from path-like structures
• Pattern Exclusion: Exclude specific patterns during extraction
• Custom Splitting: Configure split characters for path parsing
• Intelligent Matching: Focus on meaningful path components

​

Common Operations

​

Basic Similarity Search

POST: /similarity-search
{
  "searchTerm": "lorem ipsum text",
  "items": [
    "lorem ipsum sample",
    "different content",
    "ipsum lorem variation",
    "completely unrelated"
  ],
  "maxResults": 3
}

​

Smart Extraction for Paths

POST: /similarity-search
{
  "searchTerm": "product catalog",
  "items": [
    "website/pages/product-catalog/main.html",
    "website/pages/user-profile/settings.html",
    "website/pages/product-catalog-v2/index.html",
    "mobile/screens/catalog/product-list.js"
  ],
  "useSmartExtraction": true,
  "excludePatterns": ["html", "js", "v2"],
  "splitValue": "/",
  "maxResults": 5
}

​

Configuration Options

​

Required Parameters

• searchTerm: The string you want to find matches for
• items: Array of strings to search through

​

Optional Parameters

• useSmartExtraction (default: false): Enable path-aware extraction
• excludePatterns (default: []): Patterns to exclude during extraction
• splitValue (default: ’/’): Character for splitting paths
• maxResults (default: 5): Maximum results to return

​

Response Structure

​

Basic Response

{
  "searchTerm": "example search",
  "results": [
    {
      "item": "example search match",
      "distance": 5
    },
    {
      "item": "example text",
      "distance": 8
    },
    {
      "item": "sample search",
      "distance": 10
    }
  ],
  "config": {
    "useSmartExtraction": false,
    "excludePatterns": [],
    "splitValue": "",
    "maxResults": 3
  }
}

​

Smart Extraction Response

{
  "searchTerm": "product catalog",
  "results": [
    {
      "item": "website/pages/product-catalog/main.html",
      "distance": 0
    },
    {
      "item": "website/pages/product-catalog-v2/index.html",
      "distance": 2
    },
    {
      "item": "mobile/screens/catalog/product-list.js",
      "distance": 8
    }
  ],
  "config": {
    "useSmartExtraction": true,
    "excludePatterns": ["html", "js", "v2"],
    "splitValue": "/",
    "maxResults": 5
  }
}

​

Distance Scoring

​

Levenshtein Distance

• 0: Exact match
• 1-5: Very similar (minor differences)
• 6-10: Moderately similar
• 11-20: Somewhat similar
• 21+: Low similarity

​

Interpretation

• Lower distance scores indicate higher similarity
• Distance represents minimum number of edits needed to transform one string into another
• Edits include insertions, deletions, and substitutions

​

Smart Extraction Use Cases

​

File Paths

// Finding similar file paths
"components/Button.tsx" → matches "src/components/Button.tsx"

​

URL Paths

// Finding similar routes
"api/users/profile" → matches "api/v1/users/profile"

​

Category Hierarchies

// Finding similar categories
"electronics/phones" → matches "category/electronics/smartphones"

​

Important Notes

• Case Sensitivity: Matching is case-sensitive by default
• Unicode Support: Handles Unicode characters properly
• Performance: Optimized for moderate-sized string arrays
• Memory Usage: Consider memory limits for very large item arrays
• Smart Extraction: Only use when dealing with path-like structures

​

Best Practices

1. Appropriate Limits: Set reasonable maxResults to avoid overwhelming responses
2. Smart Extraction: Only enable for path-like data structures
3. Pattern Exclusion: Use excludePatterns to filter out noise (file extensions, version numbers)
4. Preprocessing: Clean input data for better matching results
5. Threshold Filtering: Consider filtering results by distance threshold
6. Performance: For large datasets, consider batching requests
7. Validation: Validate input arrays are not empty before processing