Skip to content

Adam Greenwell

Fuzzy Matching

Fuzzy matching allows users to find results even with misspellings, typos, or slight variations in spelling using configurable Levenshtein distance.

How It Works

Fuzzy matching uses the Levenshtein distance algorithm to find words that are similar to the search term, even if not exactly spelled.

Levenshtein Distance: Number of character insertions, deletions, or substitutions needed to transform one word into another.

Example:

  • Search: "computr"
  • Matches: "computer" (distance: 1 – one character difference)

Configuration

Enabling Fuzzy Matching

  1. Go to Settings → Lumenare Search → Search Quality Settings
  2. Find Fuzzy Matching section
  3. Check "Enable Fuzzy Matching"
  4. Set Fuzzy Match Threshold (1-5)
  5. Click Save Changes

Note: Fuzzy matching works immediately – no reindexing required.

Threshold Settings

Threshold 1 (Strict):

  • Allows 1 character difference
  • Example: "computr" → "computer"
  • Best for: Precision, avoiding false matches

Threshold 2 (Moderate – Recommended):

  • Allows 2 character differences
  • Example: "computr" → "computer", "compter" → "computer"
  • Best for: Balanced typo tolerance

Threshold 3 (Permissive):

  • Allows 3 character differences
  • Example: "compter" → "computer"
  • Best for: High typo tolerance

Threshold 4-5 (Very Permissive):

  • Allows 4-5 character differences
  • Example: "compter" → "computer"
  • Best for: Very forgiving, but may return less relevant results

Use Cases

Typo Tolerance

Common Typos:

  • "computr" → "computer"
  • "WordPres" → "WordPress"
  • "securty" → "security"

Why: Users make typos, especially on mobile devices.

Spelling Variations

Variations:

  • "organize" vs "organise"
  • "color" vs "colour"
  • "center" vs "centre"

Why: Different spelling conventions (US vs UK).

Partial Matches

Partial:

  • "WordPres" → "WordPress"
  • "secur" → "security"

Why: Users may type partial words.

Best Practices

Choosing Threshold

Start with 2:

  • Good balance
  • Handles most typos
  • Maintains relevance

Adjust based on:

  • Content type
  • User behavior
  • Relevance requirements
  • Performance needs

Testing

After enabling:

  1. Test with common typos
  2. Verify relevance maintained
  3. Check for false matches
  4. Adjust threshold if needed

Performance Considerations

Impact:

  • More resource-intensive than exact matching
  • Higher threshold = more processing
  • Consider disabling for very large sites if not needed

Optimization:

  • Use lower threshold if possible
  • Monitor query performance
  • Disable if not needed

Examples

Example 1: Basic Typo

Search: "computr"
Threshold: 2
Matches: "computer" (distance: 1)

Result: Finds content despite typo

Example 2: Multiple Typos

Search: "WordPres"
Threshold: 2
Matches: "WordPress" (distance: 1)

Result: Finds WordPress content

Example 3: Spelling Variation

Search: "organise"
Threshold: 2
Matches: "organize" (distance: 1)

Result: Finds content with either spelling

Positive Effects

User Experience:

  • Finds results despite typos
  • More forgiving search
  • Better mobile experience
  • Reduces zero-result queries

Accessibility:

  • Helps users with spelling difficulties
  • Accommodates different skill levels
  • More inclusive search

Considerations

Relevance:

  • May match slightly irrelevant terms
  • Higher threshold = less precision
  • Test to ensure relevance maintained

Performance:

  • More resource-intensive
  • Higher threshold = slower queries
  • Consider for large sites

Troubleshooting

Fuzzy Matching Not Working

Check:

  1. Fuzzy matching is enabled
  2. Threshold is set correctly
  3. No JavaScript errors
  4. Test with known typos

Solutions:

  • Enable fuzzy matching
  • Set appropriate threshold
  • Fix JavaScript errors
  • Test with content that should match

Too Many False Matches

Symptoms:

  • Irrelevant results appearing
  • Threshold too high

Solutions:

  • Lower threshold (try 1 instead of 2-3)
  • Review matched terms
  • Test with different threshold values
  • Consider disabling if not needed

Not Matching Obvious Typos

Symptoms:

  • Common typos not matching
  • Threshold too low

Solutions:

  • Increase threshold (try 2-3)
  • Test with specific typos
  • Review threshold setting
  • Verify fuzzy matching enabled

Advanced Tips

Combining with Synonyms

Use Together:

  • Synonyms for equivalent terms
  • Fuzzy matching for typos
  • Both improve search coverage

Example:

  • Synonym: "car, automobile"
  • Fuzzy: "automobil" → "automobile"
  • Combined: Better coverage

Content-Specific Settings

Technical Content:

  • Lower threshold (1-2)
  • Precision important
  • Avoid false matches

General Content:

  • Moderate threshold (2)
  • Balance tolerance and relevance
  • Good default

User-Generated Content:

  • Higher threshold (2-3)
  • More typos expected
  • More forgiving

← Back to Documentation Home | Phrase Boosting →