Skip to content

Adam Greenwell

Keyword Match Modes

Lumenare Search offers two keyword match modes to control how search queries are processed and what results are returned.

Current release: 2.2.01

Both modes operate on retained keyword terms after configured stop words and any enabled question interpretation. Matches can occur across indexed fields. Match mode does not require an exact phrase or validate product attributes.

Match Modes

Match ANY (OR) – Default

How it works:

  • Returns results containing at least one of the search keywords
  • Uses OR logic: keyword1 OR keyword2 OR keyword3
  • Broader, more inclusive results

Best for:

  • Blogs and news sites
  • General content sites
  • Exploratory searches
  • Maximizing result count
  • Topic discovery

Example:
Search: "WordPress security plugin"

Returns posts containing:

  • "WordPress" OR
  • "security" OR
  • "plugin"

So a post about "WordPress plugins" would match, even without "security".

Match ALL (AND) – Precision

How it works:

  • Requires every retained keyword in the ordinary exact, prefix, and suffix passes
  • Uses AND logic: keyword1 AND keyword2 AND keyword3
  • Narrower, more precise results

Best for:

  • Product catalogs
  • E-commerce sites
  • Technical documentation
  • Specific queries
  • Queries where every retained term matters

Example:
Search: "WordPress security plugin"

The ordinary passes require:

  • "WordPress" AND
  • "security" AND
  • "plugin"

So a post about "WordPress plugins" would not match those passes without "security". Supplementary synonym and fuzzy passes can add candidates; see the combinations below.

Configuration

Setting Match Mode

  1. Go to Settings → Lumenare Search → Search Quality
  2. Find Keyword Match Mode
  3. Select your preferred mode:
    • Match ANY (OR) – Broader results
    • Match ALL (AND) – Precise results
  4. Click Save Settings

Note: Match mode changes take effect immediately – no reindexing required.

Choosing the Right Mode

Use Match ANY (OR) When:

  • ✅ You want maximum results
  • ✅ Users explore topics broadly
  • ✅ Content is diverse and interconnected
  • ✅ You want related content to appear
  • ✅ Blog or news site
  • ✅ General knowledge base

Use Match ALL (AND) When:

  • ✅ You want a narrower keyword candidate set
  • ✅ Users search for specific products/concepts
  • ✅ Content is well-organized and specific
  • ✅ Each retained term matters to the query
  • ✅ E-commerce or product catalog
  • ✅ Technical documentation

Examples

Query: "WordPress security best practices"

Match ANY (OR):

  • Returns posts about WordPress
  • Returns posts about security
  • Returns posts about best practices
  • Returns posts with any combination

Match ALL (AND):

  • Ordinary passes require all four terms: WordPress, security, best, and practices
  • Much narrower results
  • More precise but fewer results

Query: "blue running shoes"

Match ANY (OR):

  • Can match indexed mentions of blue
  • Can match indexed mentions of running or shoes
  • Very broad, many irrelevant results

Match ALL (AND):

  • Ordinary passes require blue, running, and shoes
  • Terms can appear in different indexed fields
  • Product color or size is not validated as an attribute; use suitable filters for structured requirements

Normal keyword extraction keeps terms longer than two characters. A query such as size 10 does not make 10 a required keyword, so do not use match mode as a numeric attribute filter.

Switching Between Modes

You can switch match modes at any time:

  1. No reindexing required
  2. Changes take effect immediately
  3. Test with your content to see which works better
  4. You can switch back if needed

Advanced: Combining with Other Features

With Stop Words

Configured stop words are removed before matching in either mode:

  • If the is in the saved list, "the WordPress security" searches for "WordPress security"
  • Excluding the does not exclude words such as thermal or theater
  • The saved list affects query matching immediately; reindex to synchronize stored terms or restore formerly excluded words

With Synonyms

Synonyms add a supplementary exact-match lookup for configured group members, with reduced scoring. For car, automobile, vehicle, searching car can consider indexed mentions of the alternatives. This works alongside the selected match mode, but does not guarantee (car OR automobile OR vehicle) AND every other term under All. Candidate limits, score thresholds, and eligibility filters still apply.

With Fuzzy Matching

Fuzzy matching can add typo-tolerant candidates in either mode, such as computr matching computer. This broader pass does not enforce the ordinary All requirement across every original term.

Search understanding in Search & Answers uses rules for supported question wording without an AI call. Optional meaning reranking reorders an existing keyword candidate set; in 2.2.00 it considers at most 200 posts, or 20 for explicitly enabled live meaning search. It cannot discover missing posts when keyword retrieval has no eligible hits. AI answer writing is a separate choice.

Troubleshooting

Too Many Results (Match ANY)

Solutions:

  • Switch to Match ALL for precision
  • Add more stop words
  • Increase minimum characters
  • Tune weights to improve relevance

Too Few Results (Match ALL)

Solutions:

  • Switch to Match ANY for broader results
  • Enable fuzzy matching for typo tolerance
  • Add synonyms for term variations
  • Check if keywords actually exist in content
  • Check the index, active filters, and assigned Search Profile before enabling meaning reranking

Results Not Relevant

Solutions:

  • Try switching match modes
  • Adjust weights (see Tuning Guide)
  • Add stop words for irrelevant terms
  • Configure synonyms for better matching
  • Review phrase boosting settings

Best Practices

Testing Your Choice

  1. Test with Real Queries: Use actual user searches
  2. Compare Results: Try both modes with same queries
  3. Review Analytics: See what users search for
  4. Check Relevance: Are results what users expect?
  5. Iterate: Switch if needed based on feedback

Content-Specific Recommendations

Blogs/News:

  • Start with Match ANY (OR)
  • Better for topic exploration
  • More results = better discovery

E-commerce:

  • Use Match ALL (AND)
  • Precise product matching
  • Users know what they want

Documentation:

  • Try Match ANY (OR) first
  • Switch to Match ALL if too broad
  • Depends on documentation structure

← Back to Documentation Home | Weighting System →