It should at least be matching 'Planet Hollywood'. Suspect that this is due to the term 'H' not being present after tokenizing the input text.
- Set
minimum_match
to zero. This allows 'Planet' to match. The missing 'H' doesn't exclude 'Planet Hollywood'.
- 'Adler Planetarium' is still scoring higher than 'Planet Hollywood', likely due to their both being scored the same based on NGrams. 11
- Index
organization_name
twice: astext
andtext_substring
. This lets us give an extra boost to exact matches. 'Planet' now scores higher than 'Planetarium'. - Bring back some more strictness to the
minimum_match
-- setting it at 60% for now, because 'Planet H' introduces a lot of noise from matches on the 'H'.
Idea: this may be catchable using Solr's Spellcheck functionality. Solr 1.4.1 can provide hints for alternate spellings. Solr 3.1 should also be able to auto-collate corrected terms.
- Starting with Spellcheck hints.
- Updating configs as per http://blog.websolr.com/post/2748574298
- Added Spellcheck support to the Sunspot Search DSL
- Showing a simple collated query when there are no results.