Super Search 3.x

Legacy documentation for Super Search 3.x.

Fuzzy Searching

Super Search includes fuzzy searching, which basically means that searches can be set to ignore plurals, match up similar words, and suggest other words based on spelling in search terms. There are settings in the Preferences tab of the Super Search control panel to enable/disable these behaviors. There is a Lexicon utility in the control panel that needs to be built/run for the first time, and then after that, it will maintain itself without interaction. Fuzzy search terms can be weighted by a multiplier to drop them lower down the search results.

You can also indicate how close a fuzzy spelling or plural test needs to be to be considered valid using the fuzzy_distance parameter. The lesser the distance, the closer the match is to the given word. For example, detrt has a distance of 2 to detroit.

It is important to note that all of the below features require the search Lexicon to learn and become smarter over time. In other words, if you have entries that contain the word cotton, the first time a user searches on coton (misspelled), they will not see results for cotton. The next time someone searches for coton, however, will then see results for cotton. This is because the first misspelling of the word needs to be added to the Lexicon database, and can then be referenced in future searches.

Plurals

This will attempt to find plural or singluar versions of the passed terms. Ex: coat = coats, trowsers = trowser. Currently English language specific. Enabled via the preference in the Super Search control panel.

Phonetics

This will attempt to find phonetically similar terms within the search lexicon. Ex: Nolton = Noulton, Caren = Caron = Carren = Charon. Currently English language specific. Enabled via the preference in the Super Search control panel.

Spelling

This will attempt to correct any terms in the search keywords which don't exist within the site. It will attempt to find the most similar and most common term, found via tuned fitness function based on the term specific site content. Suggestions are internally cached and will only be generated once per new invalid term. Be aware the fitness function is quite intensive initally when it needs to build up it's suggestions for the first time. After overhead in usage reduces drastically. Ex: peices might include results for pieces, kite might include results for kind. Enabled via the preference in the Super Search control panel.