A new Lucene suggester based on infix matches by Michael McCandless.
From the post:
Suggest, sometimes called auto-suggest, type-ahead search or auto-complete, is now an essential search feature ever since Google added it almost 5 years ago.
Lucene has a number of implementations; I previously described
AnalyzingSuggester
. Since then,FuzzySuggester
was also added, which extendsAnalyzingSuggester
by also accepting mis-spelled inputs.Here I describe our newest suggester:
AnalyzingInfixSuggester
, now going through iterations on the LUCENE-4845 Jira issue.Unlike the existing suggesters, which generally find suggestions whose whole prefix matches the current user input, this suggester will find matches of tokens anywhere in the user input and in the suggestion; this is why it has Infix in its name.
You can see it in action at the example Jira search application that I built to showcase various Lucene features.
Lucene is a flagship open source project. It just keeps pushing the boundaries of its area of interest.