AnalyzingInfixSuggester does not use an FST; it uses a Lucene index. It's faster because 1) it indexes leading ngrams (up to 4 characters by default) so that short suggestions map to a TermQuery (longer suggestions still use PrefixQuery), which is much faster than PrefixQuery, but also 2) it uses impact sorted postings, sorting by suggest weight, so that it can early terminate on pulling suggestions.
Mike McCandless http://blog.mikemccandless.com On Thu, Mar 9, 2017 at 2:26 AM, krish mohan <krishmohan1...@gmail.com> wrote: > hi, > > When using AnalyzingInfixSuggestor, the suggest performance is faster than > the prefix query. How AnalyzingInfixSuggestor is faster than prefixQuery? > Is FST constructed in different way for AnalyzingInfixSuggestor? Or any > other reason? > > > Thanks. >
