If you want any letter and any possible substring you might be better off breaking every word into single letters with special tokens between words: ie:
the quick brown fox Becomes t h e ZZ q u i c k ZZ b r o w n ZZ f o x then you can do all the single letter searches and multi letter searches turn into phrase searches. Ie: uic (from quick) would be rewritten as "u i c" And so on. This should give you better performance and more predictable results than wildcard searches depending on the size and complexity of your data. Relevancy would be horrible since the tf/idf would always have a common denominator depending on character set but there are ways around that as well. - will -----Original Message----- From: Mike Klaas [mailto:[EMAIL PROTECTED] Sent: Friday, November 30, 2007 7:51 PM To: solr-user@lucene.apache.org Subject: Re: Tips for searching On 30-Nov-07, at 4:43 PM, Dave C. wrote: > > Thanks for the quick response Mike... > Ideally it should match more than just a single character, i.e. > "the" in "weather" or "pro" in "profile" or "000" in "18000". > > Would these cases be taken care of by the StopFilterFactory? No... you are looking for variant of WildcardQuery's. Prefix wildcards are supported (pro* -> profile), but generalize wildcard queries aren't enabled by default. There has been lots of discussion on the list if you do a search. -Mike
