: I read some documentation on the WordDelimterFilter. Just to clarify my : thinking, I understand that if I use WordDelimiterFilter and search for a : term like axd100 it will break it into two tokens "axd" and "100". But then : when I do my search should Solr match the documents containing both these : tokens?
Text Analysis for the purposes of building and searching an inverted index like Lucene is all about having "complimentary" tokenization/filtering at indextime and at query time. Yes, WordDelimiterFilter can help match on the types of things in your example, but only using it to analyzer your query strings will not provide a magic bullet -- you have to have also used it with the appropriate settings at index time in order for the correct tokens to be indexed so you can find them when you query. : In my application when I try to search for "axd 100" I get several : documents back, but when I search for axd100 with WordDelimiterFilter on, I : don't get back any results. I was assuming that if WordDelimiterFilter : breaks axd100 into two tokens - "axd" and "100", then the search should : behave exactly as if I was searching for the string "axd 100". Not exactly, a better approximation would be searching for the *phrase* "axd 100" ... but it really depends on how you configure it. Bottom line: if you don't have control over how your index is built, and it isn't build using any configuration of the WordDelimiterFilter, you are better off avoiding it at query time. -Hoss
