Hi Matthew, On 08/18/2008 at 1:39 PM, Matthew Runo wrote: > <fieldType name="text" class="solr.TextField" positionIncrementGap="100"> > <analyzer> > [...] > <filter class="solr.SynonymFilterFactory" synonyms="synonyms.txt" > ignoreCase="true" expand="true"/> > [...]
I can see from SOLR-702 that most of your synonym rules have a single term/phrase on the right-hand side. The SynonymFilterFactory section of the AnalyzersTokenizersTokenFilters wiki page <http://wiki.apache.org/solr/AnalyzersTokenizersTokenFilters#head-2c461ac74b4ddd82e453dc68fcfc92da77358d46> says: # If expand==true, "ipod, i-pod, i pod" is equivalent to the explicit mapping: ipod, i-pod, i pod => ipod, i-pod, i pod AFAICT from looking at the code, however, the "expand" option is ignored when there is an explict right-hand side of a rule (i.e. "=> something"). > a, b, c d e, f, g => something So documents containing "c d e" (or "a" or "b" or "f" or "g") will only be indexed with "something". > I'm having the behavior that searches for a, b, f, and g all > work, but the c d e does not. As Otis mentioned earlier in this thread, the above-linked wiki page mentions some gotchas about mixing phrases, synonyms, and the Lucene QueryParser. Perhaps you could address the problem by creating separate rules for your phrasal terms, e.g.: a, b, f, g => something c d e, something Using the above rule with no right-hand side, and with expand==true, both "c d e" and "something" will be indexed for documents containing "c d e". Steve
