In customer deployments (DLR), searches for TandemX and Tandem-X return different result counts. The root cause is incorrect ordering of the StopFilterFactory in the Solr analysis chain.
When TandemX is tokenized into ["Tandem", "X"], the single-character token X is not removed by the stop filter because it runs before case normalization. Reordering the filters so that case folding happens before stop word filtering ensures consistent behavior regardless of input casing.
The fix and corresponding tests have already been validated in the customer project.
In customer deployments (DLR), searches for
TandemXandTandem-Xreturn different result counts. The root cause is incorrect ordering of the StopFilterFactory in the Solr analysis chain.When
TandemXis tokenized into["Tandem", "X"], the single-character tokenXis not removed by the stop filter because it runs before case normalization. Reordering the filters so that case folding happens before stop word filtering ensures consistent behavior regardless of input casing.The fix and corresponding tests have already been validated in the customer project.