Weired wildcard search with EDisMax handler

Events happening in the community are now at Drupal community events on www.drupal.org.
ygerasimov's picture

I am using Solr 1.4.1 with EDisMax according to http://drupal.org/node/713142

When I do search for part of the word I see results properly. (In my case the search is for *verhuiz*)
But when I do search for bigger part of the work I get no results. (Nex search is for *verhuizin*).

Even I know that the original work is "verhuizing" and it does exist in my documents.

Please find attached responses and details of the Solr.

Can anyone advise what might be wrong?

AttachmentSize
solr_search_result_verhuiz.txt50.75 KB
solr_search_result_verhuizin.txt18.41 KB
solr_info.html30.08 KB

Comments

Debugging interface

gaëlg's picture

You may have a look at http://solr-url:solr-port/solr/core-name-if-any/admin/analysis.jsp?highl..., this will give you a debugging interface for the Solr textual analysis.

thank you very much for help!

ygerasimov's picture

Thank you very much for the tip. It helped me a lot.

The problem was that because of filter SnowballPorterFilter while indexing word 'verhuizining' was modified to 'verhuiz'.

After disabling this filter everything works like a charm. I understand that usage if this filter is very reasonable but... client want all words to be searchable.

Article about filters http://wiki.apache.org/solr/AnalyzersTokenizersTokenFilters

BTW as this filter is very useful, is there any possibility to let this filter 'add' short word forms to the list of searchable words instead of replacing 'long' version of the word?

Copy field

gaëlg's picture

It should be possible. All that is handled in the schema. Maybe you could use a copy field to index the same data source under a different field type. That's what is done for spelling suggestion, which needs unstemmed words. But not sure it will fit your needs.

Lucene, Nutch and Solr

Group organizers

Group categories

Projects

Group notifications

This group offers an RSS feed. Or subscribe to these personalized, sitewide feeds: