I'm trying to use the Greek stemmer with Apache Solr. If I edit the schema.xml and add the lines:

    <filter class="solr.GreekLowerCaseFilterFactory"/>
    <filter class="solr.GreekStemFilterFactory"/>

in both analyzers (index and query) of the text fieldType, the stemming works without problems and proper results are being returned. However, the suggestions that are created both in the facets and in the autocomplete form, include the stemmed version of the various keywords. I suppose that this makes sense from a developer's point of view, but the stems that are produced have no actual meaning in Greek.

I did a lot of research (I'm not a Solr expert but I'm starting to learn the basic stuff due to this issue) and I'm hitting a wall. I tried to use many variations of the solution that is described in the question: "How to index fulltext fields with and without stemming" but I can't get solr to copy the fields that I'm interested in, into the dynamic fields that I create (I can see the dynamic fields in the solr interface but no content is copied to them). I even tried to use the filter only in the query analyzer and only in the index analyzer (you can guess how that went).

Any ideas on how I can get this to work properly?


This is not a "best practice" answer but it might help you get an acceptable result (at least till you figure out how to solve this):

  • Enable just the GreekLowerCaseFilterFactory in your schema.xml (not the stemmer).
  • Use the NGramFilterFactory to get partial results in your keywords. You can find information on how to do this here.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy