Question

it is said that the Carrot integration into Solr was improved since the release of Solr 3.2 but it turns out to be different for me. I had a absolutly same configurated Solr 1.4.1 Server running were Carrot was working great and Solr 3.2 just gives me nothing but “other topics”. This ist driving me crazy because beside I get no exceptions or anything unusual. Even the result xml looks the same…

However I didn’t make many changes to the standard configuration of the clustering component:

 <searchComponent name="clustering" 
                   enable="${solr.clustering.enabled:true}"
                   class="solr.clustering.ClusteringComponent" >
    <lst name="engine">
      <str name="name">default</str>

      <str name="carrot.algorithm">org.carrot2.clustering.lingo.LingoClusteringAlgorithm</str>

      <str name="LingoClusteringAlgorithm.desiredClusterCountBase">20</str>
          <!--custom-->
      <str name="LingoClusteringAlgorithm.phraseLabelBoost">8.00</str>
      <str name="TermDocumentMatrixBuilder.titleWordsBoost">6.00</str>


      <str name="carrot.lexicalResourcesDir">clustering/carrot2</str>

      <str name="MultilingualClustering.defaultLanguage">ENGLISH</str>
    </lst>
    <lst name="engine">
      <str name="name">stc</str>
      <str name="carrot.algorithm">org.carrot2.clustering.stc.STCClusteringAlgorithm</str>
    </lst>
  </searchComponent>
  <requestHandler name="/clustering"
                  startup="lazy"
                  enable="${solr.clustering.enabled:true}"
                  class="solr.SearchHandler">
    <lst name="defaults">
      <bool name="clustering">true</bool>
      <str name="clustering.engine">default</str>
      <bool name="clustering.results">true</bool>
       <str name="carrot.title">autocomplete</str>
       <str name="carrot.url">autocomplete</str>
       <str name="carrot.snippet">autocomplete</str>
       <bool name="carrot.outputSubClusters">true</bool>

       <str name="defType">edismax</str>
       <str name="qf">
          text^0.5 autocomplete^1.2 ata^1.0 raum^1.0 system^1.0 assy^1.0 unit^1.0
       </str>
       <str name="q.alt">*:*</str>
       <str name="rows">10</str>
       <str name="fl">*,score</str>
    </lst>     
    <arr name="last-components">
      <str>clustering</str>
    </arr>
  </requestHandler>

My best guess was that carrot is not woking properly together with edismax (which wasn’t implemented in Solr 1.4.1) but that might be missleading.

I allready reindexed my data just to make sure that this is not the issue.

In the carrot workbench clustering is working well with Lingo as the algorithm. when I chose “by source” I get “other topics” as in the xml. Might Lingo be not configured well? Do have to configure anything besides solrconfig.xml to fix that?

I’m thankful for any help.

Answer 1

This happens if the ‘snippet’ you are trying to cluster on never differs or differs very little. Try adding ‘clustering.snippet=’ to your request parameters. In your settings there is a field called ‘autocomplete’ that it defaults to. Does this field have any meaningful text?

Example that makes this behaviour go away for me:

http://localhost:8983/solr/clustering?q=peter&rows=200&carrot.snippet=summary

Best regards,

/Peter W