diff options
Diffstat (limited to 'solr-8.1.1/example/example-DIH/solr/mail/conf/clustering')
3 files changed, 62 insertions, 0 deletions
diff --git a/solr-8.1.1/example/example-DIH/solr/mail/conf/clustering/carrot2/kmeans-attributes.xml b/solr-8.1.1/example/example-DIH/solr/mail/conf/clustering/carrot2/kmeans-attributes.xml new file mode 100644 index 000000000..d802465f6 --- /dev/null +++ b/solr-8.1.1/example/example-DIH/solr/mail/conf/clustering/carrot2/kmeans-attributes.xml @@ -0,0 +1,19 @@ +<!-- + Default configuration for the bisecting k-means clustering algorithm. + + This file can be loaded (and saved) by Carrot2 Workbench. + http://project.carrot2.org/download.html +--> +<attribute-sets default="attributes"> + <attribute-set id="attributes"> + <value-set> + <label>attributes</label> + <attribute key="MultilingualClustering.defaultLanguage"> + <value type="org.carrot2.core.LanguageCode" value="ENGLISH"/> + </attribute> + <attribute key="MultilingualClustering.languageAggregationStrategy"> + <value type="org.carrot2.text.clustering.MultilingualClustering$LanguageAggregationStrategy" value="FLATTEN_MAJOR_LANGUAGE"/> + </attribute> + </value-set> + </attribute-set> +</attribute-sets> diff --git a/solr-8.1.1/example/example-DIH/solr/mail/conf/clustering/carrot2/lingo-attributes.xml b/solr-8.1.1/example/example-DIH/solr/mail/conf/clustering/carrot2/lingo-attributes.xml new file mode 100644 index 000000000..5febfc320 --- /dev/null +++ b/solr-8.1.1/example/example-DIH/solr/mail/conf/clustering/carrot2/lingo-attributes.xml @@ -0,0 +1,24 @@ +<!-- + Default configuration for the Lingo clustering algorithm. + + This file can be loaded (and saved) by Carrot2 Workbench. + http://project.carrot2.org/download.html +--> +<attribute-sets default="attributes"> + <attribute-set id="attributes"> + <value-set> + <label>attributes</label> + <!-- + The language to assume for clustered documents. + For a list of allowed values, see: + http://download.carrot2.org/stable/manual/#section.attribute.lingo.MultilingualClustering.defaultLanguage + --> + <attribute key="MultilingualClustering.defaultLanguage"> + <value type="org.carrot2.core.LanguageCode" value="ENGLISH"/> + </attribute> + <attribute key="LingoClusteringAlgorithm.desiredClusterCountBase"> + <value type="java.lang.Integer" value="20"/> + </attribute> + </value-set> + </attribute-set> +</attribute-sets> diff --git a/solr-8.1.1/example/example-DIH/solr/mail/conf/clustering/carrot2/stc-attributes.xml b/solr-8.1.1/example/example-DIH/solr/mail/conf/clustering/carrot2/stc-attributes.xml new file mode 100644 index 000000000..c1bf110c8 --- /dev/null +++ b/solr-8.1.1/example/example-DIH/solr/mail/conf/clustering/carrot2/stc-attributes.xml @@ -0,0 +1,19 @@ +<!-- + Default configuration for the STC clustering algorithm. + + This file can be loaded (and saved) by Carrot2 Workbench. + http://project.carrot2.org/download.html +--> +<attribute-sets default="attributes"> + <attribute-set id="attributes"> + <value-set> + <label>attributes</label> + <attribute key="MultilingualClustering.defaultLanguage"> + <value type="org.carrot2.core.LanguageCode" value="ENGLISH"/> + </attribute> + <attribute key="MultilingualClustering.languageAggregationStrategy"> + <value type="org.carrot2.text.clustering.MultilingualClustering$LanguageAggregationStrategy" value="FLATTEN_MAJOR_LANGUAGE"/> + </attribute> + </value-set> + </attribute-set> +</attribute-sets> |