Partition Documents into Clusters

Partition clustering is a different method of clustering that allows you to create a specified number of clusters, and to place all documents into a cluster. This method is faster than using ClusterSnapshot.

This method adds all documents into a cluster, which means that the clusters generated are relatively broad and might represent several topics. You might want to use this method if you want to cluster the results of a query into a predefined number of clusters.

To partition documents into clusters

  • Send the ClusterPartition action to the IDOL Category component. Set the following action parameters:

    ClusterSGDataGen The query that you want to cluster the results for.
    ClusterSGDataGen The number of results from the specified query that you want to return and cluster.
    NumClustersClusterSGDataGen The number of clusters to create.