Generate Dynamic Clusters

Every time it runs a query, the IDOL Content component can automatically cluster the results that are available for this query, and then in turn cluster the first set of clusters further to produce subclusters. This process allows you to generate a hierarchy of dynamic clusters that allows users to navigate quickly to their area of interest.

For example:

In this example, Content has generated 10 dynamic clusters from documents that match the query War In Iraq. A plus sign (+) and a red arrow next to a cluster in this interface indicates that you can expand the cluster to display subclusters. A blue arrow indicates clusters that cannot cluster further.

Content lists each cluster with the best phrase it contains. The number in brackets indicates how many documents the cluster contains. The user can click the best phrase for a cluster to run a query with this phrase and automatically cluster the results that this query produces.

Dynamic clustering can be faster than clustering in the background, through ClusterCluster and ClusterServe2DMap. However, using ClusterCluster and ClusterServe2DMap is more powerful and accurate. Also, dynamic clustering is more suited to finding trends in smaller data sets. ClusterServe2DMap is better for clustering an entire database.