Configure Clusters

You can take a snapshot of the data content that HPE IDOL Server stores. This snapshot identifies clusters of conceptually similar documents, which enables you to generate a view of trends in the data. You do not need to generate an initial taxonomy to take a snapshot.

A set of data can contain a few large clusters or many small clusters, as well as several outliers that are not part of any cluster. Clusters can consist of highly similar documents or of less closely related ones. What constitutes optimal clustering depends on how you intend to use your clusters. However, the aim of clustering is always to generate an accurate characterization of the data content in your HPE IDOL Server.

By default HPE IDOL Server uses internal settings to produce clusters. You do not usually need to change these default settings. However, in some cases you might require more or less detail in your clusters, or the amount and nature of your data might mean that default clustering is not satisfactory.

You can adjust the size of the units on which to base clusters, the degree of conceptual similarity that documents within clusters must have, or the number of clusters to create.