New Relativity Cluster Widget

Relativity has long used pivot widgets to allow users to use to create bar graphs, stacked bar graphs, pie charts, line charts, or tables. Relativity 9.4 now includes Cluster widgets. Each cluster can be drilled down into, to view subclusters.

In order to access clusters in Relativity, go to the Document tab and click on the asterisk icon, next to the saved searches magnifying glass icon. You'll then see a list of clusters and an icon giving you the option to visualize them.

After clicking on Add Widget, and then choosing Cluster visualization you can select any cluster set to create an interactive map. You can choose between circle pack and nearby cluster visualizations.

A circle pack users circles which are bigger depending on the number of documents in each cluster. The depth of a circle pack can be increased in order to show subclusters. (There are lower level clusters inside subclusters). Hovering over a cluster will bring up a caption showing matching documents (those within the criteria of a filter or view for the cluster visualization - [conversely - the Document Breakdown pie chart in the corner shows the percentage of listed documents that are found in the visualized cluster set and percentage of listed documents that are not in the visualized cluster set] ); the total number of documents; the number of subclusters; and a coherence score. The coherence score indicates how closely related the documents in the cluster are to each other.

The nearby cluster visualization will display clusters radiating out from a single cluster that you select. The more similar the cluster, the closer it is to the select central cluster. Nearby cluster visualization is accessed by right clicking on any cluster in a circle pack. In the nearby clusters, hovering over any one circle will not only show the same details listed in the circle pack visualization, but also give a similarity score - how close the selected circle is to the main central circle on which the visualization is based.

Apply filters or a view to a cluster visualization will apply 'heat map' shading to the visualization. The circles of the visualization will be colored in shades from dark to light blue in order to show which have more of the terms of the filter.

If there are very few hits in the visualization, you can check off 'Highlight Matches' , and the circles which contain any 'hits' will turn orange.

Note that only one cluster visualization widget can be added to a dashboard at a time.