While I like Mr. Arnold’s highlights and commentary, I also encourage readers to review the source article, Scanning and Selecting Enterprise Search Results: Not as Easy as it Looks. I can see some interesting parallels to how we test and tune preservation/collection scope criteria. These issues with presentation of search or cluster results came up in my recent briefing with the Agnes Intelligence team. They were rightfully proud of their success at generating meaningful cluster names. If you are familiar with the ubiquitous concept wheel visualization commonly found in analytic platforms, you know that the top 3-5 terms do not always tell a meaningful story.