The software recognition catalog is in need of some "spring cleaning." Two areas in particular come to mind:

1) Software manufacturer names could be more consistent. Normalized data is one of the original "claims to fame" for the software recognition, and it's still pretty good. But if you look at, as an example, Adobe titles, you'll see several variations of the Adobe corporate name - I see Adobe, Adobe Systems, Adobe Systems Inc., Adobe Systems, Inc. (added comma), and Adobe Systems Incorporated - all within my 15 device lab configuration. I'm smart enough to know when building a report that I need to use a "contains" operator rather than equal-to, but not everyone may have looked at the data that carefully to notice some data is missing with a too-restrictive filter. I know the recognition business rules call for using what the publisher calls itself, but I would like to suggest there's some room to make things more consistent.

2) Categories/Subcategories have not been assigned consistently within the history of individual products. I had some raw data in front of me while building a custom report and for five version instances of a product there were four with a valid category/subcategory and one that was wildly different and really wrong. And, unfortunately, since this ugliness was assigned to the most recent version in the report, a necessary MAX function will send through the bogus categorization to the customer. Can't remember the exact product but this issue is readily apparent with a quick custom report. Not a hard problem to solve - just takes a bit of time and some sharp eyes.