Different results when using "Detect Hierarchies" in data-driven category

[blackhead2]

Hi,

I have a question regarding the "detect hierarchies" option in data-driven categories as I get different results and wondering whether this is intended or not?

I have set up a data-driven category with two levels:

For level 1 I used "{File.MD.XMP::Lightroom\hierarchicalSubject\HierarchicalSubject\0|count:true}" to get the number of keywords. As in my test database I only use person names in hierarchical keywords this number represents the number of persons.

For Level 2 I used "XMP::Lightroom\hierarchicalSubject\HierarchicalSubject\0" which represents the names of the persons.

I'm mainly interested in the pictures that show only one person which works as expected as long as "detect hierarchies" is off. The root tree shows the numbers (1,2,3...) und if I colapse node "1" I get a list with pictures that show only one person ordered by person names.

When I switch on "detect hierarchies" for level 2 (person names) I get different results as now IMatch returns e.g. below the "1" node all pictures (also where other persons are shown too) of the persons that are also shown as single persons on pictures. So IMatch creates a list of persons who are shown on pictures without other persons (same as with "detect hierarchies" off) but then shows all pictures of these persons?!

Kind of weird I know but to cut a long story short: Is it inteded that this option affects the result of the data-driven category? From the description I would expect that it only affects the formatting.

Thank you!

Regards

Jens

[Mario]

Detect hierarchies does what it says on the label. It splits the value and then produces hierarchies. It does not consider other levels or anything.

[blackhead2]

Thank you for you reply but if I understand you correct the results (returned pictures) should be the same independent from the "detect hierarchies" option, right?

But if you compare picture 3 and 5 you can see that it is not and even the results on the first level are changing. The only difference of these results is the "detect hierarchies" option used or not at the second level.

[Mario]

I don't understand what you are doing. You are (ab)using the data-driven categories by using a variable which returns a semicolon-separated list of hierarchical keywords and a count function (which is insane, from a performance perspective, just to mention this)  and then you do some stuff and then you split the results returned by that variable somehow and expect some result.

My head is already spinning even after looking at the first screen shot. Even trying to understand what you are doing to the metadata will take time, then figuring out if it can work, how it should work and if the results are correct.

I recommend you file an official bug report so I can schedule time for this for one of the next updates (probably 2018, though).

[blackhead2]

Imatch has so many features and options so I try them. I didn't add all these features And of course I know that this is not a standard data driven category. Nevertheless with the "detect hierarchies" option switched off my strange, weird, insane, etc. category returns exactly what I expected. I will try to simplify this case an will then file a bug report.

By the way. To check the performance I added another 2000 files and to update the category took far less than a second...