iwmtresolveuids: need -grab parent or -parent parent to work

UPDATE
Here's the context: Using a recognizer, we want to add up all of the scores that MT is returning and assign them to the second-level nodes in our taxonomy.

To illustrate, imagine a 3 level taxonomy where the top level node is "car", the second level node is "ford" and below that we have "mustang", "fairlane", and "fiesta" (which are all models of ford).

If "mustang" has a score of 1, "fairlane" has a score of 2 and "fiesta" has a score of 3 we want to add all of those scores up and give "ford" a score of 6.

Our second-level nodes are identified by a special vocab so we can identify them. I had thought that we could use iwmtresolveuids -implicit_add to get all the parent nodes and then we could walk the tree to add up the scores but iwmtresolveuids doesn't include elements in the descriptors even when you use the -all flag, so we can't walk the tree.

Then I thought I could put the parent information in some other element and use the -all flag but the -all flag doesn't seem to apply to nodes that are added by implicit_add, just the nodes that were originally included by MT.

From the reading I've done it looks like the only way to get the parent information is using the java api.

[Here's the original message]

I'm trying to use iwmtresolveuids to return the parent element along with the standard elements. I've tried -grab and I've tried -parent but neither is working. Is this syntax correct?

iwmtresolveuids -grab parent -implicit_add "PMO Systems and Topics"

I tried this too:

iwmtresolveuids -parent parent -implicit_add "PMO Systems and Topics"

The implicit_add flag is working correctly.

Find more posts tagged with

Comments

Ottawa_IWOV

Yikes! I remember this functionality being quite easy in the 3.x release. So, Interwoven does not ship with a standard way of getting the parent? That is completely ridiculous.

Could someone from the Interwoven MetaTagger product group (ahem, Clark Breyman), possibly provide an example of this functionality, or atleast a starting point?

Thanks.

Migrateduser

Getting the lineage of a node is an argument to the iwmtresolveUID.exe. The aurgument is -implicit_add [field name]. Say you vocab is called cars, then the line would be: "iwmtresolveUIDs -implicit_add cars" [This is documented on page 125 of the MT 4.1.2 Reference Manual - for those needing the citation.]

1. I would call this as my first executable in the post processor.

Using your example the resulting metadata record would look like this:

cars

MTS
JMVGQO8UU
fiesta
3.000000
MTS::JMVGQO8UU

MTS
JMVGPRC4S
fairlane
2.000000
MTS::JMVGPRC4S

MTS
JMVGPJESH
mustang
1.000000
MTS::JMVGPJESH

MTS
JMVGOZ4VY
ford
MTS::JMVGOZ4VY

MTS
JMVGORE8Z
car
MTS::JMVGORE8Z

2. I would then write a perl script (or Java or whatever you use) to do my tally logic and decide to keep or add the labels in the descriptors of the facet I need.

jfincke

Thanks for the quick response Morgen!

I think this is where the simplified example starts to break down. We have over 100 of the second level nodes ("Ford" equivalents). Without parent information in the metadata records we don't know that "Mustang" should bubble up to "Ford" and "Caprice" should bubble up to "Chevy".

With some clever manipulation of the taxonomy I could add a second level node field to each term in the vocabulary. Then I wouldn't even need to use "implicit_add" but I haven't been able to get related terms or custom terms to show up in the metadata using the -all flag. I started another thread about that issue.