nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

iwmtbatch...

Ottawa_IWOV

How does iwmtbatch work? I have read all the documentation, and have got it to run, but I can seem to get any results. How does it generate the metadata, will it assign multiple terms? I assume it looks for similarities with controlled vocabularies?

Any insight here would be great, I am kind of in a pinch

Cheers!

Lucas Cochrane
lcochrane@dc.com

Find more posts tagged with

Comments

Migrateduser

Hi Lucas -

iwmtbatch is the tool used to batch process documents based on the metatagger.cfg and metadata-rules.cfg in yout iw-home/local/config directory. Basically you give it a config file which tells it what files to tag, and then you let it go at it.

Here is what I do to run iwmtbatch from a TS server using a remote MT server. This is part of a script which is passed the path to the files it will tag.<p>

my $batch = "c:/batch.cfg";
open (BAT, ">$batch");
print BAT <<"EOS";
< batch>
<extension>htm</extension>
<extension>html</extension>
< path recurse=\"true\">$dir</path>
<mthost>mthost:9095</mthost>
<contextURL>http://localhost:81/iw-bin/iwmtcontext.cgi</contextURL>
< /batch>
EOS
close BAT;
`D:\\MetaTagger\\bin\\iwmtbatch.exe -batch $batch -suggest -save `;
unlink $batch;

Make sure you are getting output from the gui first. If that is working iwmtbatch will.

Migrateduser

Hi Lucas,

Here is a response from a member of MT team (thanks to Mark B.). I hope it helps.

<Resp_iwmtbatch>

iwmtbatch is a MetaTagger client that is tightly coupled with TeamSite. In its configuration file, you specify a particular TeamSite directory that it should crawl and tag using the MetaTagger. All tags are then written into the TeamSite Extended Attributes for the tagged file. In order for iwmtbatch to work, you must have a valid metadata-rules.cfg and datacapture.cfg file, as well as a metatagger.cfg file. It is the metatagger.cfg file, in conjunction with the metadata-rules.cfg and datacapture.cfg files that determine the behavior of the MetaTagger.

If you are using the MetaTagger separately from TeamSite, then you want to use an alternative program called iwgenmetadata. This program crawls a file system and tags all of the resources you specify. It then creates a file with an .imd extension that contains the metadata suggested for the file. If you are using iwgenmetadata, you only need to have the metatagger.cfg file.

The MetaTagger can assign one or more terms depending on the particular rack process you invoke and the parameters that you give to it. For instance, you can use the classify rack process and use a parameter of –numcategories 1 to have only one category assigned. The recognition process is less likely to assign one category, but it could happen, depending on the controlled vocabulary that you are using. The recognizer rack process looks for similarities with the controlled vocabulary in order to suggest terms from the controlled vocabulary. The classify rack process uses a training set to suggest categories. It does not take advantage of similarity to terms in a controlled vocabulary.

If you are not getting any terms assigned, you should try running iwgenmetadata on a test document, since that only involves configuring metatagger.cfg. Once you can successfully generate tags then you can try replicating the process with iwmtbatch by configuring your datacapture.cfg and you metadata-rules.cfg as well.
</Res_iwmtbatch>

Hope this addresses your need. If not, please provide additional detail about what you want to do, the environment, etc. and possibly send the config scripts, etc. (also some detail about the vocab.)

Thanks

Matt

Ottawa_IWOV

I cannot generate tags with iwgenmetadata at all. The problem is, I don't have access to MetaTagger Studio, because all work is being done on the server and MT 3.5 doesn't let you install the Studio on the server. Is there any other way to debug, or possibly edit vocabularies? I need something to work with here.

Thanks.

Lucas.

Lucas Cochrane
lcochrane@dc.com

NathanW

Hi Lucas

A couple of thoughts:

1) You may not get metadata suggested on your content if MT hasnt been trained particularly well - in order to fix this you will need to add additional documents to your training set and 'teach' metatagger to use the correct values - this can be a long a tedious process but it is worth it - we now have MT creating data which ranges between 80-90% accurate for 'most' of our documents. Now that we have been using it for a while and manually correcting the obvious mistakes we are going to feed a larger training set back through the process to increase the accuracy even more.

2) If you are running MT3.5 and you are doing all the work directly on the server - open a command session and type MTSTUDIO - this launches a bat file that will open studio directly off the server - you dont need to install it - its installed with the server component - the download/install is for remote use - both work really well for us.

Cheers

Nathan Wall
Department of Transport and Regional Services (Australia)

Migrateduser

You CAN use MetaTagger Studio on the server! You can't use the Studio installer on the server because the MetaTagger installer already puts it there for you.

From the documentation:

"When your administrator installs MetaTagger, the installation program installs a copy of MetaTagger Studio on the MetaTagger server host. To use Studio on the MetaTagger server host, you can launch it from the command line, by typing mtstudio and pressing Enter. "

Ottawa_IWOV

All,
Quick question. We have "MetaTagger enabled" some fields because we want the values added to the XML stream in EA's. That works fine. But all the content that exists prior to adding the functionality has the value only in TeamSite/Metadata/*.*

Is there a way to run iwmtbatch to just get that value into the XML stream? If I do a suggest_by_tag it doesn't like it an keeps saying that my tag doesn't exist. But when I simply run suggest it doesn't suggest any metadata because we are using an extractor, classifier or recognizer. Our users are simply browsing vocabulary indexes in GUI mode and manually adding subject metadata.

Any thoughts here would be GREATLY appreciated.

Thanks,

Lucas.

Lucas Cochrane
lcochrane@dc.com