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; } } }

MT recognizer for proper names ?

nipper

Has anyone tried to recognize a proper name with MT ? I see in the manual an example of Mr. Smith, but this is pretty informal, we call him Joe. So how would I extract Joe Smith ? Two capitals ? Write a 1st name vocab ?

Tips/Pointers/RTFMs appreciated.

Andy

Find more posts tagged with

Comments

Migrateduser

Andy,

MT 4.1.1 includes a pretty decent example name.person extractor model. The extraction rules run about 20KB ....

\examples\fields\person_names_en

Cheers,
C

yuvpurush

Andy,

Even i tried the recognizer for propernames and was not successfule. Im posting what steps i followed. Can you tell me that you've also did it in the same way or differrent.

** I have created a recognizer called "Names" and created a taxonomy by importing the vocabulary from the file "Name.Person.xml" which is present in that example.

** Under Build-> Settings -> Recognizer tab -> Advanced, i copied the "trainRecognizer" script from the file "Name.Person_trainrecognizer.cfg" which is present i that example.

**In TestIndex -> Advanced, copied the "Parsing rule" from the file "Name.Person.script" which is present i that example.

** Then i build the index. But when i tested the index with some sample keywords, which is verymuch present in the vocabulary (eg Daniel), it is not coming. If i didn't change the script, it is working well. As the example given one script, im using the given script and not the default one.

** Then i committed the index and added this taxonomy to the content processor for type "text"

**I then copied the "Name.Person.mtx.xml" on "mthome/conf/mtstudio_home/Names/db_home" folder

** Created a script called combo.script with the value "parse -richTokens; recognize; resolve -report_all -only_annotate;extract -config Name.Person.mtx.xml"

** Then i ran the iwmtdebug with "iwmtdebug -db Names -script combo.script -dump_xml -lang english sample.txt"

I got the output as

****************************************************************
<metadata>

Timing Summary:

Initialization time: 0
Execution time : 1
Total time : 1
Total size : 21264
Rate : 20.7656K/sec
Rack Metrics:
parse : 1 seconds
recognize : 0 seconds
resolve : 0 seconds
extract : 0 seconds
****************************************************************

Using CI viewer also im not getting any output.

the sample.txt contains some person names.

Rgds
yuv

Migrateduser

Try using the following commands or attached batch file to build and deploy the model.

---------------------------------------------------BEGIN install.bat
rem --- delete the previous version of the model
rmdir %DB_HOME%\mtstudio_home\name.person.en /S /Q

rem --- build the model lexicon
call iwmtbuild.bat -new name.person.en -username %1 -password %2 Name.Person.xml -case -lang english -config Name.Person_trainrecognizer.cfg -script_tag Name.Person.script

rem --- deploy the model lexicon
call iwmtdeploytaxonomy.bat -username %1 -password %2 name.person.en

rem --- deploy the extractor script
copy Name.Person.mtx.xml %DB_HOME%
---------------------------------------------------END install.bat
You will need to use ciadmin to add the model to a text content processor. Alternatively, you can replace your metatagger.cfg with the attached metatagger.cfg which has an example of the configuration.

You can run a test using the CLT:
iwgenmetadata -suggest -save test.txt

yuvpurush

Hi,Thanks for the reply.

After i ran the batch file, it created all the corresponding folder and files under dbhome/mtstudio_home

But when i tested with the sample provided, it is not extracting. Im not sure were im worng. Attached the console output

***********************************************************
F:\installs\Interwoven\MetaTagger\examples\fields\person_names_en>install admin admin

F:\installs\Interwoven\MetaTagger\examples\fields\person_names_en>rem --- delete the previous version of the model

F:\installs\Interwoven\MetaTagger\examples\fields\person_names_en>rem --- build the modellexicon

F:\installs\Interwoven\MetaTagger\examples\fields\person_names_en>call iwmtbuild.bat -new name.person.en -username admin -password admin Name.Person.xml -case -lang english -config Name.Person_trainrecognizer.cfg -script_tag Name.Person.script

MTAdminClient.getRunningHosts id=19224790
1 file(s) copied.
******************************************************
I then made changes in the metatagger.cfg and tested like the same

F:\installs\Interwoven\MetaTagger\examples\files>iwgenmetadata -suggest -save test.txt
Processing: test.txt

The output is

*********************

<metadata>

english
cp_1252
test.txt

*************************

Migrateduser

Did you restart the MT server after updating metatagger.cfg?

ExamplePie(visible).rptdesign

yuvpurush

Hi,

Sorry.. i forgot to restart.. now it is working fine.. thanks a lot for your help.

Rgds
Yuv