Reputation: 107
I am trying to use the StanfordNamedEntityRecognizer for German. I loaded a model from dkpro. Unfortunately, the model is not recognized. Instead the NER tries to use a different model, that is not provided anywhere.
How can I tell the AE which model to use? For example instead of "model-ner-de-nemgp" use "model-ner-de-Person".
This is my main ruta file:
PACKAGE org.apache.uima.ruta.novel;
IMPORT PACKAGE de.tudarmstadt.ukp.dkpro.core.api.lexmorph.type.pos FROM desc.type.POS AS pos;
IMPORT PACKAGE de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Sentence FROM desc.type.LexicalUnits AS sentence;
IMPORT PACKAGE de.tudarmstadt.ukp.dkpro.core.api.ner.type.NamedEntity FROM desc.type.NamedEntity;
UIMAFIT de.tudarmstadt.ukp.dkpro.core.opennlp.OpenNlpSegmenter;
UIMAFIT de.tudarmstadt.ukp.dkpro.core.stanfordnlp.StanfordPosTagger;
UIMAFIT de.tudarmstadt.ukp.dkpro.core.stanfordnlp.StanfordParser;
UIMAFIT de.tudarmstadt.ukp.dkpro.core.stanfordnlp.StanfordNamedEntityRecognizer;
uima.tcas.DocumentAnnotation{-CONTAINS(pos.POS)} -> {
uima.tcas.DocumentAnnotation{-> SETFEATURE("language", "de")};
EXEC(OpenNlpSegmenter);
EXEC(StanfordPosTagger, {pos.POS});
EXEC(StanfordNamedEntityRecognizer);
};
Upvotes: 1
Views: 66
Reputation: 107
Inspecting the corresponding descriptor.xml I found the configurationParameter "modelVariant". I added the following lines to my Ruta script which solved my request.
This sets the modelVariant for the StanfordNamedEntityRecognizer to my prefered model named "germeval2014.hgc_175m_600.crf".
Document{-> CONFIGURE(StanfordNamedEntityRecognizer, "modelVariant" = "germeval2014.hgc_175m_600.crf")};
Upvotes: 1