Add topic domain classification in XSLT pass2
Generated with mallet based on the old training data in /vol/work/TE via
calling a Java function from XSLT.
Resolves #6
diff --git a/Makefile b/Makefile
index 9b38bed..674c69a 100644
--- a/Makefile
+++ b/Makefile
@@ -17,7 +17,7 @@
MAKE ?= make -j $(shell nproc)
KORAPXML2CONLLU_HEAP ?= $(shell echo "$$(($(MAX_THREADS) * 2100))")
KORAPXML2CONLLU ?= java -Xmx$(KORAPXML2CONLLU_HEAP)m -jar lib/korapxml2conllu.jar
-SAXON ?= java -cp lib/saxon9ee.jar:lib/xml-resolver-1.2.jar net.sf.saxon.Transform -expand:off -catalog:"lib/dtds/xhtml11/xhtmlcatalog.xml;lib/dtds/xhtml/dtd/xhtmlcatalog.xml"
+SAXON ?= java -cp lib/saxon9ee.jar:lib/xml-resolver-1.2.jar:lib/textclassifier.jar net.sf.saxon.Transform -expand:off -catalog:"lib/dtds/xhtml11/xhtmlcatalog.xml;lib/dtds/xhtml/dtd/xhtmlcatalog.xml"
.DELETE_ON_ERROR:
@@ -92,6 +92,10 @@
mkdir -p models
curl -sL -o $@ https://corpora.ids-mannheim.de/tools/$@
+models/dereko_domains_s.classifier:
+ mkdir -p models
+ curl -sL -o $@ https://corpora.ids-mannheim.de/tools/$@
+
%.marmot-malt.zip: %.zip models/de.marmot models/german.mco
$(KORAPXML2CONLLU) -T $(MAX_THREADS) -t marmot:models/de.marmot -P malt:models/german.mco $< | conllu2korapxml > $@