xquery/nieuwe_namen.xq

.gitignore

@@ -32,3 +32,4 @@ bin/week2files
 20??
 corpus
 data
+tmp

xquery/howto

@@ -0,0 +1,12 @@
+alto *.data.dz tq:nieuwe_namen.xq | sort | uniq > items.txt
+
+voor elk item dit bijwerken:
+
+    type Item struct {
+        count int
+        tags  map[string]int
+    }
+
+    items := make(map[string]Item)
+
+

xquery/nieuwe_namen.xq

@@ -0,0 +1,20 @@
+for $x in //node[((@cat="mwu" and node[@pt="spec"]) or (@pt and @*="eigen" and not(@rel="mwp"))) and not(@his="normal") and not(@his_1="decap" or @his_1="0")]
+  return ( <i>
+<m> {replace(data(/alpino_ds/sentence/@sentid), "\.[^.]*$", "")} </m>
+{
+for $i in data(/alpino_ds/metadata/meta[@name="cat"]/@value)
+  return <c> {$i} </c>
+}
+{
+for $i in data(/alpino_ds/metadata/meta[@name="tag"]/@value)
+  return <t> {$i} </t>
+}
+<w> { data($x//@word) } </w>
+</i>, '&#xa;' )
+
+(:
+
+<m>{ data(/alpino_ds/sentence/@sentid) }</m>
+<m>{ replace(data(/alpino_ds/sentence/@sentid), "\.[^.]*$", "") }</m>
+
+:)