Fix duplicate notation 'n10'

dini-ag-kim · Jul 9, 2024 · e11fc5a · e11fc5a
1 parent 9bbee5c
commit e11fc5a
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 17 deletions.
diff --git a/create_faechersystematik_ttl.py b/create_faechersystematik_ttl.py
@@ -10,6 +10,11 @@
 df_2nd_level = pd.read_csv(url_2nd_level, encoding="ISO-8859-1", sep=';', quotechar='"', header=None, engine ='python', dtype=str, usecols=[0, 2, 3], names=["notation", "label", "broader"])
 df_3rd_level = pd.read_csv(url_3rd_level, encoding="ISO-8859-1", sep=';', quotechar='"', header=None, engine ='python', dtype=str, usecols=[0, 2, 3], names=["notation", "label", "broader"])
 
+df_1st_level['notation'] = df_1st_level['notation'].str.lstrip("0")
+df_2nd_level['broader'] = df_2nd_level['broader'].str.lstrip("0")
+df_1st_level['notation'] = df_1st_level['notation'].apply(lambda n: "00" if n == "10" else n)
+df_2nd_level['broader'] = df_2nd_level['broader'].apply(lambda n: "00" if n == "10" else n)
+
 dict_1st_level = df_1st_level.to_dict("records")
 dict_2nd_level = df_2nd_level.to_dict("records")
 dict_3rd_level = df_3rd_level.to_dict("records")
@@ -37,18 +42,18 @@
 
 
 for idx, i in enumerate(dict_1st_level):
-    top_level = dict_1st_level[idx]['notation'].lstrip("0")
+    top_level = dict_1st_level[idx]['notation']
     g.add((URIRef('n%s' % top_level), RDF['type'], skos['Concept']))
     g.add((URIRef('n%s' % top_level), skos['topConceptOf'], (URIRef('scheme'))))
     g.add((URIRef('n%s' % top_level), skos['prefLabel'], Literal(dict_1st_level[idx]['label'], lang='de')))
     g.add((URIRef('n%s' % top_level), skos['notation'], Literal(top_level)))
     g.add((URIRef('scheme'), skos['hasTopConcept'], (URIRef('n%s' % top_level))))
     for idx_2, i_2 in enumerate(dict_2nd_level):
-        if dict_2nd_level[idx_2]['broader'].lstrip("0") == top_level:
+        if dict_2nd_level[idx_2]['broader'] == top_level:
             level_2_notation = dict_2nd_level[idx_2]['notation']
             g.add((URIRef('n%s' % level_2_notation), RDF['type'], skos['Concept']))
             g.add((URIRef('n%s' % level_2_notation), skos['prefLabel'], Literal(dict_2nd_level[idx_2]['label'], lang='de')))
-            g.add((URIRef('n%s' % level_2_notation), skos['broader'], (URIRef('n%s' % dict_2nd_level[idx_2]['broader'].lstrip('0')))))
+            g.add((URIRef('n%s' % level_2_notation), skos['broader'], (URIRef('n%s' % dict_2nd_level[idx_2]['broader']))))
             g.add((URIRef('n%s' % level_2_notation), skos['notation'], Literal(level_2_notation)))
             g.add((URIRef('n%s' % level_2_notation), skos['inScheme'], (URIRef('scheme'))))
             for idx_3, i_3 in enumerate(dict_3rd_level):

diff --git a/hochschulfaechersystematik.ttl b/hochschulfaechersystematik.ttl
@@ -1734,6 +1734,11 @@
     skos:notation "70" ;
     skos:prefLabel "Wirtschaftsingenieurwesen mit ingenieurwissenschaftlichem Schwerpunkt"@de .
 
+<n00> a skos:Concept ;
+    skos:notation "00" ;
+    skos:prefLabel "Außerhalb der Studienbereichsgliederung/Sonstige Fächer"@de ;
+    skos:topConceptOf <scheme> .
+
 <n01> a skos:Concept ;
     skos:broader <n1> ;
     skos:inScheme <scheme> ;
@@ -1746,6 +1751,12 @@
     skos:notation "06" ;
     skos:prefLabel "Informations- und Bibliothekswissenschaften"@de .
 
+<n10> a skos:Concept ;
+    skos:broader <n1> ;
+    skos:inScheme <scheme> ;
+    skos:notation "10" ;
+    skos:prefLabel "Anglistik, Amerikanistik"@de .
+
 <n2> a skos:Concept ;
     skos:notation "2" ;
     skos:prefLabel "Sport"@de ;
@@ -1884,23 +1895,11 @@
     skos:prefLabel "Ernährungs- und Haushaltswissenschaften"@de .
 
 <n83> a skos:Concept ;
-    skos:broader <n10> ;
+    skos:broader <n00> ;
     skos:inScheme <scheme> ;
     skos:notation "83" ;
     skos:prefLabel "Außerhalb der Studienbereichsgliederung"@de .
 
-<n10> a skos:Concept ;
-    skos:broader <n1> ;
-    skos:inScheme <scheme> ;
-    skos:notation "10" ;
-    skos:prefLabel "Anglistik, Amerikanistik"@de .
-
-<n00> a skos:Concept ;
-    skos:notation "00" ;
-    skos:prefLabel "Außerhalb der Studienbereichsgliederung/Sonstige Fächer"@de ;
-    skos:topConceptOf <scheme> .
-
-
 <n36> a skos:Concept ;
     skos:broader <n4> ;
     skos:inScheme <scheme> ;
@@ -2107,8 +2106,8 @@
     vann:preferredNamespacePrefix "hfs" ;
     vann:preferredNamespaceUri "https://w3id.org/kim/hochschulfaechersystematik/" ;
     skos:hasTopConcept <n0>,
-        <n1>,
         <n00>,
+        <n1>,
         <n15>,
         <n2>,
         <n20>,