Add submission 3

app.py

@@ -3,7 +3,7 @@ from flask.json import jsonify
 
 from src.main import assemble_wikidata_groundtruth_english, assemble_dbpedia_groundtruth_english, \
     assemble_wikidata_triples, \
-    assemble_dbpedia_triples, test_if_data_is_enlargeable_for_dbpedia
+    assemble_dbpedia_triples, test_if_data_is_enlargeable_for_dbpedia, assemble_similarity_triples
 
 app = Flask(__name__)
 
@@ -12,6 +12,11 @@ app = Flask(__name__)
 # Todo: Füge eine Route ein die die zusätlichen Werte in die Triple-Dateien einbringt.
 # Todo: Es soll vorher immer erst der Grunddatensatz erzeugt werden und dann die zusätzlichen Informationen ergänzt werden.
 
+@app.route('/coreferences')
+def coreferences():
+    return jsonify(assemble_similarity_triples())
+
+
 @app.route('/dbpedia/collect/enlargeable')
 def dbpedia_data_can_enlarge():
     return jsonify(test_if_data_is_enlargeable_for_dbpedia())

src/lib/keys.py

@@ -3,8 +3,10 @@ from enum import Enum
 
 class ResultKeys(Enum):
     movie = 'Movie'
+    movie_link = 'MovieLink'
     title = 'Title'
     director = 'Director'
+    director_link = 'DirectorLink'
     author = 'Author'
     cast = 'Cast'
     published = 'Published'

src/main.py

@@ -256,3 +256,47 @@ def assemble_dbpedia_triples() -> List:
                                           value=production_companies).as_ete_triple()]
     FileWriter(destination='static/dbpedia/dbpedia.txt', data=triples).as_string(new_line=True)
     return triples
+
+
+def assemble_similarity_triples() -> List:
+    """
+    This method assembles the similarity between the wikidata and the dbpedia datase.
+    This will be done in two instances.
+
+    First level:
+        -> Take a look the owl:sameAs
+    Second level:
+        -> Take a look if two instances have the same name and the same director/directors
+
+    :return: None
+    """
+    dbpedia_data = FileReader(source='static/dbpedia/dbpedia_groundtruth.txt').as_json()
+    wikidata_data = FileReader(source='static/wikidata/wikidata_groundtruth.txt').as_json()
+
+    triples = []
+    for wiki in wikidata_data:
+        wiki_movie = Decapper(wiki[ResultKeys.movie.value]).unpack()
+        for dbpedia in dbpedia_data:
+            if Decapper(wiki[ResultKeys.movie.value]).unpack() == \
+                    Decapper(dbpedia[ResultKeys.movie_link.value]).unpack():
+                triples += [Statement(subject=wiki_movie,
+                                      predicate='http://www.w3.org/2002/07/owl#sameAs',
+                                      value=Decapper(dbpedia[ResultKeys.movie.value]).unpack()
+                                      ).as_ete_triple()
+                            ]
+                break
+
+            if Decapper(wiki[ResultKeys.title.value]).unpack() == Decapper(dbpedia[ResultKeys.title.value]).unpack():
+                wiki_set = set(
+                    Decapper(wiki[ResultKeys.director.value]).unpack().split(ResultKeys.line_separator.value))
+                dbpedia_set = set(
+                    Decapper(dbpedia[ResultKeys.director_link.value]).unpack().split(ResultKeys.line_separator.value))
+                if len(wiki_set & dbpedia_set) > 0:
+                    triples += [Statement(subject=wiki_movie,
+                                          predicate='http://www.w3.org/2002/07/owl#sameAs',
+                                          value=Decapper(dbpedia[ResultKeys.movie.value]).unpack()
+                                          ).as_ete_triple()
+                                ]
+                    break
+    FileWriter(destination='static/coreferences/coreferences.txt', data=triples).as_string(new_line=True)
+    return triples

static/dbpedia/dbpedia_groundtruth.sparql

@@ -15,10 +15,13 @@ SELECT DISTINCT (?movie AS ?Movie)
 (GROUP_CONCAT(DISTINCT ?releaseDate;separator="|") AS ?Published)
 (GROUP_CONCAT(DISTINCT ?subject;separator="|") AS ?Subject)
 (GROUP_CONCAT(DISTINCT ?genre;separator="|") AS ?Genre)
-(GROUP_CONCAT(DISTINCT ?duration/60;separator="|") AS ?Duration)
+(GROUP_CONCAT(DISTINCT ?duration/60;separator="|") AS ?Duration) # todo: change to original value
 (GROUP_CONCAT(DISTINCT ?abstract;separator="|") AS ?Description)
 (GROUP_CONCAT(DISTINCT ?distributor;separator="|") AS ?Distributor)
 (GROUP_CONCAT(DISTINCT ?productionCompanies;separator="|") AS ?ProductionCompanies)
+(GROUP_CONCAT(DISTINCT ?sameDirector;separator="|") AS ?DirectorLink)
+(GROUP_CONCAT(DISTINCT ?sameMovie;separator="|") AS ?MovieLink)
+
 WHERE {
    {
     SELECT DISTINCT ?movie
@@ -50,7 +53,13 @@ WHERE {
     }GROUP BY ?movie HAVING (?min_year >= 1970)
    }
    OPTIONAL{?movie foaf:name ?name FILTER(LANG(?name)="en").}
-   OPTIONAL{?movie dbo:director ?director.}
+   OPTIONAL{?movie owl:sameAs ?sameMovie FILTER CONTAINS (LCASE(STR(?sameMovie)), 'wikidata.org').}
+   OPTIONAL{
+       ?movie dbo:director ?director.
+       OPTIONAL{
+          ?director owl:sameAs ?sameDirector FILTER CONTAINS (LCASE(STR(?sameDirector)), 'wikidata.org').
+       }
+   }
    OPTIONAL{?movie dbo:author ?author.}
    OPTIONAL{?movie dbo:starring ?cast_member.}
    OPTIONAL{?movie dbo:releaseDate ?releaseDate.}