def transform(self, X, y=None):
X = sameas_linker(X,
column=self.column,
new_attribute_name=self.new_attribute_name,
progress=self.progress,
endpoint=self.endpoint,
result_filter=self.result_filter,
uri_data_model=self.uri_data_model,
bundled_mode=self.bundled_mode,
prefix_lookup=self.prefix_lookup,
caching=self.caching)
return X
def test1_wikidata_bundled(self):
df = pd.DataFrame({
"word": ["they", "they", "she", "she", "he"],
"uri": ["http://www.wikidata.org/entity/L1372", "http://www.wikidata.org/entity/L493", "http://www.wikidata.org/entity/L1370", "http://www.wikidata.org/entity/L496", "http://www.wikidata.org/entity/L1371"]
})
expected_result_df = pd.DataFrame({
"word": ["they", "they", "she", "she", "he"],
"uri": ["http://www.wikidata.org/entity/L1372", "http://www.wikidata.org/entity/L493", "http://www.wikidata.org/entity/L1370", "http://www.wikidata.org/entity/L496", "http://www.wikidata.org/entity/L1371"],
"same_as_1": ["http://www.wikidata.org/entity/L371", "http://www.wikidata.org/entity/L371", "http://www.wikidata.org/entity/L484", "http://www.wikidata.org/entity/L484", "http://www.wikidata.org/entity/L485"]
})
result = sameas_linker(df, "uri", new_attribute_name="same_as", endpoint=WikiData, bundled_mode=True)
pd.testing.assert_frame_equal(result, expected_result_df)
def test4_dbpedia_unbundled_filter(self):
df = pd.DataFrame({
"label": ["Universität Mannheim", "Universität Bremen"],
"uri": ["http://dbpedia.org/resource/University_of_Mannheim", "http://dbpedia.org/resource/University_of_Bremen"]
})
expected_result_df = pd.DataFrame({
"label": ["Universität Mannheim", "Universität Bremen"],
"uri": ["http://dbpedia.org/resource/University_of_Mannheim", "http://dbpedia.org/resource/University_of_Bremen"],
"same_as_1": ["http://yago-knowledge.org/resource/University_of_Mannheim", "http://yago-knowledge.org/resource/University_of_Bremen"],
"same_as_2": ["http://rdf.freebase.com/ns/m.0b6dry", "http://rdf.freebase.com/ns/m.04fd75"],
"same_as_3": ["http://www.wikidata.org/entity/Q317070", "http://www.wikidata.org/entity/Q500692"],
"same_as_4": ["http://wikidata.dbpedia.org/resource/Q317070", "http://wikidata.dbpedia.org/resource/Q500692"]
})
result = sameas_linker(df, "uri", new_attribute_name="same_as", endpoint=DBpedia, result_filter=[
"yago", "freebase", "wiki"], bundled_mode=False)
pd.testing.assert_frame_equal(result, expected_result_df)