def transform(self, X, y=None):
X = label_linker(X,
column=self.column,
new_attribute_name=self.new_attribute_name,
progress=self.progress,
endpoint=self.endpoint,
result_filter=self.result_filter,
language=self.language,
max_hits=self.max_hits,
label_property=self.label_property,
prefix_lookup=self.prefix_lookup,
caching=self.caching)
return X
def test3_dbpedia_de_gu(self):
df = pd.DataFrame({
"state": ["Baden-Württemberg", "Bayern", "Sachsen"]
})
expected_result_df = pd.DataFrame({
"state": ["Baden-Württemberg", "Bayern", "Sachsen"],
"new_link_1": ["http://dbpedia.org/resource/Baden-Württemberg", "http://dbpedia.org/resource/Bavaria", "http://dbpedia.org/resource/Saxony"]
})
result = label_linker(df, "state", endpoint=DBpedia,
language="de", graph_uri="http://dbpedia.org")
pd.testing.assert_frame_equal(result, expected_result_df)
def test2_dbpedia_en(self):
df = pd.DataFrame({
"state": ["Baden-Württemberg", "Bayern", "Sachsen"]
})
expected_result_df = pd.DataFrame({
"state": ["Baden-Württemberg", "Bayern", "Sachsen"],
"new_link_1": ["http://dbpedia.org/resource/Baden-Württemberg", "http://dbpedia.org/resource/Bayern", "http://dbpedia.org/resource/Sachsen"],
"new_link_2": ["http://www.wikidata.org/entity/Q985", "http://www.wikidata.org/entity/Q255654", "http://www.wikidata.org/entity/Q16882470"],
"new_link_3": ["http://www.wikidata.org/entity/Q20825585", "http://www.wikidata.org/entity/Q4874432", np.nan],
"new_link_4": [np.nan, "http://www.wikidata.org/entity/Q18148056", np.nan]
})
result = label_linker(df, "state", endpoint=DBpedia, language="en")
pd.testing.assert_frame_equal(result, expected_result_df)
def test4_dbpedia_de(self):
df = pd.DataFrame({
"state": ["Baden-Württemberg", "Bayern", "Sachsen"]
})
expected_result_df = pd.DataFrame({
"state": ["Baden-Württemberg", "Bayern", "Sachsen"],
"new_link_1": ["http://dbpedia.org/resource/Baden-Württemberg", "http://www.wikidata.org/entity/Q980", "http://dbpedia.org/resource/Saxony"],
"new_link_2": ["http://www.wikidata.org/entity/Q985", "http://dbpedia.org/resource/Bavaria", "http://www.wikidata.org/entity/Q467095"],
"new_link_3": ["http://www.wikidata.org/entity/Q21036291", "http://www.wikidata.org/entity/Q255654", "http://www.wikidata.org/entity/Q1202"],
"new_link_4": [np.nan, "http://www.wikidata.org/entity/Q4874432", "http://www.wikidata.org/entity/Q101985"],
"new_link_5": [np.nan, "http://www.wikidata.org/entity/Q18148056", "http://www.wikidata.org/entity/Q3873416"],
"new_link_6": [np.nan, np.nan, "http://www.wikidata.org/entity/Q27505"],
"new_link_7": [np.nan, np.nan, "http://www.wikidata.org/entity/Q16882470"],
"new_link_8": [np.nan, np.nan, "http://www.wikidata.org/entity/Q1325543"],
"new_link_9": [np.nan, np.nan, "http://www.wikidata.org/entity/Q1435378"],
"new_link_10": [np.nan, np.nan, "http://www.wikidata.org/entity/Q1537344"],
"new_link_11": [np.nan, np.nan, "http://www.wikidata.org/entity/Q1365112"]
})
result = label_linker(df, "state", endpoint=DBpedia, language="de")
pd.testing.assert_frame_equal(result, expected_result_df)