def process_spotlight_api(text):
try:
entities = spotlight.annotate(
"http://spotlight.dbpedia.org/rest/annotate",
text,
confidence=0.1,
support=0
)
except:
return []
link_matches = HyperLink.extract_all_url(text)
initial_entities = []
for entity in entities:
occ = 0
if occ is not 0:
occ = text.count('"', 0, entity["offset"] + len(entity["serviceForm"]) - 1)
start = entity["offset"] + occ
end = entity["offset"] + len(entity["surfaceForm"]) + occ
possible_link = False
for link_match in link_matches:
if link_match["start"] <= start and link_match["end"] >= end:
possible_link = True
if not possible_link:
e = {
"label": entity["surfaceForm"],
"startOffset": start,
"endOffset": end,
"confidence": entity["similarityScore"],
"provenance": "dbpediaspotlight",
"types": []
}
types = []
for data_type in entity["types"].split(","):
link = data_type
if "DBpedia:" in data_type:
link = "http://en.dbpedia.org/resource/" + data_type.split(":")[1]
if "Freebase:" in data_type:
link = "http://www.freebase.com" + data_type.split(":")[1]
dbpedia_type = {
"typeURI": None,
"typeLabel": data_type,
"entityURI": link,
"confidence": entity["similarityScore"],
"wikiURI": DbpediaLink.get_english_wikipedia_link_from_english_resource(link)
}
types.append(dbpedia_type)
e["types"].append(types)
initial_entities.append(e)
return initial_entities
def process_textrazor_api(text):
client = TextRazor(
api_key='67ef1ca06614f7d202b23f1444bd7ee1ea2f916b3ecf488f8d39f800',
extractors=[
"entities",
"topics",
"words",
"phrases",
"dependency-trees",
"senses"
]
)
try:
response = client.analyze(text)
except:
return []
link_matches = HyperLink.extract_all_url(text)
initial_entities = []
for entity in response.entities():
possible_link = False
for link_match in link_matches:
if link_match["start"] <= entity.starting_position and link_match["end"] >= entity.ending_position:
possible_link = True
if not possible_link:
e = {
"label": entity.matched_text,
"startOffset": entity.starting_position,
"endOffset": entity.ending_position,
"confidence": entity.confidence_score,
"relevance": entity.relevance_score,
"provenance": "textrazor",
"wikipediaLink": entity.wikipedia_link,
"types": []
}
for dbpedia_type in entity.dbpedia_types:
wiki_link = "http://en.wikipedia.org/wiki/" + dbpedia_type
dbpedia_type_list = {
"typeURI": None,
"typeLabel": dbpedia_type,
"wikiURI": wiki_link,
"entityURI": DbpediaLink.get_english_resource_from_english_wikipedia_link(wiki_link),
"confidence": entity.confidence_score
}
e["types"].append(dbpedia_type_list)
for freebase_type in entity.freebase_types:
freebase_link = "http://www.freebase.com" + freebase_type
freebase_type_list = {
"typeURI": None,
"typeLabel": "Freebase:" + freebase_type.replace(" ", ""),
"wikiURI": None,
"entityURI": freebase_link,
"confidence": entity.confidence_score
}
e["types"].append(freebase_type_list)
wiki_type_list = {
"typeURI": None,
"typeLabel": [],
"wikiURI": entity.wikipedia_link,
"entityURI": DbpediaLink.get_english_resource_from_english_wikipedia_link(entity.wikipedia_link),
"confidence": entity.confidence_score
}
e["types"].append(wiki_type_list)
initial_entities.append(e)
return initial_entities
def process_textrazor_api(text):
client = TextRazor(
api_key="67ef1ca06614f7d202b23f1444bd7ee1ea2f916b3ecf488f8d39f800",
extractors=["entities", "topics", "words", "phrases", "dependency-trees", "senses"],
)
try:
response = client.analyze(text)
except:
return []
link_matches = HyperLink.extract_all_url(text)
initial_entities = []
for entity in response.entities():
possible_link = False
for link_match in link_matches:
if link_match["start"] <= entity.starting_position and link_match["end"] >= entity.ending_position:
possible_link = True
if not possible_link:
e = {
"label": entity.matched_text,
"startOffset": entity.starting_position,
"endOffset": entity.ending_position,
"confidence": entity.confidence_score,
"relevance": entity.relevance_score,
"provenance": "textrazor",
"wikipediaLink": entity.wikipedia_link,
"types": [],
}
for dbpedia_type in entity.dbpedia_types:
wiki_link = "http://en.wikipedia.org/wiki/" + dbpedia_type
dbpedia_type_list = {
"typeURI": None,
"typeLabel": dbpedia_type,
"wikiURI": wiki_link,
"entityURI": DbpediaLink.get_english_resource_from_english_wikipedia_link(wiki_link),
"confidence": entity.confidence_score,
}
e["types"].append(dbpedia_type_list)
for freebase_type in entity.freebase_types:
freebase_link = "http://www.freebase.com" + freebase_type
freebase_type_list = {
"typeURI": None,
"typeLabel": "Freebase:" + freebase_type.replace(" ", ""),
"wikiURI": None,
"entityURI": freebase_link,
"confidence": entity.confidence_score,
}
e["types"].append(freebase_type_list)
wiki_type_list = {
"typeURI": None,
"typeLabel": [],
"wikiURI": entity.wikipedia_link,
"entityURI": DbpediaLink.get_english_resource_from_english_wikipedia_link(entity.wikipedia_link),
"confidence": entity.confidence_score,
}
e["types"].append(wiki_type_list)
initial_entities.append(e)
return initial_entities
def process_spotlight_api(text):
try:
entities = spotlight.annotate(
"http://spotlight.dbpedia.org/rest/annotate",
text,
confidence=0.1,
support=0)
except:
return []
link_matches = HyperLink.extract_all_url(text)
initial_entities = []
for entity in entities:
occ = 0
if occ is not 0:
occ = text.count('"', 0,
entity["offset"] + len(entity["serviceForm"]) - 1)
start = entity["offset"] + occ
end = entity["offset"] + len(entity["surfaceForm"]) + occ
possible_link = False
for link_match in link_matches:
if link_match["start"] <= start and link_match["end"] >= end:
possible_link = True
if not possible_link:
e = {
"label": entity["surfaceForm"],
"startOffset": start,
"endOffset": end,
"confidence": entity["similarityScore"],
"provenance": "dbpediaspotlight",
"types": []
}
types = []
for data_type in entity["types"].split(","):
link = data_type
if "DBpedia:" in data_type:
link = "http://en.dbpedia.org/resource/" + data_type.split(
":")[1]
if "Freebase:" in data_type:
link = "http://www.freebase.com" + data_type.split(":")[1]
dbpedia_type = {
"typeURI":
None,
"typeLabel":
data_type,
"entityURI":
link,
"confidence":
entity["similarityScore"],
"wikiURI":
DbpediaLink.
get_english_wikipedia_link_from_english_resource(link)
}
types.append(dbpedia_type)
e["types"].append(types)
initial_entities.append(e)
return initial_entities
def process_nerd_api(text):
try:
timeout = 10
text = urllib.quote_plus(text)
n = nerd.NERD("nerd.eurecom.fr", 'akkqfgos0p85mcubcfgp82rn92d23enu')
entities = n.extract(text, "combined", timeout)
except:
return []
link_matches = HyperLink.extract_all_url(text)
initial_entities = []
for entity in entities:
possible_link = False
for link_match in link_matches:
if link_match["start"] <= entity["startChar"] and link_match[
"end"] >= entity["endChar"]:
possible_link = True
if not possible_link:
e = {
"label": entity["label"],
"startOffset": entity["startChar"],
"endOffset": entity["endChar"],
"confidence": entity["confidence"],
"provenance": "nerd-" + entity["extractor"],
"types": []
}
if entity["extractorType"]:
all_types = entity["extractorType"].split(",")
for extracted_type in all_types:
if "dbpedia" in extracted_type:
type_data = {
"typeURI":
extracted_type,
"typeLabel":
None,
"wikiURI":
DbpediaLink.
get_english_resource_from_english_wikipedia_link(
entity["uri"]),
"entityURI":
entity["uri"],
"confidence":
entity["confidence"]
}
else:
type_data = {
"typeURI":
None,
"typeLabel":
extracted_type,
"wikiURI":
DbpediaLink.
get_english_resource_from_english_wikipedia_link(
entity["uri"]),
"entityURI":
entity["uri"],
"confidence":
entity["confidence"]
}
e["types"].append(type_data)
if entity["nerdType"]:
nerd_type_data = {
"typeURI":
entity["nerdType"],
"typeLabel":
entity["nerdType"].split("#")[1],
"wikiURI":
DbpediaLink.
get_english_resource_from_english_wikipedia_link(
entity["uri"]),
"entityURI":
entity["uri"],
"confidence":
entity["confidence"]
}
e["types"].append(nerd_type_data)
initial_entities.append(e)
return initial_entities
def process_tagme_api(text):
try:
lang = "en"
url = "http://tagme.di.unipi.it/tag?"
include_categories = "true"
include_all_spots = "true"
api_key = read_api_key()
link_matches = HyperLink.extract_all_url(text)
request = "key=" + api_key + \
"&include_categories=" + include_categories + \
"&lang=" + lang + \
"&include_all_spots=" + include_all_spots + \
"&text=" + urllib.quote_plus(text)
entities = requests.post(url + request)
entities = json.loads(entities.text)
except:
return []
initial_entities = []
for entity in entities["annotations"]:
possible_link = False
for link_match in link_matches:
if link_match["start"] <= entity["start"] and link_match[
"end"] >= entity["end"]:
possible_link = True
if not possible_link:
e = {
"label": entity["spot"],
"startOffset": entity["start"],
"endOffset": entity["end"],
"confidence": entity["rho"],
"provenance": "tagme",
"types": []
}
types = []
for data_type in entity["dbpedia_categories"]:
wiki_url = "http://en.wikipedia.org/wiki/" + data_type.replace(
" ", "_")
dbpedia_type = {
"typeURI":
None,
"typeLabel":
data_type,
"entityURI":
DbpediaLink.
get_english_resource_from_english_wikipedia_link(wiki_url),
"confidence":
None,
"wikiURI":
wiki_url
}
types.append(dbpedia_type)
e["types"].append(types)
initial_entities.append(e)
initial_entities = {v["label"]: v for v in initial_entities}.values()
return initial_entities
def process_nerd_api(text):
try:
timeout = 10
text = urllib.quote_plus(text)
n = nerd.NERD("nerd.eurecom.fr", "akkqfgos0p85mcubcfgp82rn92d23enu")
entities = n.extract(text, "combined", timeout)
except:
return []
link_matches = HyperLink.extract_all_url(text)
initial_entities = []
for entity in entities:
possible_link = False
for link_match in link_matches:
if link_match["start"] <= entity["startChar"] and link_match["end"] >= entity["endChar"]:
possible_link = True
if not possible_link:
e = {
"label": entity["label"],
"startOffset": entity["startChar"],
"endOffset": entity["endChar"],
"confidence": entity["confidence"],
"provenance": "nerd-" + entity["extractor"],
"types": [],
}
if entity["extractorType"]:
all_types = entity["extractorType"].split(",")
for extracted_type in all_types:
if "dbpedia" in extracted_type:
type_data = {
"typeURI": extracted_type,
"typeLabel": None,
"wikiURI": DbpediaLink.get_english_resource_from_english_wikipedia_link(entity["uri"]),
"entityURI": entity["uri"],
"confidence": entity["confidence"],
}
else:
type_data = {
"typeURI": None,
"typeLabel": extracted_type,
"wikiURI": DbpediaLink.get_english_resource_from_english_wikipedia_link(entity["uri"]),
"entityURI": entity["uri"],
"confidence": entity["confidence"],
}
e["types"].append(type_data)
if entity["nerdType"]:
nerd_type_data = {
"typeURI": entity["nerdType"],
"typeLabel": entity["nerdType"].split("#")[1],
"wikiURI": DbpediaLink.get_english_resource_from_english_wikipedia_link(entity["uri"]),
"entityURI": entity["uri"],
"confidence": entity["confidence"],
}
e["types"].append(nerd_type_data)
initial_entities.append(e)
return initial_entities
def process_tdh_api(text):
url = "http://entityclassifier.eu/thd/api/v2/extraction?"
lang = "en"
content_format = "json"
entity_types = ["ne", "ce"]
priority_entity_linking = "true"
api_key = read_api_key()
link_matches = HyperLink.extract_all_url(text)
initial_entities = []
for entity_type in entity_types:
request = "apikey=" + api_key + \
"&format=" + content_format + \
"&lang=" + lang + \
"&priority_entity_linking=" + priority_entity_linking + \
"&entity_type=" + entity_type
try:
entities = requests.post(url + request, urllib.quote_plus(text))
entities = json.loads(entities.text)
except:
return []
for entity in entities:
possible_link = False
for link_match in link_matches:
if link_match["start"] <= entity["startOffset"] and link_match[
"end"] >= entity["endOffset"]:
possible_link = True
if not possible_link:
e = {
"label": entity["underlyingString"],
"startOffset": entity["startOffset"],
"endOffset": entity["endOffset"],
"confidence": None,
"provenance": "thd",
"types": []
}
types = []
try:
for data_type in entity["types"]:
thd_type = {
"typeURI":
data_type["typeURI"],
"typeLabel":
data_type["typeLabel"],
"entityURI":
data_type["entityURI"],
"confidence":
data_type["salience"]["confidence"],
"wikiURI":
DbpediaLink.
get_english_wikipedia_link_from_english_resource(
data_type["entityURI"]),
}
types.append(thd_type)
e["types"].append(types)
except KeyError:
continue
initial_entities.append(e)
initial_entities = {v["label"]: v for v in initial_entities}.values()
return initial_entities
