def __sample_file(self, dir, file):
"""Creates a local from a specific file in a given directory.
:param dir: directory (relative to path_to_dbpedia)
:param file:
"""
t = Triple()
p = NTriplesParser(t)
infile = os.path.join(self.path_to_dbpedia, dir, file)
outfile = os.path.join(self.output_dir, dir, file)
print("Processing file " + file + " ...")
i = 0
with FileUtils.open_file_by_type(infile) as fin:
fout = FileUtils.open_file_by_type(
outfile,
mode="w") # output file will be of the same type as the input
for line in fin:
try:
p.parsestring(line.decode("utf-8"))
except ParseError: # skip lines that couldn't be parsed
continue
if t.subject() is None: # only if parsed as a triple
continue
subj = self.prefix.get_prefixed(
t.subject()) # prefixing subject
if subj in self.sample_entities:
fout.write(line)
i += 1
if i % 100000 == 0:
print(str(i // 1000) + "K lines processed")
fout.close()
def batch_identification(self):
"""Annotates, in a batch, queries with identified target types, and outputs results."""
queries = json.load(FileUtils.open_file_by_type(self.__query_file))
f_trec_out = None
if "trec_output_file" in self.__config: # for TREC-formatted outputting
f_trec_out = FileUtils.open_file_by_type(
self.__config["trec_output_file"], mode="w")
results = dict()
for query_id in sorted(queries):
PLOGGER.info("Identifying target types for [{}] {}".format(
query_id, queries[query_id]))
results[query_id] = self.identify(queries[query_id])
# Output resulting scores in TREC format if required
if f_trec_out:
type_to_score = dict()
for d in results.get(query_id, {}).get("results", {}).values():
type_to_score[d["type"]] = d["score"]
ret_res = RetrievalResults(type_to_score)
ret_res.write_trec_format(query_id,
self.__config["run_id"],
f_trec_out,
max_rank=self.__config["num_docs"])
json.dump(results,
FileUtils.open_file_by_type(self.__output_file, mode="w"),
indent=4,
sort_keys=True)
PLOGGER.info("Output file: {}".format(self.__output_file))
if f_trec_out:
f_trec_out.close()
def __init__(self):
self.persons_ids = {}
self.ids_persons = {}
for line in FileUtils.read_file_as_list(PERSONS_IDS_F):
person, id = line.split("\t", maxsplit=1)
self.persons_ids[person] = id
self.ids_persons[id] = person
self.professions_ids = {}
self.ids_professions = {}
for line in FileUtils.read_file_as_list(PROFESSIONS_IDS_F):
prof, id = line.split("\t", maxsplit=1)
self.professions_ids[prof] = id
self.ids_professions[id] = prof
self.nationalities_ids = {}
self.ids_nationalities = {}
for line in FileUtils.read_file_as_list(NATIONALITIES_IDS_F):
nation, id = line.split("\t", maxsplit=1)
self.nationalities_ids[nation] = id
self.ids_nationalities[id] = nation
self.nationalities_countries = {}
self.countries_nationalities = {}
for line in FileUtils.read_file_as_list(COUNTRIES_NATIONALITIES_F):
country, nation = line.split("\t", maxsplit=1)
self.nationalities_countries[nation] = country
self.countries_nationalities[country] = nation
def load_nordlys_config(file_name):
"""Loads nordlys config file. If local file is provided, global one is ignored."""
config_path = os.sep.join([BASE_DIR, "config"])
local_config = os.sep.join([config_path, "local", file_name])
if os.path.exists(local_config):
return FileUtils.load_config(local_config)
else:
return FileUtils.load_config(os.sep.join([config_path, file_name]))
def generate_features(self, kb_file, output_file1, output_file2):
"""Generate features of freq-person-nationality"""
fout1 = open(output_file1, "w")
fout2 = open(output_file2, "w")
# write tsv header
header = ["person", "nationality", "freq_person_nationality_noun"]
fout1.write("\t".join(header) + "\n")
header = ["person", "nationality", "freq_person_nationality_adj"]
fout2.write("\t".join(header) + "\n")
with FileUtils.open_file_by_type(kb_file) as kb_f:
line_count = 1
for line in kb_f:
print(line_count)
line_count += 1
person_id, nat_id, noun, adj = line.strip().split("\t")
values_noun = [person_id, nat_id]
values_adj = [person_id, nat_id]
nats = [noun, adj]
fpn_noun, fpn_adj = self.get_per_nat_tf(person_id, nats)
values_noun.append(str(fpn_noun))
values_adj.append(str(fpn_adj))
fout1.write("\t".join(values_noun) + "\n")
fout2.write("\t".join(values_adj) + "\n")
fout1.close()
fout2.close()
def __load_entity_abstracts(self, filename):
prefix = URIPrefix()
t = Triple()
p = NTriplesParser(t)
lines_counter = 0
PLOGGER.info("Loading entity abstracts from {}".format(filename))
for line in FileUtils.read_file_as_list(filename):
# basic line parsing
line = line.decode("utf-8") if isinstance(line, bytes) else line
try:
p.parsestring(line)
except ParseError: # skip lines that couldn't be parsed
continue
if t.subject() is None: # only if parsed as a triple
continue
# Subject and object identification
subj = prefix.get_prefixed(t.subject())
obj = ""
if type(t.object()) is URIRef:
# PLOGGER.error("Error: it is URIRef the parsed obj")
pass
else:
obj = t.object().encode("utf-8")
if len(obj) == 0:
continue # skip empty objects
self.__entity_abstracts[subj] = obj
lines_counter += 1
if lines_counter % 10000 == 0:
PLOGGER.info("\t{}K lines processed".format(lines_counter // 1000))
pass
PLOGGER.info("\n### Loading entity abstracts... Done.")
def __load_items_stats(self, items_tfidf_fpath):
"""Loads pre-computed tf-idf stats for items.
:param items_tfidf_fpath:
:return:
"""
item_term_weights = {}
for line in FileUtils.read_file_as_list(items_tfidf_fpath):
item, term, _, _, weight = line.split("\t", maxsplit=4)
if len(term) < 4: # avoid short terms
continue
# some cleanings
if term.startswith("_"):
term = term[1:]
if term.endswith("_"):
term = term[:-1]
if term.startswith(WP_PREFIX): # it's a person name
term = term.split(WP_PREFIX)[-1] # remove prefix
term = "_".join([
word[0].upper() + word[1:] for word in term.split("_")
]) # capitalize every word
item_d = item_term_weights.get(item, {})
item_d[term] = float(weight)
item_term_weights[item] = item_d
return item_term_weights
def main(args):
s_t = time.time() # start time
config = FileUtils.load_config(args.config) if args.config != "" else get_config()
r = Retrieval(config)
r.batch_retrieval()
e_t = time.time() # end time
print("Execution time(min):\t" + str((e_t - s_t) / 60) + "\n")
def main(args):
config = FileUtils.load_config(args.config)
el = EL(config, Entity())
if args.query:
res = el.link(args.query)
pprint(res)
else:
el.batch_linking()
def main(args):
config = FileUtils.load_config(args.config)
tti = TTI(config)
if args.query:
res = tti.identify(args.query)
pprint(res)
else:
tti.batch_identification()
def main(args):
config = FileUtils.load_config(args.config)
er = ER(config, ElasticCache(DBPEDIA_INDEX))
if args.query:
res = er.retrieve(args.query)
pprint(res)
else:
er.batch_retrieval()
def main(args):
config = FileUtils.load_config(args.config)
er = ER(config)
if args.query:
res = er.retrieve(args.query)
pprint(res)
else:
er.batch_retrieval()
def make_persons_fb_ids(persons_fpath):
"""Our person ID is the Freebase ID where the prefix 'm.' is replaced with 'fb_'.
:param persons_fpath: 'persons'-named file path.
"""
with open(PERSONS_IDS_F, "w") as f_out:
for line in FileUtils.read_file_as_list(persons_fpath):
person, raw_fb_id = line.split("\t", maxsplit=1)
fb_id = "fb_" + raw_fb_id[2:]
f_out.write("{}\t{}\n".format(person, fb_id))
def make_professions_kb_translation(dest_translation_fpath, persons_ids_fpath, professions_ids_fpath,
person_items_fpath):
"""A person ID -to- item ID translation schema is convenient.
:param dest_translation_fpath: destination file path of IDs translation.
:param persons_ids_fpath: person IDs file path.
:param professions_ids_fpath: profession IDs file path.
:param person_items_fpath: path to the file with person items (a '.kb'-extension file).
:return:
"""
persons_ids = {}
for line in FileUtils.read_file_as_list(persons_ids_fpath):
person, id = line.split("\t", maxsplit=1)
persons_ids[person] = id
professions_ids = {}
for line in FileUtils.read_file_as_list(professions_ids_fpath):
prof, id = line.split("\t", maxsplit=1)
professions_ids[prof] = id
person_items = {}
for line in FileUtils.read_file_as_list(person_items_fpath):
person, item = line.split("\t", maxsplit=1)
person_items.get(person, []).append(item)
translations = {}
for person, items in person_items.items():
if person not in persons_ids:
continue
person_id = persons_ids[person]
items_ids = []
for item in items:
if item not in professions_ids:
continue
items_ids.append(professions_ids[item])
translations[person_id] = items_ids
with open(dest_translation_fpath, "w") as f_out:
for person_id, items_ids in translations.items():
for item_id in items_ids:
f_out.write("{}\t{}\n".format(person_id, item_id))
def generate_features(self, kb_file, output_file):
"""Generating features related to term statistics"""
fout = open(output_file, "w")
# write tsv header
header = ["person", "profession"]
for k in self.K_VALUES:
header.append("sumProfTerms_" + str(k))
header.append("simCos_" + str(k))
fout.write("\t".join(header) + "\n")
with FileUtils.open_file_by_type(kb_file) as kb_f:
for line in kb_f:
person_id, prof_id = line.strip().split("\t")
values = [person_id, prof_id]
person_tf, num_sent = self.get_person_tf(person_id)
for k in self.K_VALUES:
# we take top-K profession terms
# Compute sumProfTerms: \sum_{t \in T_k(pr)}\sum_{s \in S(pe)} tf(t,s) w(t,pr)
# where w(t,pe )= TFIDF(t,pr) = \frac{\sum_{s \in S(pr)} tf(t,s)}
sum_prof_terms = 0
for term, tf in person_tf.items():
pt = self.__stats.get(prof_id, {}).get(term, {})
if pt.get("rank",
100000) > k: # skip term if not in top-K
continue
sum_prof_terms += tf * pt.get("tfidf", 0)
values.append(str(sum_prof_terms))
# compute simCosK
# where K is the top-K terms for the profession
vec_pr = [] # construct prof vector
vec_pe = [] # construct person vector
if prof_id in self.__stats:
for term, s in self.__stats[prof_id].items():
if s["rank"] <= k:
vec_pr.append(s["tfidf"])
idf = s["tfidf"] / s[
"tf"] # we back-generate IDF from profession's TF-IDF
vec_pe.append(person_tf.get(term, 0) * idf)
cos = cos_sim(vec_pr, vec_pe)
else:
cos = 0 # in some exceptional cases the profession does not have any sentences
values.append(str(cos))
fout.write("\t".join(values) + "\n")
print(values)
fout.close()
def main(args):
config = FileUtils.load_config(args.config)
type2entity_file = os.path.expanduser(os.path.join(config.get("type2entity_file", "")))
entity_abstracts_file = os.path.expanduser(os.path.join(config.get("entity_abstracts_file", "")))
if (not os.path.isfile(type2entity_file)) or (not os.path.isfile(entity_abstracts_file)):
exit(1)
indexer = IndexerDBpediaTypes(config)
indexer.build_index(force=True)
PLOGGER.info("Index build: <{}>".format(indexer.name))
def add_file(self, filename, reverse_triple=False, predicate_prefix=None):
"""Adds contents from an NTriples file to MongoDB.
:param filename: NTriples file.
:param reverse_triple: if set True, the subject and object values are swapped.
:param predicate_prefix: prefix to be added to predicates.
:param subjects_redirecter: redirects dict.
"""
print("Processing " + filename + "...")
t = Triple()
p = NTriplesParser(t)
self.__m_id = None # document id for MongoDB -- subj
self.__m_contents = None # document contents for MongoDB -- pred, obj
i = 0
with FileUtils.open_file_by_type(filename) as f:
for line in f:
try:
p.parsestring(line.decode("utf-8"))
except ParseError: # skip lines that couldn't be parsed
continue
if t.subject() is None: # only if parsed as a triple
continue
# subject prefixing
subj = self.__prefix.get_prefixed(t.subject())
# predicate prefixing
pred = self.__prefix.get_prefixed(t.predicate())
if predicate_prefix is not None:
pred = predicate_prefix + pred
# Object prefixing
if type(t.object()) is URIRef:
obj = self.__prefix.get_prefixed(t.object())
else:
obj = t.object()
if len(obj) == 0:
continue # skip empty objects
# write or append
if reverse_triple: # reverse subj and obj
self._next_triple(obj, pred, subj)
else: # normal mode
self._next_triple(subj, pred, obj)
i += 1
if i % 100000 == 0:
print(
str(i // 1000) + "K lines processed from " + filename)
# process last triple
self._write_to_mongo()
def make_relation_item_ids(rel_items_fpath):
"""Our relation ID is the original relation where the prefix any dash or blank is replaced with underscore.
:param rel_items_fpath: items file path.
"""
basename = os.path.basename(rel_items_fpath) # professions or nationalities
with open(os.sep.join([DATA_DIR, "{}_ids.tsv".format(basename)]), "w") as f_out:
for item in FileUtils.read_file_as_list(rel_items_fpath):
id = item.lower().replace(" ", "_").replace("-", "_")
f_out.write("{}\t{}\n".format(item, id))
def main(args):
conf = FileUtils.load_config(args.config)
el = EL(conf, Entity(), ElasticCache(DBPEDIA_INDEX), FeatureCache())
if conf.get("gen_model", False):
LTR.train(conf)
elif args.query:
res = el.link(args.query)
pprint(res)
else:
el.batch_linking()
def build(self):
"""Builds word2vec collection from GoogleNews 300-dim pre-trained corpus."""
self.__mongo = Mongo(MONGO_HOST, MONGO_DB, self.__collection)
self.__mongo.drop()
infile = FileUtils.open_file_by_type(self.__w2v_fname)
i = 0
for line in infile:
term, vector = self.__parse_line(line)
self.__mongo.add(term, {'vector': vector})
i += 1
if i % 1000 == 0:
PLOGGER.info(str(i / 1000) + "K lines are loaded.")
def dump_differences(self, output_file):
"""Outputs query-level differences between two methods into a tab-separated file.
The first method is considered the baseline, the differences are with respect to that.
Output format: queryID res1 res2 diff(res2-res1)
"""
te_method1 = TrecEval()
te_method1.evaluate(self.__qrels, self.__run1_file)
te_method2 = TrecEval()
te_method2.evaluate(self.__qrels, self.__run2_file)
data = []
for query_id in te_method1.get_query_ids():
res1 = te_method1.get_score(query_id, self.__metric)
res2 = te_method2.get_score(query_id, self.__metric)
data.append([query_id, res1, res2, round(res2 - res1, 4)])
# sorts based on the differences desc
sorted_data = sorted(data, key=lambda l: l[3], reverse=True)
FileUtils.dump_tsv(output_file,
sorted_data,
header=["queryID", "method1", "method2", "diff"])
def generate_features(self, kb_file, output_file):
"""Core function for generating into output_file the features, with person-item data from kb_file.
:param kb_file: path to the file with person items (a '.kb'-extension file).
:param output_file:
:return:
"""
feat_w2v_approx = FeaturesW2VSimApprox()
with open(output_file, "w") as f_out:
# write tsv header
header = ["person_id", "prof_id"]
for k in self.K_VALUES:
header.append("simCos_w2v_" + str(k))
f_out.write("\t".join(header) + "\n")
for line in FileUtils.read_file_as_list(kb_file):
person_id, prof_id = line.split(
"\t") # strip() done in read_file_as_list()
values = [person_id, prof_id]
person_tf, num_sent = self.get_person_tf(person_id)
for k in self.K_VALUES:
# we take top-K profession terms
# compute simCosK
# where K is the top-K terms for the profession
term_weights_pr = {
} # dict from top-K profession terms to their tfidf weights
term_weights_pe = {
} # dict from top-K person terms to their tfidf weights
if prof_id in self.__stats:
for term, s in self.__stats[prof_id].items():
if s["rank"] <= k:
term_weights_pr[term] = float(s["tfidf"])
idf = s["tfidf"] / s[
"tf"] # we back-generate IDF from profession's TF-IDF
term_weights_pe[term] = person_tf.get(term,
0) * idf
vec_pr = feat_w2v_approx.get_vector(term_weights_pr)
vec_pe = feat_w2v_approx.get_vector(term_weights_pe)
cos = cos_sim(vec_pr, vec_pe)
else:
cos = 0 # in some exceptional cases the profession does not have any sentences
values.append(str(cos))
f_out.write("\t".join(values) + "\n")
def main(args):
config = FileUtils.load_config(args.config)
if "_uri" not in config["index_name"]:
print("index name might not be correct, please check again!")
exit(0)
indexer = IndexerDBpediaURI(config)
fields_file = config.get("fields_file", "output/field_counts.json")
if "fields_file" not in config:
field_counts2json(fields_file)
indexer.build()
print("Index build: " + config["index_name"])
def main(args):
config = FileUtils.load_config(args.config)
dbpedia_path = config.get("dbpedia_files_path", "")
# Check DBpedia files
PLOGGER.info("Checking needed DBpedia files under {}".format(dbpedia_path))
for fname in [ENTITY_ABSTRACTS_FILE] + ENTITY_TYPES_FILES:
if os.path.isfile(os.sep.join([dbpedia_path, fname])):
PLOGGER.info(" - {}: OK".format(fname))
else:
PLOGGER.error(" - {}: Missing".format(fname))
exit(1)
indexer = IndexerDBpediaTypes(config)
indexer.build_index(force=True)
def __load_entity_abstracts(self):
num_lines = 0
filename = os.sep.join([self.__dbpedia_path, ENTITY_ABSTRACTS_FILE])
PLOGGER.info("Loading entity abstracts from {}".format(filename))
for line in FileUtils.read_file_as_list(filename):
entity, abstract = self.__parse_line(line)
if abstract and len(abstract) > 0: # skip empty objects
self.__entity_abstracts[entity] = abstract
num_lines += 1
if num_lines % 10000 == 0:
PLOGGER.info(" {}K lines processed".format(num_lines // 1000))
PLOGGER.info(" Done.")
def main(args):
config = FileUtils.load_config(args.config)
if "_uri" not in config["index_name"]:
PLOGGER.error("index name might not be correct, please check again!")
exit(0)
if "fields_file" not in config:
fields_count = compute_field_counts()
else:
fields_count = json.load(config["fields_file"])
indexer = IndexerDBpediaURI(config, fields_count)
indexer.build()
PLOGGER.info("Index build: " + config["index_name"])
def main(args):
config = FileUtils.load_config(args.config)
elastic_term = ElasticCache(config["text_index"])
lambdas = config.get("lambdas", [0.9, 0.1])
queries = json.load(open(config["query_file"], "r"))
mappings = json.load(open(config["mapping_file"], "r"))
annots = load_annot(config["annot_file"])
run = load_run(config["run_file"])
instances = Instances()
# gets the results
out_file = open(config["output_file"], "w")
qid_int = 0
for qid, query in sorted(queries.items()):
print("Scoring ", qid, "...")
results, libsvm_str = {}, ""
query_len = len(elastic_term.analyze_query(query).split())
scorer = ScorerELR(ElasticCache(config["uri_index"]), annots[qid],
query_len, lambdas)
for doc_id, p_T_d in sorted(run[qid].items()):
query_mappings = get_mapping_query(annots[qid], mappings)
p_E_d = scorer.score_doc(doc_id, query_mappings)
properties = {
'doc_id': doc_id,
'query': query,
'qid': qid,
'qid_int': qid_int
}
features = {'p_T_d': p_T_d, 'p_E_d': p_E_d}
ins = Instance(qid + "_" + doc_id,
features=features,
properties=properties)
instances.add_instance(ins)
# libsvm_str += ins.to_libsvm(qid_prop="qod_int")
results[doc_id] = (lambdas[0] * p_T_d) + (lambdas[1] * p_E_d)
qid_int += 1
# Write trec format
out_str = trec_format(results, qid, "elr")
out_file.write(out_str)
out_file.close()
print("Output file:", config["output_file"])
instances.to_json(config["json_file"])
print("Output file:", config["json_file"])
def main(args):
config = FileUtils.load_config(args.config)
type2entity_file = os.path.expanduser(config.get("type2entity_file", ""))
if not os.path.isfile(type2entity_file):
print("invalid path to type-to-entity source file: ", type2entity_file)
exit(1)
entity_abstracts_file = os.path.expanduser(
config.get("entity_abstracts_file", ""))
if not os.path.isfile(entity_abstracts_file):
print("invalid path to entity abstracts source file: ",
entity_abstracts_file)
exit(1)
indexer = IndexerDBpediaTypes(config, type2entity_file,
entity_abstracts_file)
indexer.build_index(force=True)
print("Index build: <{}>".format(indexer.name))
def main(args):
example_config = {"index_name": "toy_index",
# "query_file": "data/queries/test_queries.json",
"first_pass": {
"num_docs": 1000,
"field": "content",
# "model": "LMJelinekMercer",
# "model_params": {"lambda": 0.1}
},
"second_pass": {
"field": "content",
"model": "lm",
"smoothing_method": "jm",
"smoothing_param": 0.1
},
"output_file": "output/test_retrieval.txt"
}
config = FileUtils.load_config(args.config) if args.config != "" else example_config
r = Retrieval(config)
r.batch_retrieval()
def read_fb2dbp_file(self, is_39=False):
"""Reads the file and generates an initial mapping of Freebase to DBpedia IDs.
Only proper DBpedia entities are considered; i.e. redirect and disambiguation pages are ignored.
"""
fb2dbp_file = self.__fb2dbp_file_39 if is_39 else self.__fb2dbp_file
print("Processing " + fb2dbp_file + "...")
t = Triple()
p = NTriplesParser(t)
i = 0
fb2dbp_mapping = defaultdict(set)
with FileUtils.open_file_by_type(fb2dbp_file) as f:
for line in f:
try:
p.parsestring(line.decode("utf-8"))
except ParseError: # skip lines that couldn't be parsed
continue
if t.subject() is None: # only if parsed as a triple
continue
# prefixing
dbp_id = self.__prefix.get_prefixed(t.subject())
fb_id = self.__prefix.get_prefixed(t.object())
# if reading 3.9 file, converts ID to 2015-10 version
if is_39:
dbp_id = EntityUtils.convert_39_to_201510(dbp_id)
fb2dbp_mapping[fb_id].add(dbp_id)
# if reading 2015-10 file, keeps only the proper DBpedia entities
else:
entity_utils = EntityUtils(
self.__mongo_dbpedia.find_by_id(dbp_id))
if entity_utils.is_entity():
fb2dbp_mapping[fb_id].add(dbp_id)
i += 1
if i % 1000 == 0:
print(str(i // 1000) + "K lines are processed!")
return fb2dbp_mapping
