def distance(self, query, entity):
q_len = len(query)
e_len = len(entity)
if q_len == 0 or e_len == 0:
return 1.0
else:
query = dict(sorted(query.iteritems(), key=operator.itemgetter(1), reverse=True)[:100])
entity = dict(sorted(entity.iteritems(), key=operator.itemgetter(1), reverse=True)[:100])
query_wrs = list(self.iter_word_reps(query))
entity_wrs = list(self.iter_word_reps(entity))
a = entity_wrs if e_len >= q_len else query_wrs
b = entity_wrs if q_len > e_len else query_wrs
total = 0.0
for wb in b:
max_sim = 0.0
for wa in a:
sim = dense_cosine_distance(query, entity)
#log.debug("SIM: %f", sim)
if sim > max_sim:
max_sim = sim
total += max_sim
#llm_sim = sum(max(self.dense_cosine(wa, wb) for wa in a) for wb in b) / float(len(b))
#log.debug('LLM SIM: %f - %i - %i - %f' % (total, len(b), len(a), total / len(b)))
return 1.0 - (total / len(b))
def distance(self, query, entity):
q_len = len(query)
e_len = len(entity)
if q_len == 0 or e_len == 0:
return 1.0
else:
query = dict(
sorted(query.iteritems(),
key=operator.itemgetter(1),
reverse=True)[:100])
entity = dict(
sorted(entity.iteritems(),
key=operator.itemgetter(1),
reverse=True)[:100])
query_wrs = list(self.iter_word_reps(query))
entity_wrs = list(self.iter_word_reps(entity))
a = entity_wrs if e_len >= q_len else query_wrs
b = entity_wrs if q_len > e_len else query_wrs
total = 0.0
for wb in b:
max_sim = 0.0
for wa in a:
sim = dense_cosine_distance(query, entity)
#log.debug("SIM: %f", sim)
if sim > max_sim:
max_sim = sim
total += max_sim
#llm_sim = sum(max(self.dense_cosine(wa, wb) for wa in a) for wb in b) / float(len(b))
#log.debug('LLM SIM: %f - %i - %i - %f' % (total, len(b), len(a), total / len(b)))
return 1.0 - (total / len(b))
def distance(self, query, entity):
query = self.bow_to_dbow(query)
entity = self.bow_to_dbow(entity)
return dense_cosine_distance(query, entity)
def distance(self, query, entity):
query = self.bow_to_dbow(query)
entity = self.bow_to_dbow(entity)
return dense_cosine_distance(query, entity)
