def rank_alignment(embeds1,
embeds2,
mapping,
top_k,
threads_num,
metric='inner',
normalize=False,
csls_k=0,
accurate=True):
t = time.time()
if mapping is None:
mr_12, mrr_12, hits_12, hits_12_list = greedy_alignment(
embeds1, embeds2, top_k, threads_num, metric, normalize, csls_k,
accurate)
else:
test_embeds1_mapped = np.matmul(embeds1, mapping)
mr_12, mrr_12, hits_12, hits_12_list = greedy_alignment(
test_embeds1_mapped, embeds2, top_k, threads_num, metric,
normalize, csls_k, accurate)
cost = time.time() - t
rank_candidates_num = len(hits_12_list)
if accurate:
print(
"alignment results: hits@{} = {}, mr = {:.4f}, mrr = {:.4f}, rank_candidates: {}, time = {:.3f} s. "
.format(top_k, hits_12, mr_12, mrr_12, rank_candidates_num, cost))
return mr_12, mrr_12, hits_12, hits_12_list
def valid(embeds1, embeds2, mapping, top_k, threads_num, metric='inner', normalize=False, csls_k=0, accurate=False):
if mapping is None:
_, hits1_12, mr_12, mrr_12 = greedy_alignment(embeds1, embeds2, top_k, threads_num,
metric, normalize, csls_k, accurate)
else:
test_embeds1_mapped = np.matmul(embeds1, mapping)
_, hits1_12, mr_12, mrr_12 = greedy_alignment(test_embeds1_mapped, embeds2, top_k, threads_num,
metric, normalize, csls_k, accurate)
return hits1_12, mrr_12
def rank_alignment_bidirection(embeds1,
embeds2,
mapping,
top_k,
threads_num,
metric='inner',
normalize=False,
csls_k=0,
accurate=True):
t = time.time()
if mapping is None:
mr_12, mrr_12, hits_12, hits_12_list = greedy_alignment(
embeds1, embeds2, top_k, threads_num, metric, normalize, csls_k,
accurate)
mr_21, mrr_21, hits_21, hits_21_list = greedy_alignment(
embeds2, embeds1, top_k, threads_num, metric, normalize, csls_k,
accurate)
else:
test_embeds1_mapped = np.matmul(embeds1, mapping)
mr_12, mrr_12, hits_12, hits_12_list = greedy_alignment(
test_embeds1_mapped, embeds2, top_k, threads_num, metric,
normalize, csls_k, accurate)
mr_21, mrr_21, hits_21, hits_21_list = greedy_alignment(
embeds2, test_embeds1_mapped, top_k, threads_num, metric,
normalize, csls_k, accurate)
mr = round((mr_12 + mr_12) / 2, 4)
mrr = round((mrr_12 + mrr_21) / 2, 4)
hits = ((hits_12 + hits_21) / 2).tolist()
for i in range(len(hits)):
hits[i] = round(hits[i], 4)
hits = np.array(hits)
cost = time.time() - t
rank_candidates_num = len(hits_12_list)
if accurate:
if csls_k > 0:
print(
"alignment results with csls: csls={}, hits@{} = {}, mr = {:.4f}, mrr = {:.4f}, rank_candidates: {}, time = {:.3f} s "
.format(csls_k, top_k, hits, mr, mrr, rank_candidates_num,
cost))
else:
print(
"alignment results: hits@{} = {}, mr = {:.4f}, mrr = {:.4f}, rank_candidates: {}, time = {:.3f} s. "
.format(top_k, hits, mr, mrr, rank_candidates_num, cost))
else:
if csls_k > 0:
print(
"quick results with csls: csls={}, hits@{} = {}, time = {:.4f} s "
.format(csls_k, top_k, hits, cost))
else:
print(
"alignment results: hits@{} = {}, mr = {:.4f}, mrr = {:.4f}, rank_candidates: {}, time = {:.3f} s. "
.format(top_k, hits, mr, mrr, rank_candidates_num, cost))
return mr, mrr, hits, hits_12_list, hits_21_list