def test_1_gram(in_path, out_path):
lfreq, ltotal = build_pfdict(dic_path)
# print(lfreq, ltotal)
seg = []
start_time = time.time()
segmentation_mr(in_path, seg, lfreq, ltotal)
end_time = time.time()
print(round((end_time - start_time) * 1000), 'ms')
output_2_file(seg, out_path)
# 分词评价
analysis(out_path, gold_std_path)
def test_2_gram(in_path, out_path):
lfreq, ltotal = build_pfdict(dic_path)
pfreq = pro_prefix_dictionary()
seg = []
start_time = time.time()
segmentation_graph(in_path, seg, lfreq, pfreq)
end_time = time.time()
print(round((end_time - start_time) * 1000), 'ms')
output_2_file(seg, out_path)
# 分词评价
analysis(out_path, gold_std_path)
def main():
lfreq, ltotal = build_pfdict(dic_path)
pfreq = pro_prefix_dictionary()
seg = []
start_time = time.time()
segmentation_bigram(sent_path, seg, lfreq, pfreq)
end_time = time.time()
print(round((end_time - start_time) * 1000), 'ms')
output_2_file(seg, seg_LM_bg_path)
analysis(seg_LM_bg_path, gold_std_path)
def main():
seg_text = forward_maximum_matching(dic_path, sent_path)
# print(seg_text)
output_2_file(seg_text, seg_FMM_path)