def main(f):
word_dic = defaultdict(int)
sl = knock30.parse(f)
for s in sl:
for word in s:
word_dic[word["surface"]] += 1
word_freq = [v for k, v in sorted(word_dic.items(), key=lambda x:x[1], reverse=True)]
plt.hist(word_freq)
plt.savefig("output_knock38.png")
def main(f):
word_dic = defaultdict(int)
sl = knock30.parse(f)
for s in sl:
for word in s:
word_dic[word["surface"]] += 1
for k, v in sorted(word_dic.items(), key=lambda x: x[1],
reverse=True)[0:10]:
print k, v
def main(f):
word_dic = defaultdict(int)
sl = knock30.parse(f)
for s in sl:
for word in s:
word_dic[word["surface"]] += 1
word_freq = sorted(word_dic.values(), reverse=True)
plt.xscale("log")
plt.yscale("log")
plt.bar(range(len(word_freq)), word_freq)
plt.savefig("output_knock39.png")
def main(f):
word_dic = defaultdict(int)
sl = knock30.parse(f)
for s in sl:
for word in s:
word_dic[word["surface"]] += 1
top_ten = []
top_ten_labels = []
for k, v in sorted(word_dic.items(), key=lambda x:x[1], reverse=True)[0:10]:
top_ten.append(v)
top_ten_labels.append(k)
print top_ten_labels
plt.bar(range(10), top_ten, tick_label=top_ten_labels)
plt.savefig("output_knock37.png")
def main(f):
sl = knock30.parse(f)
flag_of_noun = False
flag2 = False
for dic_list in sl:
for d in dic_list:
if flag2 and d["pos"] == "名詞":
print meisiku + d["surface"]
frag_of_noun = False
flag2 = False
elif flag2 and d["pos"] != "名詞":
flag_of_noun = False
flag2 = False
elif flag_of_noun and d["surface"] == "の":
flag2 = True
meisiku += "の"
elif flag_of_noun and d["surface"] != "の":
flag_of_noun = False
elif d["pos"] == "名詞":
flag_of_noun = True
meisiku = d["surface"]
def main(f):
sl = knock30.parse(f)
flag_of_noun = False
flag_of_noun2 = False
meisi = ""
for dic_list in sl:
for d in dic_list:
if flag_of_noun2 and d["pos"] != "名詞":
print meisi
meisi = ""
flag_of_noun = False
flag_of_noun2 = False
elif flag_of_noun and d["pos"] != "名詞":
flag_of_noun = False
meisi = ""
elif flag_of_noun and d["pos"] == "名詞":
flag_of_noun2 = True
meisi += d["surface"]
elif not (flag_of_noun) and d["pos"] == "名詞":
flag_of_noun = True
meisi += d["surface"]
def main(f):
sl = knock30.parse(f)
for dic_list in sl:
for d in dic_list:
if d["pos"] == "動詞":
print d["surface"]
def main(f):
sl = knock30.parse(f)
for dic_list in sl:
for d in dic_list:
if d["pos"] == "名詞" and d["pos1"] == "サ変接続":
print d["surface"]