intro.ipynb
6.31 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# dh_msra 说明\n",
"0. **下载地址：** [Github](https://github.com/SophonPlus/ChineseNlpCorpus/raw/master/datasets/dh_msra/dh_msra.zip)\n",
"1. **数据概览：** 5 万多条中文命名实体识别标注数据（[IOB2](https://dl.acm.org/citation.cfm?id=977059) 格式，符合 [CoNLL 2002](https://www.clips.uantwerpen.be/conll2002/ner/) 和 [CRF++](https://taku910.github.io/crfpp/#format) 标准）\n",
"2. **推荐实验：** 中文命名实体识别\n",
"2. **数据来源：** 不详\n",
"3. **原数据集：** [zh-NER-TF](https://github.com/Determined22/zh-NER-TF)，网上搜集，具体作者、来源不详，可能是来自于 MSRA 的语料\n",
"4. **加工处理：**\n",
" 1. 将原来 2 个文件 (train 和 test) 整合到 1 个文件中"
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"import codecs\n",
"import random\n",
"\n",
"import numpy as np"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
"path = 'dh_msra_文件夹_所在_路径'"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# 1. dh_msra.txt"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 加载数据"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [],
"source": [
"def load_iob2(file_path):\n",
" '''加载 IOB2 格式的数据'''\n",
" token_seqs = []\n",
" label_seqs = []\n",
" tokens = []\n",
" labels = []\n",
" with codecs.open(file_path) as f:\n",
" for index, line in enumerate(f):\n",
" items = line.strip().split()\n",
" if len(items) == 2:\n",
" token, label = items\n",
" tokens.append(token)\n",
" labels.append(label)\n",
" elif len(items) == 0:\n",
" if tokens:\n",
" token_seqs.append(tokens)\n",
" label_seqs.append(labels)\n",
" tokens = []\n",
" labels = []\n",
" else:\n",
" print('格式错误。行号：{} 内容：{}'.format(index, line))\n",
" continue\n",
" \n",
" if tokens: # 如果文件末尾没有空行，手动将最后一条数据加入序列的列表中\n",
" token_seqs.append(tokens)\n",
" label_seqs.append(labels) \n",
" \n",
" return np.array(token_seqs), np.array(label_seqs)\n",
"\n",
"\n",
"def show_iob2(token_seqs, label_seqs, num=5, shuffle=True):\n",
" '''显示 IOB2 格式数据'''\n",
" if shuffle:\n",
" length = len(token_seqs)\n",
" indexes = [random.randrange(0, length) for i in range(num)] \n",
" zip_seqs = zip(token_seqs[indexes], label_seqs[indexes])\n",
" else:\n",
" zip_seqs = zip(token_seqs[0:num], label_seqs[0:num])\n",
" \n",
" for tokens, labels in zip_seqs:\n",
" for token, label in zip(tokens, labels):\n",
" print('{}/{} '.format(token, label), end='')\n",
" print('\\n')"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"55289 55289\n",
"\n",
"目/O 前/O “/O 继/B-PER 生/I-PER ”/O 共/O 产/O 仔/O 5/O 胎/O ，/O 产/O 下/O 小/O 老/O 虎/O 1/O 8/O 只/O ，/O 堪/O 称/O 虎/O 妈/O 妈/O 中/O 的/O 英/O 雄/O 。/O \n",
"\n",
"历/O 史/O 的/O 内/O 涵/O 是/O 很/O 丰/O 富/O 的/O ，/O 经/O 典/O 作/O 家/O 的/O 论/O 断/O 固/O 然/O 有/O 其/O 权/O 威/O 性/O 和/O 合/O 理/O 性/O ，/O 但/O 历/O 史/O 学/O 家/O 显/O 然/O 不/O 能/O 局/O 限/O 于/O 此/O 。/O \n",
"\n",
"5/O 月/O 3/O 0/O 日/O 在/O 中/B-LOC 国/I-LOC 革/I-LOC 命/I-LOC 军/I-LOC 事/I-LOC 博/I-LOC 物/I-LOC 馆/I-LOC 开/O 幕/O 的/O 全/O 国/O 禁/O 毒/O 展/O 览/O ，/O 在/O 社/O 会/O 上/O 引/O 起/O 了/O 强/O 烈/O 的/O 反/O 响/O 。/O \n",
"\n",
"另/O 外/O ，/O 还/O 有/O 一/O 个/O 惊/O 人/O 的/O 发/O 现/O ：/O 有/O 的/O 发/O 展/O 中/O 国/O 家/O 人/O 均/O 国/O 民/O 资/O 源/O 非/O 常/O 丰/O 富/O ，/O 但/O 发/O 展/O 不/O 起/O 来/O 的/O 原/O 因/O 在/O 于/O 教/O 育/O 水/O 平/O 太/O 低/O 、/O 对/O 技/O 术/O 的/O 理/O 解/O 和/O 把/O 握/O 太/O 低/O 、/O 管/O 理/O 水/O 平/O 太/O 低/O 等/O 等/O ，/O 一/O 句/O 话/O ，/O 智/O 力/O 资/O 本/O 太/O 贫/O 乏/O 。/O \n",
"\n",
"这/O 还/O 要/O 看/O 进/O 一/O 步/O 深/O 入/O 调/O 查/O 的/O 结/O 果/O 。/O \n",
"\n"
]
}
],
"source": [
"token_seqs, label_seqs = load_iob2(path+'dh_msra.txt')\n",
"\n",
"print(len(token_seqs), len(label_seqs))\n",
"print() \n",
"show_iob2(token_seqs, label_seqs)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 标签说明\n",
"\n",
"| 标签 | 说明 |\n",
"| ---- | ---- |\n",
"| LOC | 地点 (LOCATION) |\n",
"| ORG | 机构 (ORGANIZATION) |\n",
"| PER | 人物 (PERSON) |"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"{'B-LOC', 'B-ORG', 'B-PER', 'I-LOC', 'I-ORG', 'I-PER', 'O'}"
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"set([label for labels in label_seqs for label in labels])"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.5"
},
"widgets": {
"state": {},
"version": "1.1.2"
}
},
"nbformat": 4,
"nbformat_minor": 2
}