File size: 6,288 Bytes
d4774de
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Example code for Named Entity Recognition with MITIE"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cell 1: Import Libraries and Set Up Paths\n",
    "Make sure to import necessary libraries and set up the path for the MITIE library.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from mitie import ner_training_instance, ner_trainer"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cell 2: Create Training Examples\n",
    "Define training examples by tokenizing sentences and annotating entities. Each entity is labeled with its respective tag, such as \"person\" or \"organization\"."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# First training example\n",
    "sample = ner_training_instance([\"Mam\", \"na\", \"imię\", \"Alicja\", \"Kowalska\", \"i\", \"pracuję\", \"w\", \"Cleantext\", \".\"])\n",
    "sample.add_entity(range(3,5), \"PERSON\")  # Alicja Kowalska as a person\n",
    "sample.add_entity(range(8,9), \"ORGANIZATION\")  # Cleantext as an organization\n",
    "\n",
    "# Second training example\n",
    "sample_2 = ner_training_instance([\"Wczoraj\", \"spotkałem\", \"się\", \"z\", \"Robertem\", \"Nowakiem\", \"z\", \"Global\", \"Tech\", \".\"])\n",
    "sample_2.add_entity(range(4,6), \"PERSON\")  # Robert Nowak as a person\n",
    "sample_2.add_entity(range(7,9), \"ORGANIZATION\")  # Global Tech as an organization\n",
    "\n",
    "samples = [sample, sample_2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cell 3: Initialize the Trainer\n",
    "Load the feature extractor and add the training examples to the trainer. Set the number of threads for faster processing."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer = ner_trainer(\"./model/total_word_feature_extractor.dat\")\n",
    "\n",
    "for sample in samples:\n",
    "    trainer.add(sample)\n",
    "\n",
    "trainer.num_threads = 8"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cell 4: Train the Model\n",
    "Train the named entity recognizer and save the trained model to disk for future use."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training to recognize 2 labels: 'PERSON', 'ORGANIZATION'\n",
      "Part I: train segmenter\n",
      "words in dictionary: 200000\n",
      "num features: 271\n",
      "now do training\n",
      "C:           20\n",
      "epsilon:     0.01\n",
      "num threads: 8\n",
      "cache size:  5\n",
      "max iterations: 2000\n",
      "loss per missed segment:  3\n",
      "C: 20   loss: 3 \t0.5\n",
      "C: 35   loss: 3 \t0.5\n",
      "C: 20   loss: 4.5 \t0.5\n",
      "C: 5   loss: 3 \t0.5\n",
      "C: 20   loss: 1.5 \t0.5\n",
      "C: 21   loss: 3 \t0.5\n",
      "C: 20   loss: 3.1 \t0.5\n",
      "C: 19   loss: 3 \t0.5\n",
      "C: 20   loss: 3 \t0.5\n",
      "best C: 20\n",
      "best loss: 3\n",
      "num feats in chunker model: 4095\n",
      "train: precision, recall, f1-score: 1 1 1 \n",
      "Part I: elapsed time: 0 seconds.\n",
      "\n",
      "Part II: train segment classifier\n",
      "now do training\n",
      "num training samples: 4\n",
      "C: 200   f-score: 0.5\n",
      "C: 400   f-score: 0.5\n",
      "C: 300   f-score: 0.5\n",
      "C: 100   f-score: 0.5\n",
      "C: 0.01   f-score: 0.5\n",
      "C: 50.005   f-score: 0.5\n",
      "C: 25.0075   f-score: 0.5\n",
      "C: 12.5088   f-score: 0.5\n",
      "C: 6.25938   f-score: 0.5\n",
      "C: 3.13469   f-score: 0.5\n",
      "C: 1.57234   f-score: 0.5\n",
      "C: 0.791172   f-score: 0.5\n",
      "C: 0.400586   f-score: 0.5\n",
      "best C: 0.791172\n",
      "test on train: \n",
      "2 0 \n",
      "0 2 \n",
      "\n",
      "overall accuracy: 1\n",
      "Part II: elapsed time: 1 seconds.\n",
      "df.number_of_classes(): 2\n"
     ]
    }
   ],
   "source": [
    "ner = trainer.train()\n",
    "ner.save_to_disk(\"./output/ner_model.dat\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cell 5: Test the Trained Model\n",
    "Test the trained model with a sample sentence and display the detected entities along with their tags.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tags: ['PERSON', 'ORGANIZATION']\n",
      "\n",
      "Discovered entities: [(range(3, 5), 'PERSON', 0.18746685653645725), (range(6, 8), 'ORGANIZATION', 0.22672175876072712)]\n",
      "\n",
      "Amount of entities: 2\n",
      "    PERSON: Anną Kowalską\n",
      "    ORGANIZATION: Politechnice Warszawskiej\n"
     ]
    }
   ],
   "source": [
    "# Show possible tags\n",
    "print(\"Tags:\", ner.get_possible_ner_tags())\n",
    "\n",
    "# Test the model\n",
    "tokens = [\"Spotkałem\", \"się\", \"z\", \"Anną\", \"Kowalską\", \"w\", \"Politechnice\", \"Warszawskiej\", \".\"]\n",
    "entities = ner.extract_entities(tokens)\n",
    "\n",
    "# Print the results\n",
    "print(\"\\nDiscovered entities:\", entities)\n",
    "print(\"\\nAmount of entities:\", len(entities))\n",
    "for e in entities:\n",
    "    range = e[0]\n",
    "    tag = e[1]\n",
    "    entity_text = \" \".join(tokens[i] for i in range)\n",
    "    print(\"    \" + tag + \": \" + entity_text)\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "mitie-polish-oNJ6WEaN-py3.10",
   "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": "3.10.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}