File size: 12,708 Bytes
8210aca |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 |
{
"cells": [
{
"cell_type": "code",
"execution_count": null,
"id": "1fc75ebf",
"metadata": {},
"outputs": [],
"source": [
"## datasets==2.0.0 pandas==1.4.2"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "c9cc126c",
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"import numpy as np\n",
"import pandas as pd\n",
"import re\n",
"from tqdm import tqdm\n",
"from datasets import Dataset, DatasetDict\n",
"import pickle\n",
"import json"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "2adeaf52",
"metadata": {},
"outputs": [],
"source": [
"def get_list_values(text):\n",
" return text.split()\n",
"\n",
"def replc_t_n(text):\n",
" return re.sub(\"\\t|\\n\", \" \", text).strip()\n",
"\n",
"def read_file(filepath, readlines=False):\n",
" with open(filepath, \"r\") as f:\n",
" if readlines:\n",
" txt = f.readlines()\n",
" else:\n",
" txt = f.read()\n",
" return txt"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "26a51547",
"metadata": {},
"outputs": [],
"source": [
"def split_text_on_labeled_tokens(text, labels):\n",
" \"\"\"\n",
" Split text on labeled token\n",
"\n",
" :param text: input text\n",
" :type text: string\n",
" :param labels: token labels with position in text \n",
" :type labels: list\n",
" :return: list of splited text on tokens, list of entity label for each token\n",
" :rtype: list, list\n",
" \"\"\"\n",
" ### inner function\n",
" def chunk_text_labeling(text, start, end, is_ner = False):\n",
" \"\"\"\n",
" Labeling part of text by text position\n",
"\n",
" :param text: input text\n",
" :type text: string\n",
" :param start: start position of entity in text \n",
" :type start: int\n",
" :param end: end position of entity in text \n",
" :type end: int\n",
" :param is_ner: part of text is named entity or not \n",
" :type is_ner: bool\n",
" \"\"\"\n",
" chunk_iter = 0\n",
" ner_chunk = text[start: end].split()\n",
" for part_of_chunk in ner_chunk:\n",
" split_text.append(part_of_chunk)\n",
" if is_ner:\n",
" if chunk_iter == 0:\n",
" ner_label.append(\"B-\"+ner)\n",
" else:\n",
" ner_label.append(\"I-\"+ner)\n",
" chunk_iter += 1\n",
" else:\n",
" ner_label.append(\"O\") \n",
" ### inner function\n",
" \n",
" init_start = 0\n",
" split_text = []\n",
" ner_label = []\n",
" for ner, start, end in labels:\n",
"\n",
" if start > init_start:\n",
"\n",
" chunk_text_labeling(text, init_start, start) \n",
" chunk_text_labeling(text, start, end, True)\n",
" init_start = end\n",
" else:\n",
" chunk_text_labeling(text, start, end, True)\n",
" init_start = end\n",
" \n",
" return split_text, ner_label"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "0ba5da7e",
"metadata": {},
"outputs": [],
"source": [
"def grouped_and_sort_labeled_data(annotation_file):\n",
" \"\"\"\n",
" Get list of entities with corresponding position in text\n",
"\n",
" :param annotation_file: List of entities\n",
" :type annotation_file: list\n",
" :return: list entitiens sorted by start position in text\n",
" :rtype: list\n",
" \"\"\"\n",
" df_ann = pd.DataFrame([get_list_values(replc_t_n(i)) for i in annotation_file if \";\" not in i]) \n",
" df_ann[2] = df_ann[2].astype(\"int\")\n",
" df_ann[3] = df_ann[3].astype(\"int\")\n",
" grouped = df_ann.groupby([1, 2])[3].min().reset_index()\n",
" \n",
" return grouped.sort_values(by=2)[[1,2,3]].values"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "46fdb74b",
"metadata": {},
"outputs": [],
"source": [
"def check_isalnum(text):\n",
" return any(i.isalnum() for i in text)\n",
"\n",
"def keep_only_alnum(text):\n",
" return \"\".join([i if i.isalnum() else \" \" for i in text]).strip()\n",
"\n",
"def drop_punct(seq, labels):\n",
" \"\"\"\n",
" Drop punctuation from labeled data\n",
"\n",
" :param seq: List of tokens\n",
" :type seq: list\n",
" :param labels: List of entities\n",
" :type labels: list\n",
" \"\"\"\n",
" new_seq = []\n",
" new_labels = []\n",
" for i in range(len(seq)):\n",
" if seq[i].isalnum():\n",
" new_seq.append(seq[i])\n",
" new_labels.append(labels[i]) \n",
" return new_seq, new_labels\n",
"\n",
"def drop_duplicate_tokens(seq, labels):\n",
" new_seq = []\n",
" new_labels = []\n",
" for i in range(len(seq)):\n",
" if (i != 0) & (seq[i-1] == seq[i]):\n",
" continue\n",
" else:\n",
" new_seq.append(seq[i])\n",
" new_labels.append(labels[i])\n",
" return new_seq, new_labels\n",
"\n",
"def prepare_sequences(seqs, labels):\n",
" clear_tokens = [keep_only_alnum(i) if check_isalnum(i) else i for i in seqs]\n",
" d_p_tokens, d_p_labels = drop_punct(clear_tokens, labels)\n",
" return drop_duplicate_tokens(d_p_tokens, d_p_labels)\n",
" \n",
"\n",
"def map_label_to_id(ids_dict, labels):\n",
" \"\"\"\n",
" Convert string label to corresponding id\n",
"\n",
" :param ids_dict: {\"age\": 0, \"event\": 1.....}\n",
" :type ids_dict: dict\n",
" :param labels: List of entities [\"age\", \"event\", \"O\"....]\n",
" :type labels: list\n",
" \"\"\"\n",
" return [ids_dict[i] for i in labels]"
]
},
{
"cell_type": "markdown",
"id": "5b735210",
"metadata": {},
"source": [
"### Preparing files in folders"
]
},
{
"cell_type": "markdown",
"id": "23ad44bb",
"metadata": {},
"source": [
"#### The data have been taken from https://github.com/dialogue-evaluation/RuNNE"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "1c2748fa",
"metadata": {},
"outputs": [],
"source": [
"folders = [\"train\", \"test\", \"dev\"]"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "ab65dc18",
"metadata": {},
"outputs": [],
"source": [
"for folder in folders:\n",
" base_path = f\"RuNNE/data/{folder}\"\n",
" temp_folder = os.listdir(base_path)\n",
" \n",
" ## getting list filenames of annotation\n",
" files_with_ann = [i for i in temp_folder if \".ann\" in i]\n",
"\n",
" all_sequences = []\n",
" all_labels = []\n",
" \n",
" for f_ann in tqdm(files_with_ann):\n",
" \n",
" ## getting filename for text by replaced of extension\n",
" txt_file = f_ann.replace(\".ann\", \".txt\")\n",
"\n",
" ann = read_file(base_path +\"/\"+ f_ann, readlines=True)\n",
" txt = read_file(base_path +\"/\"+ txt_file)\n",
" \n",
" ## check len, because in dev folder there are empty files\n",
" if len(ann) == 0:\n",
" continue\n",
" labels = grouped_and_sort_labeled_data(ann)\n",
" \n",
" ## splitting text on tokens and labeling each of them\n",
" split_text, ner_label = split_text_on_labeled_tokens(txt, labels)\n",
" seq_split_indexes = [i for i, v in enumerate(split_text) if v == \".\"]\n",
" \n",
" ## adding prepared data from each file to general list\n",
" prev = 0\n",
" for i in seq_split_indexes:\n",
" \n",
" short_text = split_text[prev: i]\n",
" short_label = ner_label[prev: i]\n",
" \n",
" clear_tokens, clear_label = prepare_sequences(short_text, short_label)\n",
" \n",
" all_sequences.append(clear_tokens)\n",
" all_labels.append(clear_label)\n",
" ## we don't take into account the dots in text \n",
" prev = i+1\n",
" \n",
" ## save data to file for each part of splitted dataset\n",
" df_folder = pd.DataFrame({\"sequences\": all_sequences, \"labels\": all_labels})\n",
" with open(f'{folder}_data.pickle', 'wb') as f:\n",
" pickle.dump(df_folder, f)\n",
" print(f\"For folder <{folder}> prepared <{df_folder.shape[0]}> sequences\")"
]
},
{
"cell_type": "markdown",
"id": "cc61030b",
"metadata": {},
"source": [
"### Creating DatasetDict fro prepared data"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "8c24ab41",
"metadata": {},
"outputs": [],
"source": [
"## load 3 dataframe and init them into transformer dataset\n",
"dsd = DatasetDict()\n",
"for folder in folders:\n",
" with open(f'{folder}_data.pickle', 'rb') as f:\n",
" data = pickle.load(f)\n",
" dsd[folder] = Dataset.from_pandas(data)"
]
},
{
"cell_type": "markdown",
"id": "04e76e90",
"metadata": {},
"source": [
"### Creating dictionary for labels ids "
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "ce021634",
"metadata": {},
"outputs": [],
"source": [
"## get unique entyties\n",
"for_df = []\n",
"for folder in folders:\n",
" with open(f'{folder}_data.pickle', 'rb') as f:\n",
" for_df.append(pickle.load(f))\n",
"lbls = pd.concat(for_df)[\"labels\"].values\n",
"\n",
"dd = dict()\n",
"ids = 0\n",
"for ll in lbls:\n",
" for lbl in ll:\n",
" if lbl not in dd:\n",
" dd[lbl] = ids\n",
" ids += 1\n",
"\n",
" \n",
"# # count each entity\n",
"# countss = dict()\n",
"# for ll in lbls:\n",
"# for lbl in ll:\n",
"# if lbl not in countss:\n",
"# countss[lbl] = 1\n",
"# else:\n",
"# countss[lbl] += 1\n",
"\n",
"# del countss[\"O\"]\n",
"# sorted_counts = {k: v for k, v in sorted(countss.items(), key=lambda item: item[0].split(\"-\")[1])}\n",
"\n",
"# for k, v in sorted_counts.items():\n",
"# print(\"- \"+k+f\": {v}\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "58000df7",
"metadata": {},
"outputs": [],
"source": [
"## sort mapper\n",
"\n",
"ll = [i for i in dd.keys() if i != \"O\"] \n",
"ll_sort = (sorted(ll, key=lambda x: x.split(\"-\")[1]))\n",
"new_dd = {k: v for v, k in enumerate([\"O\"] + ll_sort)}\n",
" \n",
" \n",
"reverse_dd = {v: k for k, v in new_dd.items()}\n",
"with open('id_to_label_map.pickle', 'wb') as f:\n",
" pickle.dump(reverse_dd, f)"
]
},
{
"cell_type": "markdown",
"id": "b30a7098",
"metadata": {},
"source": [
"### Creating new column with numerical labels"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "51fd6b38",
"metadata": {},
"outputs": [],
"source": [
"dsd_with_ids = dsd.map(\n",
" lambda x: {\"ids\": [map_label_to_id(new_dd, i) for i in x[\"labels\"]]}, batched=True, remove_columns = \"labels\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "b7ecf94f",
"metadata": {},
"outputs": [],
"source": [
"dsd_with_ids.push_to_hub(\"\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "5eb5f3fa",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "hf_env",
"language": "python",
"name": "hf_env"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.8.10"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
|