notebook: add demos for CoNLL-like exporter and Flair dataset loader

Export-To-CoNLL.ipynb

@@ -0,0 +1,76 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "7b378ac1-1035-4f5d-9c16-68c14178389b",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from pathlib import Path\n",
+    "\n",
+    "from ftfy import fix_encoding"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "1a8017c7-f7a1-4b17-804b-c4d404511969",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Download and unzip archive from: https://www.dfki.uni-kl.de/cybermapping/data/CO-Fun-1.0-anonymized.zip\n",
+    "base_path = Path(\"./prepared-data-and-code/NER/CRF/\")\n",
+    "\n",
+    "splits = {\n",
+    "    \"train\": base_path / \"train_set.txt\",\n",
+    "    \"dev\": base_path / \"dev_set.txt\",\n",
+    "    \"test\": base_path / \"test_set.txt\",\n",
+    "}"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "ab1deaef-adf3-4c4e-8da2-0bcc833615b3",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "for split_name, split_file in splits.items():\n",
+    "    with open(split_file, \"rt\") as f_p:\n",
+    "        buffer = f_p.readlines()\n",
+    "    buffer = \"[\" + \"\".join(buffer) + \"]\"\n",
+    "    data = eval(buffer)\n",
+    "\n",
+    "    with open(f\"{split_name}.tsv\", \"wt\") as f_out:\n",
+    "        for sentence in data:\n",
+    "            for line in sentence:\n",
+    "                token, pos, ner = line\n",
+    "                token = fix_encoding(token)\n",
+    "                f_out.write(f\"{token}\\t{pos}\\t{ner}\\n\")\n",
+    "            f_out.write(\"\\n\")"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "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": "3.11.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

FlairDatasetTest.ipynb

@@ -0,0 +1,152 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "87270bc7-2da0-490e-be8a-8cb06f38507b",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from flair.datasets.sequence_labeling import ColumnCorpus\n",
+    "\n",
+    "from pathlib import Path"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "08ebe948-4179-4a75-8c6b-0e2fcad293d7",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "columns = {0: \"text\", 2: \"ner\"}"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "af10c49b-002a-41f4-afde-172ac07251c0",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "2024-03-25 12:57:38,583 Reading data from .\n",
+      "2024-03-25 12:57:38,585 Train: train.tsv\n",
+      "2024-03-25 12:57:38,586 Dev: dev.tsv\n",
+      "2024-03-25 12:57:38,586 Test: test.tsv\n"
+     ]
+    }
+   ],
+   "source": [
+    "corpus = ColumnCorpus(\".\",\n",
+    "                      column_format=columns,\n",
+    "                      train_file=\"train.tsv\",\n",
+    "                      dev_file=\"dev.tsv\",\n",
+    "                      test_file=\"test.tsv\",\n",
+    "                      comment_symbol=None)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "efd08709-9308-4244-a730-0d51b738520c",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "'Corpus: 758 train + 94 dev + 96 test sentences'"
+      ]
+     },
+     "execution_count": 4,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "str(corpus)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "6d857c95-ae9f-4aab-803a-0a7e03ff632d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "base_path = Path(\"./prepared-data-and-code/\")\n",
+    "\n",
+    "indices_for_split = {\n",
+    "    \"train\": base_path / \"train_indices.txt\",\n",
+    "    \"dev\": base_path / \"valid_indices.txt\",\n",
+    "    \"test\": base_path / \"test_indices.txt\",\n",
+    "}"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "2529666f-793f-4d8c-a7aa-9b7d83975310",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def get_number_of_total_indices(filename: str) -> int:\n",
+    "    indices_counter = 0\n",
+    "    with open(filename, \"rt\") as f_p:\n",
+    "        for line in f_p:\n",
+    "            line = line.strip()\n",
+    "            if not line:\n",
+    "                continue\n",
+    "            indices_counter += 1\n",
+    "    return indices_counter"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "49a69247-ef66-445b-b9e8-adcaac282e14",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "train_indices = get_number_of_total_indices(indices_for_split[\"train\"])\n",
+    "dev_indices = get_number_of_total_indices(indices_for_split[\"dev\"])\n",
+    "test_indices = get_number_of_total_indices(indices_for_split[\"test\"])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "id": "fb22489b-07d2-4403-a30c-6d92f00d252e",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "assert len(corpus.train) == train_indices\n",
+    "assert len(corpus.dev) == dev_indices\n",
+    "assert len(corpus.test) == test_indices"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "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": "3.11.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}