nbs/evaluation-gpt3.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "0f9ae7f9",
   "metadata": {},
   "source": [
    "# Evaluierung Lemmatisierung mit GPT-3"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3bcc9b77",
   "metadata": {},
   "source": [
    "## Inhalt\n",
    "1. [Analyse der Ausgabeformate](#Ausgabeformate)\n",
    "2. [Einlesen der Ausgaben](#Einlesen)\n",
    "3. [Ergebnisse und Analysen](#Ergebnisse)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "d46bfb81",
   "metadata": {},
   "outputs": [],
   "source": [
    "import glob\n",
    "import json\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.pyplot import figure\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "de1b04ba",
   "metadata": {},
   "source": [
    "## Ausgabeformate\n",
    "\n",
    "Ausgabeformate auf die Eingabeaufforderung `Lemmatisiere die Tokenliste: ['token1', 'token2', ...]` fallen sehr unterschiedlich aus. Die häufigsten wurden ausgezählt:\n",
    "\n",
    "* `list-notation`: `['lemma']`\n",
    "* `list-notation-without-quot`: `[lemma]`\n",
    "* `:-notation`: `token:lemma\\n`\n",
    "* `--notation`: `token-lemma\\n`\n",
    "* `comma-separated`: `lemma,lemma,lemma`\n",
    "* `whitespace-separated`: `lemma lemma lemma`\n",
    "* `+-separated`: `lemma+lemma+lemma`\n",
    "* `|-separated`: `lemma|lemma|lemma`\n",
    "* `_-separated`: `lemma_lemma_lemma`\n",
    "* `error`: alle übrigen, unter anderem Tuple-Notation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "947c309a",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"gpt3_outputs/formats.json\", \"r\") as fp:\n",
    "    formats = json.load(fp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "f6b9ee2a",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.DataFrame.from_dict(formats, orient='columns')\n",
    "df = df.set_index('dataset')\n",
    "df['error_num'] = df.error.str.len()  # number of errors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "2f672fb6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "list-notation                                                              4785\n",
       "list-notation-without-quot                                                  188\n",
       ":-notation                                                                    0\n",
       "--notation                                                                    0\n",
       "comma-separated                                                             272\n",
       "whitespace-separated                                                        675\n",
       "+-separated                                                                   1\n",
       "|-separated                                                                 275\n",
       "_-separated                                                                 545\n",
       "other                                                                        []\n",
       "error                         [[('(', '('), ('Bei', 'bei'), ('den', 'der'), ...\n",
       "error_num                                                                    56\n",
       "dtype: object"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e5920cae",
   "metadata": {},
   "source": [
    "Die meisten Ausgaben sind in Listennotation (4785), weitere sind durch Leerzeichen (675) oder Unterstriche (545) getrennt."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "71b4f726",
   "metadata": {},
   "source": [
    "## Einlesen"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "c3ab4364",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"results-gpt3.json\", \"r\") as fp:\n",
    "    results = json.load(fp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "977f2e87",
   "metadata": {},
   "outputs": [],
   "source": [
    "d = pd.DataFrame.from_dict(results, orient='columns')\n",
    "#d = d.set_index('dataset')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "35703920",
   "metadata": {},
   "outputs": [],
   "source": [
    "acc = []\n",
    "prec = []\n",
    "rec = []\n",
    "for entry in d.metrics:\n",
    "    acc.append(entry['overall']['accuracy'])\n",
    "    prec.append(entry['overall']['adj_precision'])\n",
    "    rec.append(entry['overall']['adj_recall'])\n",
    "d['accuracy'] = acc\n",
    "d['precision'] = prec\n",
    "d['recall'] = rec"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d88aade2",
   "metadata": {},
   "source": [
    "## Ergebnisse"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "608faa3e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='dataset'>"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "d.plot.bar(x='dataset', y='accuracy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "e89a8de9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "d2 = d[['dataset', 'precision', 'recall']]\n",
    "d2.set_index('dataset')\n",
    "d2.plot.bar()"
   ]
  }
 ],
 "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": "3.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}