evaluation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "\n",
    "import nltk\n",
    "from nltk.tokenize import word_tokenize\n",
    "from nltk.translate import bleu_score\n",
    "\n",
    "from euskalToken import EuskalToken\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "def avg_of_sentence_bleu_scores(groundtruth_filename, prediction_filename, language):\n",
    "    \n",
    "    list_of_scores = []\n",
    "    \n",
    "    if language == \"euskara\":\n",
    "        tokenize = EuskalToken().tokenize\n",
    "    elif language == \"english\":\n",
    "        tokenize = word_tokenize\n",
    "            \n",
    "    with open(groundtruth_filename, encoding=\"utf8\") as gtfile, open(prediction_filename, encoding=\"utf8\") as predfile:\n",
    "   \n",
    "        for i, (gt_sentence, predicted_sentence) in enumerate(zip(gtfile, predfile)):   \n",
    "        \n",
    "            gt_list_of_tokens = tokenize(gt_sentence)\n",
    "            predicted_list_of_tokens = predicted_sentence.split()\n",
    "\n",
    "            list_of_scores.append(bleu_score.sentence_bleu([gt_list_of_tokens], predicted_list_of_tokens))\n",
    "            \n",
    "        array_of_scores = np.asarray(list_of_scores)\n",
    "        avg_bleu_scores = array_of_scores.mean()\n",
    "          \n",
    "    return avg_bleu_scores\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Non smoothed BLEU score: 0.14%\n"
     ]
    }
   ],
   "source": [
    "groundtruth_filename = \"./evaldata/berriak.en\"\n",
    "prediction_filename = \"./predictions/predicted_berriak.unsup.en\"\n",
    "\n",
    "# Without smoothing\n",
    "score = avg_of_sentence_bleu_scores(groundtruth_filename, prediction_filename, \"english\")\n",
    "print(\"Non smoothed BLEU score: {:.2%}\".format(score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Non smoothed BLEU score: 0.48%\n"
     ]
    }
   ],
   "source": [
    "groundtruth_filename = \"./evaldata/berriak.en\"\n",
    "prediction_filename = \"./predictions/predicted_berriak.semisup.en\"\n",
    "\n",
    "# Without smoothing\n",
    "score = avg_of_sentence_bleu_scores(groundtruth_filename, prediction_filename, \"english\")\n",
    "print(\"Non smoothed BLEU score: {:.2%}\".format(score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Non smoothed BLEU score: 0.03%\n"
     ]
    }
   ],
   "source": [
    "groundtruth_filename = \"./evaldata/berriak.eu\"\n",
    "prediction_filename = \"./predictions/predicted_berriak.unsup.eu\"\n",
    "\n",
    "# Without smoothing\n",
    "score = avg_of_sentence_bleu_scores(groundtruth_filename, prediction_filename, \"euskara\")\n",
    "print(\"Non smoothed BLEU score: {:.2%}\".format(score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Non smoothed BLEU score: 0.17%\n"
     ]
    }
   ],
   "source": [
    "groundtruth_filename = \"./evaldata/berriak.eu\"\n",
    "prediction_filename = \"./predictions/predicted_berriak.semisup.eu\"\n",
    "\n",
    "# Without smoothing\n",
    "score = avg_of_sentence_bleu_scores(groundtruth_filename, prediction_filename, \"euskara\")\n",
    "print(\"Non smoothed BLEU score: {:.2%}\".format(score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 55333 embeddings with dimension 300\n",
      "55333\n"
     ]
    }
   ],
   "source": [
    "import io\n",
    "\n",
    "embeddings_handler = io.open(\"train_vocab.eu\", 'r', encoding='utf-8', newline='\\n', errors='ignore')\n",
    "\n",
    "n, d = map(int, embeddings_handler.readline().split())\n",
    "print(\"Found {} embeddings with dimension {}\".format(n, d))\n",
    "\n",
    "vocab = []\n",
    "\n",
    "for line in embeddings_handler:\n",
    "    tokens = line.rstrip().split(' ')\n",
    "    vocab.append(tokens[0])\n",
    "    \n",
    "print(len(vocab))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "def chunks(l, n):\n",
    "    for i in range(0, len(l), n):\n",
    "        yield l[i:i+n]\n",
    "        \n",
    "def compute_oov(lines, vocab):\n",
    "    calc_oov = []\n",
    "    \n",
    "    for line in lines:\n",
    "        tokens = line.rstrip().split()\n",
    "        num_oov = 0.\n",
    "        \n",
    "        for token in tokens:\n",
    "            if token not in vocab:\n",
    "                num_oov += 1\n",
    "                \n",
    "        calc_oov.append(num_oov/len(tokens))\n",
    "        \n",
    "    return calc_oov\n",
    "\n",
    "def filter_oov(lines, vocab):\n",
    "    filtered = []\n",
    "    \n",
    "    for line in lines:\n",
    "        tokens = line.rstrip().split()\n",
    "        num_oov = 0.\n",
    "        \n",
    "        for token in tokens:\n",
    "            if token not in vocab:\n",
    "                num_oov += 1\n",
    "                \n",
    "        if num_oov/len(tokens) <= 0.1:\n",
    "            filtered.append(line)\n",
    "        \n",
    "    return filtered"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processed 100000 sentences\n",
      "Processed 200000 sentences\n",
      "Processed 300000 sentences\n",
      "Processed 400000 sentences\n",
      "Processed 500000 sentences\n",
      "Processed 600000 sentences\n",
      "Processed 700000 sentences\n",
      "Processed 800000 sentences\n",
      "Processed 900000 sentences\n",
      "Processed 1000000 sentences\n",
      "Processed 1100000 sentences\n",
      "Processed 1200000 sentences\n",
      "Processed 1300000 sentences\n",
      "Processed 1400000 sentences\n",
      "Processed 1500000 sentences\n",
      "Processed 1600000 sentences\n",
      "Processed 1700000 sentences\n",
      "Processed 1800000 sentences\n",
      "Processed 1900000 sentences\n",
      "Processed 2000000 sentences\n",
      "Processed 2100000 sentences\n",
      "Processed 2200000 sentences\n",
      "Processed 2300000 sentences\n",
      "Processed 2400000 sentences\n"
     ]
    }
   ],
   "source": [
    "import math\n",
    "import concurrent.futures\n",
    "from itertools import repeat\n",
    "\n",
    "total = 0\n",
    "perc_oov = []\n",
    "\n",
    "# Sentences have already been tokenized\n",
    "with open(\"sentences.eu\", \"r\") as f:\n",
    "    sentences = f.readlines()\n",
    "    \n",
    "num_chunks = 1000\n",
    "chunked_sentences = chunks(sentences, num_chunks)\n",
    "vocab_repeated = repeat(vocab, math.ceil(len(sentences)/num_chunks))\n",
    "oov_stats = []    \n",
    "    \n",
    "with concurrent.futures.ProcessPoolExecutor() as executor:\n",
    "    for updated_sents in executor.map(compute_oov, chunked_sentences, vocab_repeated):\n",
    "        oov_stats.extend(updated_sents)\n",
    "        \n",
    "        if len(oov_stats) % 100000 == 0:\n",
    "            print(\"Processed {} sentences\".format(len(oov_stats)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.mlab as mlab\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "n, bins, patches = plt.hist(oov_stats, 10, facecolor='blue', alpha=0.5)\n",
    "plt.title(\"OOV percentage in Wikipedia Basque corpus\")\n",
    "plt.xlabel(\"OOV percentage\")\n",
    "plt.ylabel(\"# of sentences\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processed 2485520 sentences\n"
     ]
    }
   ],
   "source": [
    "# Sentences have already been tokenized\n",
    "with open(\"parallel_sents.eu.train\", \"r\") as f:\n",
    "    sentences = f.readlines()\n",
    "    \n",
    "num_chunks = 1000\n",
    "chunked_sentences = chunks(sentences, num_chunks)\n",
    "vocab_repeated = repeat(vocab, math.ceil(len(sentences)/num_chunks))\n",
    "oov_stats_paco = []    \n",
    "    \n",
    "with concurrent.futures.ProcessPoolExecutor() as executor:\n",
    "    for updated_sents in executor.map(compute_oov, chunked_sentences, vocab_repeated):\n",
    "        oov_stats_paco.extend(updated_sents)\n",
    "        \n",
    "        if len(oov_stats_paco) % 100000 == 0:\n",
    "            print(\"Processed {} sentences\".format(len(oov_stats)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax1 = plt.subplots()\n",
    "n, bins, patches = ax1.hist(oov_stats, 10, ls='dashed', alpha = 0.9, lw=3, color= 'g', label=\"Wikipedia\")\n",
    "ax2 = ax1.twinx()\n",
    "n, bins, patches = ax2.hist(oov_stats_berriak, 10, ls='dotted', alpha = 0.5, lw=3, color= 'r', label=\"Berriak\")\n",
    "ax1.set_ylabel('# of sentences')\n",
    "ax1.set_xlabel('<OOV> proportion')\n",
    "ax2.set_ylabel('# of sentences')\n",
    "ax2.tick_params('y')\n",
    "fig.tight_layout()\n",
    "ax1.legend(loc='upper right')\n",
    "ax2.legend(loc='lower right')\n",
    "ax1.set_title(\"<OOV> in Wikipedia and Berriak Basque\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(20,20))\n",
    "n, bins, patches = plt.hist(oov_stats, 10, facecolor='blue')\n",
    "n, bins, patches = plt.hist(oov_stats_paco, 10, facecolor='red')\n",
    "n, bins, patches = plt.hist(oov_stats_berriak, 10, facecolor='green')\n",
    "plt.title(\"OOV percentage in PACO Basque training corpus\")\n",
    "plt.xlabel(\"OOV percentage\")\n",
    "plt.ylabel(\"# of sentences\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Sentences have already been tokenized\n",
    "with open(\"evaldata/berriak.eu\", \"r\") as f:\n",
    "    sentences = f.readlines()\n",
    "    \n",
    "num_chunks = 1000\n",
    "chunked_sentences = chunks(sentences, num_chunks)\n",
    "vocab_repeated = repeat(vocab, math.ceil(len(sentences)/num_chunks))\n",
    "oov_stats_berriak = []    \n",
    "    \n",
    "with concurrent.futures.ProcessPoolExecutor() as executor:\n",
    "    for updated_sents in executor.map(compute_oov, chunked_sentences, vocab_repeated):\n",
    "        oov_stats_berriak.extend(updated_sents)\n",
    "        \n",
    "        if len(oov_stats_berriak) % 100000 == 0:\n",
    "            print(\"Processed {} sentences\".format(len(oov_stats_berriak)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fb91403e940>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "n, bins, patches = plt.hist(oov_stats, 10, facecolor='blue', alpha=0.5)\n",
    "plt.title(\"OOV percentage in Berriak Basque corpus\")\n",
    "plt.xlabel(\"OOV percentage\")\n",
    "plt.ylabel(\"# of sentences\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"all_sentences.eu\", \"r\") as f:\n",
    "    sentences = f.readlines()\n",
    "    \n",
    "num_chunks = 1000\n",
    "chunked_sentences = chunks(sentences, num_chunks)\n",
    "vocab_repeated = repeat(vocab, math.ceil(len(sentences)/num_chunks))\n",
    "result = []    \n",
    "    \n",
    "with concurrent.futures.ProcessPoolExecutor() as executor:\n",
    "    for updated_sents in executor.map(filter_oov, chunked_sentences, vocab_repeated):\n",
    "        result.extend(updated_sents)\n",
    "        \n",
    "        if len(result) % 100000 == 0:\n",
    "            print(\"Processed {} sentences\".format(len(result)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"clean_mono.eu\", 'w') as file:\n",
    "    file.writelines(\"%s\" % line for line in result)"
   ]
  }
 ],
 "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}