data/Train.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Summarization Using Deep Learning Prototype\n",
    "\n",
    "This notebook is a prototype of trying to use deep learning for automatic extractive text summarization. To simplify the problem I have made the following assumptions. \n",
    "\n",
    "- The input article is 30 sentences or less.\n",
    "- A labeled dataset will be created by choosing up to the top 3 sentences that maximize the rouge score.\n",
    "- The training task is to then automatically chose which sentences should be included in the summary. \n",
    "- The input is a list of sentences in the article, and 1 sentence from the article.\n",
    "- The output is whether that sentence should be included in the summary or not. (1 or 0).\n",
    "- To classify the entire document we will have to loop over each sentence to ask the model if it should be included or not. \n",
    "\n",
    "The structure for this model gets lots of ideas from this paper:\n",
    "\n",
    "_SummaRuNNer: A Recurrent Neural Network based Sequence Model for\n",
    "Extractive Summarization of Documents_\n",
    "\n",
    "Ramesh Nallapati, Feifei Zhai, Bowen Zhou\n",
    "\n",
    "https://arxiv.org/pdf/1611.04230.pdf\n",
    "\n",
    "\n",
    "I have already made a clean dataset in the data cleaning notebook. So I can go ahead and load it. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# run the following command if you don't have the spacy model downloaded yet.\n",
    "# This must be done before you import spacy. https://spacy.io/models/en#en_core_web_sm\n",
    "# ! python -m spacy download en_core_web_sm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import tensorflow_hub as hub\n",
    "import tensorflow_datasets as tfds\n",
    "from tensorflow.keras import *\n",
    "import spacy\n",
    "import pandas as pd\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Params\n",
    "\n",
    "article_train_limit = None # How many articles we allow to be trained.\n",
    "max_article_length = 30\n",
    "batch = 128\n",
    "fine_tune_USE = False # Whether or not our we should allow additional training on the sentence embedder.\n",
    "checkpoint_dir = 'tfmodel.spacy.average_pooling.class_weights'\n",
    "\n",
    "data_dir = \"spacy_fmeasure-1\"\n",
    "train_data = os.path.join(data_dir,'train.labeled.parquet')\n",
    "validation_data = os.path.join(data_dir,'validation.labeled.parquet')\n",
    "test_data = os.path.join(data_dir,'test.labeled.parquet')\n",
    "\n",
    "# True means to balance data using number of samples\n",
    "# False means to balance it using class weights\n",
    "balance_data = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['article', 'summary', 'sentences', 'labels', 'rouge'], dtype='object')"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df=pd.read_parquet('test.labeled.parquet')\n",
    "df.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here is a description of the columms\n",
    "\n",
    "- **article** the original article.\n",
    "- **summary** the summary from the dataset. \n",
    "- **sentences** an array with the article split into individual sentences\n",
    "- **labels** array same length as `sents` identifying which sentences should be in the summary.\n",
    "- **rouge** The rouge1 score for the 3 labels=1 sentences against the summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# If the article is less than max_article_length, we need to pad and mask the empty sentences.\n",
    "def make_mask(sents):\n",
    "    length = len(sents)\n",
    "    return ([True] * length) +([False] * (max_article_length-length))\n",
    "\n",
    "# Just use empty strings to pad\n",
    "def pad_sents(sents):\n",
    "    length = len(sents)\n",
    "    if not isinstance(sents, list):\n",
    "        sents = sents.tolist()\n",
    "    return sents + [\"\"] * (max_article_length-length)\n",
    "\n",
    "def explode_examples(parquet_file_name, article_train_limit=None, balance_data=True):\n",
    "    df=pd.read_parquet(parquet_file_name)\n",
    "    df['sent_count'] = df.sentences.map(lambda x: len(x))\n",
    "    df = df[df.sent_count<=max_article_length].copy()\n",
    "\n",
    "    df = df.iloc[0:article_train_limit].copy()\n",
    "    df['mask'] = df.sentences.map(make_mask)\n",
    "\n",
    "    # Here we pair the label and individual sentence then explode these into individual training examples.\n",
    "    # This has the side effect of excluding the padded strings because the labels can be shorter than the\n",
    "    # sentences. e.g. list(zip(['a','b','c'], [1,2])) ==  [('a', 1), ('b', 2)]    (Note no 'c')\n",
    "    df['paired'] = df.apply(lambda x: list(zip(x.labels, x.sentences)), axis=1)\n",
    "    df = df.explode('paired')\n",
    "    df['label'] = df.paired.map(lambda x: x[0])\n",
    "    df['sent'] = df.paired.map(lambda x: x[1])\n",
    "    df['sentences'] = df.sentences.map(pad_sents)\n",
    "    df = df[['sentences', 'sent', 'label', 'mask']]\n",
    "    if balance_data:\n",
    "        positive = df[df.label==1].shape[0]\n",
    "        balanced = df[df.label==0].sample(frac=1).iloc[0:positive].copy()\n",
    "        df = balanced.append(df[df.label==1].copy())\n",
    "    df = df.sample(frac=1)\n",
    "    \n",
    "    x = {\"article_sentences\":tf.constant(df.sentences.values.tolist()), \n",
    "             \"sent\":tf.constant(df.sent.values.tolist()), \n",
    "             \"mask\": tf.constant(df['mask'].values.tolist())}\n",
    "    y = tf.convert_to_tensor(df.label.values, dtype=tf.int8)\n",
    "    return tf.data.Dataset.from_tensor_slices((x, y)).batch(batch, drop_remainder=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataset = explode_examples(train_data, article_train_limit=article_train_limit, balance_data=balance_data)\n",
    "validation_dataset = explode_examples(validation_data, balance_data=balance_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Porportion of positive examples 0.047206480999946204\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{0: 0.5247726711289393, 1: 10.591765990787787}"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df = pd.read_parquet(train_data)\n",
    "train_df['article_length'] = train_df.labels.map(len)\n",
    "train_df = train_df[train_df.article_length <= max_article_length]\n",
    "total_labels = train_df.article_length.sum()\n",
    "total_pos = train_df.labels.map(sum).sum()\n",
    "print(f'Porportion of positive examples {total_pos/total_labels}')\n",
    "pos_weight = total_labels / (2 * total_pos)\n",
    "neg_weight = total_labels / (2* (total_labels-total_pos))\n",
    "class_weights = {0: neg_weight, 1: pos_weight}\n",
    "if balance_data:\n",
    "    class_weights = {0:1, 1:1}\n",
    "class_weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<BatchDataset shapes: ({article_sentences: (128, 30), sent: (128,), mask: (128, 30)}, (128,)), types: ({article_sentences: tf.string, sent: tf.string, mask: tf.bool}, tf.int8)>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here is the description of the transformed data\n",
    "\n",
    "#### X\n",
    "- **article_sentences** The sentences padded to a fixed length, this could be duplicated for as many sentences are in the article.\n",
    "- **sent** The individual sentence for the training example\n",
    "- **mask** To keep track of which sentences are just paddin\n",
    "\n",
    "#### Y:\n",
    "- **label** Whether or not this example should be included in the summary"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Graph\n",
    "\n",
    "Below you can see the summary and image. \n",
    "For the forward pass we have the sentences that get embedded using Universal Sentence Encoder and then squashed into 1 vector using a GRU. \n",
    "Then the sentence to be classified is embedded using Universal Sentence encoder as well. \n",
    "These two vectors are concatenated and dense layers bring it down to 1 output. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# This can take a minute to load.\n",
    "# Loading this one multiple times sometimes causes errors. \n",
    "# If running the cells below multiple times and you get some weird un explainable error, restart\n",
    "# the kernel to see if it helps. \n",
    "\n",
    "#If you get the error SavedModel file does not exist at: /var/folders/cl....\n",
    "# Delete it ex: rm -r /var/folders/cl/63p3pg596jq18sjxn0zdb8k40000gn/T/tfhub_modules/063d866c06683311b44b4992fd46003be952409c\n",
    "\n",
    "use_embed_sent = hub.keras_layer.KerasLayer(\"https://tfhub.dev/google/universal-sentence-encoder/4\", name=\"sent_embedder\", trainable=fine_tune_USE)\n",
    "use_embed_article_sentences = hub.keras_layer.KerasLayer(\"https://tfhub.dev/google/universal-sentence-encoder/4\", name=\"article_sents_embedder\", trainable=fine_tune_USE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "article_sentences (InputLayer)  [(None, 30)]         0                                            \n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_article_sentences_1 [(None,)]            0           article_sentences[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "sent (InputLayer)               [(None,)]            0                                            \n",
      "__________________________________________________________________________________________________\n",
      "article_sents_embedder (KerasLa (None, 512)          256797824   tf_op_layer_article_sentences_1d[\n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_sent_reshaped (Tens [(None,)]            0           sent[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_article_sentences_r [(None, 30, 512)]    0           article_sents_embedder[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "mask (InputLayer)               [(None, 30)]         0                                            \n",
      "__________________________________________________________________________________________________\n",
      "sent_embedder (KerasLayer)      (None, 512)          256797824   tf_op_layer_sent_reshaped[0][0]  \n",
      "__________________________________________________________________________________________________\n",
      "bidirectional (Bidirectional)   (None, 1024)         3151872     tf_op_layer_article_sentences_res\n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_sent_embed_reshaped [(None, 512)]        0           sent_embedder[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_concated (TensorFlo [(None, 1536)]       0           bidirectional[0][0]              \n",
      "                                                                 tf_op_layer_sent_embed_reshaped[0\n",
      "__________________________________________________________________________________________________\n",
      "dense (Dense)                   (None, 128)          196736      tf_op_layer_concated[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "dense_1 (Dense)                 (None, 1)            129         dense[0][0]                      \n",
      "==================================================================================================\n",
      "Total params: 516,944,385\n",
      "Trainable params: 3,348,737\n",
      "Non-trainable params: 513,595,648\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "article_sentences = tf.keras.layers.Input(shape=[max_article_length],dtype=tf.string, name=\"article_sentences\", ragged=False)\n",
    "sent = tf.keras.layers.Input(shape=[], name='sent', dtype=tf.string)\n",
    "mask = tf.keras.layers.Input(shape=[max_article_length], dtype=tf.bool, name=\"mask\")\n",
    "sent_reshaped = tf.reshape(sent, [-1], name=\"sent_reshaped\")\n",
    "sent_embed = use_embed_sent(sent_reshaped)\n",
    "sent_embed_reshaped = tf.reshape(sent_embed, [-1,512], name=\"sent_embed_reshaped\")\n",
    "article_sentences_1d = tf.reshape(article_sentences, [-1], name=\"article_sentences_1d\")\n",
    "article_sentences_embed = use_embed_article_sentences(article_sentences_1d)\n",
    "article_sentences_restored = tf.reshape(article_sentences_embed, [-1, max_article_length, 512], name=\"article_sentences_restored\")\n",
    "article_sentences_gru = tf.keras.layers.Bidirectional(tf.keras.layers.GRU(512,name=\"gru\", dropout=0.2))(article_sentences_restored, mask=mask)\n",
    "concatenated = tf.concat([article_sentences_gru, sent_embed_reshaped], axis=1, name=\"concated\")\n",
    "d1 = layers.Dense(128, activation='relu')(concatenated)\n",
    "d2 = layers.Dense(1, activation='sigmoid')(d1)\n",
    "model = Model(inputs=[article_sentences, sent, mask], outputs=[d2])\n",
    "model.compile(loss=losses.BinaryCrossentropy(), \n",
    "              metrics=[tf.keras.metrics.BinaryAccuracy(), tf.keras.metrics.Recall()],\n",
    "              optimizer=tf.keras.optimizers.RMSprop(learning_rate=0.001))\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf.keras.utils.plot_model(model, show_shapes=True, show_layer_names=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "checkpoint_filepath = checkpoint_dir\n",
    "model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
    "    filepath=checkpoint_filepath,\n",
    "    monitor='val_binary_accuracy',\n",
    "    mode='max',\n",
    "    save_best_only=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "17425/17425 [==============================] - ETA: 0s - loss: 0.5479 - binary_accuracy: 0.6497 - recall: 0.7792WARNING:tensorflow:From /home/jzeimen/anaconda3/envs/p36/lib/python3.6/site-packages/tensorflow/python/ops/resource_variable_ops.py:1817: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "If using Keras pass *_constraint arguments to layers.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/jzeimen/anaconda3/envs/p36/lib/python3.6/site-packages/tensorflow/python/ops/resource_variable_ops.py:1817: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "If using Keras pass *_constraint arguments to layers.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: tfmodel.spacy.average_pooling.class_weights/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: tfmodel.spacy.average_pooling.class_weights/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "17425/17425 [==============================] - 5046s 290ms/step - loss: 0.5479 - binary_accuracy: 0.6497 - recall: 0.7792 - val_loss: 0.5946 - val_binary_accuracy: 0.6307 - val_recall: 0.8229\n",
      "Epoch 2/5\n",
      "17425/17425 [==============================] - 4995s 287ms/step - loss: 0.5533 - binary_accuracy: 0.6630 - recall: 0.7741 - val_loss: 0.6596 - val_binary_accuracy: 0.6177 - val_recall: 0.8220\n",
      "Epoch 3/5\n",
      "17425/17425 [==============================] - ETA: 0s - loss: 0.5962 - binary_accuracy: 0.6637 - recall: 0.7443INFO:tensorflow:Assets written to: tfmodel.spacy.average_pooling.class_weights/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: tfmodel.spacy.average_pooling.class_weights/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "17425/17425 [==============================] - 5019s 288ms/step - loss: 0.5962 - binary_accuracy: 0.6637 - recall: 0.7443 - val_loss: 0.6109 - val_binary_accuracy: 0.6714 - val_recall: 0.7314\n",
      "Epoch 4/5\n",
      "17425/17425 [==============================] - 4995s 287ms/step - loss: 0.6333 - binary_accuracy: 0.6589 - recall: 0.7324 - val_loss: 0.6297 - val_binary_accuracy: 0.6531 - val_recall: 0.7537\n",
      "Epoch 5/5\n",
      "17425/17425 [==============================] - 4993s 287ms/step - loss: 0.6634 - binary_accuracy: 0.6545 - recall: 0.7253 - val_loss: 0.6998 - val_binary_accuracy: 0.6256 - val_recall: 0.7553\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_dataset, \n",
    "                    validation_data=validation_dataset,\n",
    "                    epochs=5,\n",
    "                    callbacks=[model_checkpoint_callback],\n",
    "                    class_weight=class_weights)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'loss': [0.5478658676147461,\n",
       "  0.5533235669136047,\n",
       "  0.5962039828300476,\n",
       "  0.6333487629890442,\n",
       "  0.6633638143539429],\n",
       " 'binary_accuracy': [0.6496717929840088,\n",
       "  0.6630277037620544,\n",
       "  0.663658082485199,\n",
       "  0.6589078307151794,\n",
       "  0.6545377373695374],\n",
       " 'recall': [0.7791612148284912,\n",
       "  0.774080216884613,\n",
       "  0.7442969083786011,\n",
       "  0.7323778867721558,\n",
       "  0.7252739667892456],\n",
       " 'val_loss': [0.5946314930915833,\n",
       "  0.6596103310585022,\n",
       "  0.6108503341674805,\n",
       "  0.6296756267547607,\n",
       "  0.6997542977333069],\n",
       " 'val_binary_accuracy': [0.630679726600647,\n",
       "  0.6177093386650085,\n",
       "  0.6713985204696655,\n",
       "  0.6531038284301758,\n",
       "  0.6255671977996826],\n",
       " 'val_recall': [0.8229360580444336,\n",
       "  0.8220049738883972,\n",
       "  0.7313780188560486,\n",
       "  0.7537243962287903,\n",
       "  0.7552762031555176]}"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history.history"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'tfmodel.spacy.average_pooling.class_weights'"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model=None\n",
    "tf.keras.backend.clear_session()\n",
    "checkpoint_dir"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "saved_model = tf.keras.models.load_model(checkpoint_dir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# See info here https://spacy.io/models/en#en_core_web_sm\n",
    "from rouge_score import rouge_scorer\n",
    "scorer = rouge_scorer.RougeScorer(['rouge1'], use_stemmer=True)\n",
    "nlp = spacy.load('en_core_web_sm')\n",
    "\n",
    "def predict(sentences, model):\n",
    "    sentences = sentences[:max_article_length]\n",
    "    original_length = len(sentences)\n",
    "    sentences = pad_sents(sentences)\n",
    "    mask = make_mask(sentences)\n",
    "    scores = model({\"article_sentences\":tf.constant([sentences]*max_article_length), \n",
    "             \"sent\":tf.constant(sentences), \n",
    "             \"mask\": tf.constant([mask]*max_article_length)})\n",
    "    scores = scores.numpy().squeeze()\n",
    "    return scores, sentences\n",
    "\n",
    "def predict_top_sents(sentences, model, n=1):\n",
    "    scores, sentences = predict(sentences, model)\n",
    "    df = pd.DataFrame({'score': scores, 'sentence': sentences})\n",
    "    return \" \".join(df.nlargest(n,'score').sort_index()['sentence'].values.tolist())\n",
    "\n",
    "def calculate_rouge(sentences, model, summary, scorer):\n",
    "    gen_summary = predict_top_sents(sentences, model)\n",
    "    score = scorer.score(summary, gen_summary)\n",
    "    return score['rouge1'].fmeasure\n",
    "\n",
    "def calculate_naive_rouge(sentences, summary, scorer):\n",
    "    score = scorer.score(summary, \" \".join(sentences[:1]))\n",
    "    return score['rouge1'].fmeasure\n",
    "\n",
    "def naive_rouge_pd(row):\n",
    "    row['naive_rouge'] = calculate_naive_rouge(row['sentences'], row['summary'], scorer)\n",
    "    return row\n",
    "\n",
    "def rouge_pd(row):\n",
    "    row['rouge'] = calculate_rouge(row['sentences'], saved_model, row['summary'], scorer)\n",
    "    return row"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_dataset = explode_examples(test_data, None, False)\n",
    "saved_model.evaluate(test_dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_df = pd.read_parquet(test_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/jzeimen/anaconda3/envs/p3/lib/python3.7/site-packages/tqdm/std.py:668: FutureWarning: The Panel class is removed from pandas. Accessing it from the top-level namespace will also be removed in the next version\n",
      "  from pandas import Panel\n"
     ]
    }
   ],
   "source": [
    "from tqdm import tqdm\n",
    "tqdm.pandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 11490/11490 [00:21<00:00, 539.18it/s]\n"
     ]
    }
   ],
   "source": [
    "df = test_df.progress_apply(naive_rouge_pd, axis=1)\n",
    "df = df.progress_apply(rouge_pd, axis=1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>rouge_fmeasure</th>\n",
       "      <th>rouge</th>\n",
       "      <th>naive_rouge</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>11490.000000</td>\n",
       "      <td>11490.000000</td>\n",
       "      <td>11490.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.434478</td>\n",
       "      <td>0.284209</td>\n",
       "      <td>0.267632</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.091923</td>\n",
       "      <td>0.124016</td>\n",
       "      <td>0.126278</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.094203</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.370968</td>\n",
       "      <td>0.192308</td>\n",
       "      <td>0.179559</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.426966</td>\n",
       "      <td>0.268657</td>\n",
       "      <td>0.263158</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.487805</td>\n",
       "      <td>0.363636</td>\n",
       "      <td>0.350000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.840580</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       rouge_fmeasure         rouge   naive_rouge\n",
       "count    11490.000000  11490.000000  11490.000000\n",
       "mean         0.434478      0.284209      0.267632\n",
       "std          0.091923      0.124016      0.126278\n",
       "min          0.094203      0.000000      0.000000\n",
       "25%          0.370968      0.192308      0.179559\n",
       "50%          0.426966      0.268657      0.263158\n",
       "75%          0.487805      0.363636      0.350000\n",
       "max          1.000000      1.000000      0.840580"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Rouge_fmeasure is from the best in a greedy search\n",
    "# rouge is from our precision\n",
    "# naive_rouge is when we just use the first 3 sentences\n",
    "df[['rouge_fmeasure', 'rouge', 'naive_rouge']].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['This is the moment that a crew of firefighters struggled to haul a giant  pig out of a garden swimming pool.',\n",
       " 'The prize porker, known  as Pigwig, had fallen into the pool in an upmarket neighbourhood in Ringwood, Hampshire.',\n",
       " 'His owners had been taking him for a walk around the garden when the animal plunged into the water and was unable to get out.',\n",
       " 'A team from Dorset Fire and Rescue struggled to haul the huge black pig out of swimming pool water .',\n",
       " 'The prize porker known as Pigwig had fallen into the water and had then been unable to get out again .',\n",
       " 'Two fire crews and a specialist animal rescue team had to use slide boards and strops to haul the huge black pig from the small pool.',\n",
       " \"A spokesman for Dorset Fire and Rescue Service said: 'At 4.50pm yesterday the service received a call to a pig stuck in a swimming pool. '\",\n",
       " \"One crew of firefighters from Ferndown and a specialist animal rescue unit from Poole were mobilised to this incident. '\",\n",
       " 'Once in attendance the crew secured the pig with strops, and requested the attendance of another appliance which was mobilised from Ringwood by our colleagues in Hampshire Fire and Rescue Service.',\n",
       " 'Firefighters were also called out to a horse which had fallen into a swimming pool in Heyshott, West Sussex .',\n",
       " \"The exhausted animal had to be winched to using an all-terrain crane but appeared no worse for wear after its tumble . '\",\n",
       " \"The crew rescued the pig from the swimming pool using specialist animal rescue slide boards, strops and lines to haul the pig from the swimming pool.'\",\n",
       " \"But Pigwig wasn't the only animal who needed rescuing after taking an unexpected swim .\",\n",
       " 'Crews in West Sussex were called out to a swimming pool where this time a horse had fallen in.',\n",
       " 'Wet and very bedraggled, the exhausted animal put up no opposition when firefighters arrived to hoist her out of the small garden pool in\\xa0Heyshott.',\n",
       " 'The two-hour rescue operation ended with the wayward horse being fitted with straps under her belly and lifted up into the air with an all-terrain crane before being swung around and deposited back on dry land.',\n",
       " 'A fire brigade spokesman said that she appeared none the worse for her impromptu swim after stepping over the edge of the domestic pool.']"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "row=df.iloc[5]\n",
    "article = row.sentences.tolist()\n",
    "article"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For this article the following is our predicted best sentence to summarize the article."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"The crew rescued the pig from the swimming pool using specialist animal rescue slide boards, strops and lines to haul the pig from the swimming pool.'\""
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predict_top_sents(article, saved_model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The actual summary:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Giant pig fell into the swimming pool at his home in Ringwood, Hampshire .\n",
      "It took the efforts of a team of firefighters to winch him out of the water .\n",
      "A wayward horse also had to be rescued from a swimming pool in Sussex .\n"
     ]
    }
   ],
   "source": [
    "print(row.summary)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And this is what best matched the given summary. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Firefighters were also called out to a horse which had fallen into a swimming pool in Heyshott, West Sussex .'"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from itertools import compress\n",
    "best_possible_match = \" \".join(list(compress(row.sentences.tolist(), row.labels.tolist())))\n",
    "best_possible_match"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is how our predicted summary scored compared with the actual summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'rouge1': Score(precision=0.4230769230769231, recall=0.2558139534883721, fmeasure=0.3188405797101449)}"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scorer.score(row.summary, predict_top_sents(article, saved_model))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And this is with the best sentence possible. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'rouge1': Score(precision=0.6842105263157895, recall=0.3023255813953488, fmeasure=0.41935483870967744)}"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scorer.score(row.summary, best_possible_match)"
   ]
  }
 ],
 "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": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}