pos_tagger_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "data = pd.read_csv(\"data_last.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "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>Unnamed: 0</th>\n",
       "      <th>doc_no</th>\n",
       "      <th>sentence_no</th>\n",
       "      <th>word</th>\n",
       "      <th>tag</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>2800</td>\n",
       "      <td>0</td>\n",
       "      <td>İster</td>\n",
       "      <td>CONJ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>2800</td>\n",
       "      <td>0</td>\n",
       "      <td>tango</td>\n",
       "      <td>NOUN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2800</td>\n",
       "      <td>0</td>\n",
       "      <td>yap</td>\n",
       "      <td>VERB</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>2800</td>\n",
       "      <td>0</td>\n",
       "      <td>ister</td>\n",
       "      <td>CONJ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>2800</td>\n",
       "      <td>0</td>\n",
       "      <td>göbek</td>\n",
       "      <td>NOUN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0  doc_no  sentence_no   word   tag\n",
       "0           0    2800            0  İster  CONJ\n",
       "1           1    2800            0  tango  NOUN\n",
       "2           2    2800            0    yap  VERB\n",
       "3           3    2800            0  ister  CONJ\n",
       "4           4    2800            0  göbek  NOUN"
      ]
     },
     "execution_count": 159,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {},
   "outputs": [],
   "source": [
    "data=data.drop(\"Unnamed: 0\", axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "517"
      ]
     },
     "execution_count": 161,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "words = list(set(data[\"word\"].values))\n",
    "n_words = len(words); n_words"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "11"
      ]
     },
     "execution_count": 162,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tags = list(set(data[\"tag\"].values))\n",
    "n_tags = len(tags); n_tags"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {},
   "outputs": [],
   "source": [
    "class SentenceGetter(object):\n",
    "    \n",
    "    def __init__(self, data):\n",
    "        self.n_sent = 1\n",
    "        self.data = data\n",
    "        self.empty = False\n",
    "        agg_func = lambda s: [(w, t) for w, t in zip(s[\"word\"].values.tolist(),\n",
    "                                                           s[\"tag\"].values.tolist())]\n",
    "        self.grouped = self.data.groupby(\"sentence_no\").apply(agg_func)\n",
    "        self.sentences = [s for s in self.grouped]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "metadata": {},
   "outputs": [],
   "source": [
    "getter = SentenceGetter(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[[('İster', 'CONJ'),\n",
       "  ('tango', 'NOUN'),\n",
       "  ('yap', 'VERB'),\n",
       "  ('ister', 'CONJ'),\n",
       "  ('göbek', 'NOUN'),\n",
       "  ('at!', 'VERB'),\n",
       "  ('Dünya', 'NOUN'),\n",
       "  ('Sağlık', 'NOUN'),\n",
       "  (\"Örgütü'nün\", 'NOUN'),\n",
       "  ('rakamlarına', 'NOUN'),\n",
       "  ('göre', 'ADV'),\n",
       "  (\"2000'de\", 'NUM'),\n",
       "  ('66', 'NUM'),\n",
       "  ('olan', 'ADJ'),\n",
       "  ('ortalama', 'ADJ'),\n",
       "  ('ömür', 'NOUN'),\n",
       "  (\"2009'da\", 'NUM'),\n",
       "  (\"68'e\", 'NUM'),\n",
       "  ('yükseldi.', 'VERB'),\n",
       "  ('Akıl', 'NOUN'),\n",
       "  ('hastaneleri', 'NOUN'),\n",
       "  ('tarih', 'NOUN'),\n",
       "  ('olacak.', 'VERB'),\n",
       "  ('Akciğer', 'NOUN'),\n",
       "  ('tomografisi', 'NOUN'),\n",
       "  ('hayat', 'NOUN'),\n",
       "  ('kurtarıyor.', 'VERB')],\n",
       " [('Ama', 'CONJ'),\n",
       "  ('iskeletimiz', 'NOUN'),\n",
       "  ('uzun', 'ADJ'),\n",
       "  ('yaşam', 'NOUN'),\n",
       "  ('süresine', 'NOUN'),\n",
       "  ('hazır', 'ADJ'),\n",
       "  ('değil.', 'VERB'),\n",
       "  ('Sağlık', 'NOUN'),\n",
       "  ('Bakanı', 'NOUN'),\n",
       "  ('Recep', 'NOUN'),\n",
       "  ('Akdağ,', 'NOUN'),\n",
       "  ('yeni', 'ADJ'),\n",
       "  ('yılla', 'NOUN'),\n",
       "  ('birlikte', 'ADP'),\n",
       "  ('ağır', 'ADJ'),\n",
       "  ('ruh', 'NOUN'),\n",
       "  ('sağlığı', 'NOUN'),\n",
       "  ('problemi', 'NOUN'),\n",
       "  ('olan', 'ADJ'),\n",
       "  ('hastaların', 'NOUN'),\n",
       "  ('tedavi', 'NOUN'),\n",
       "  ('edileceği', 'ADJ'),\n",
       "  ('toplum', 'NOUN'),\n",
       "  ('temelli', 'NOUN'),\n",
       "  ('ruh', 'NOUN'),\n",
       "  ('sağlığı', 'NOUN'),\n",
       "  ('merkezlerinin', 'NOUN'),\n",
       "  ('sayısının', 'NOUN'),\n",
       "  ('artırılacağını', 'ADJ'),\n",
       "  ('bildirdi.', 'VERB'),\n",
       "  ('Başta', 'NOUN'),\n",
       "  ('sigaranın', 'NOUN'),\n",
       "  ('neden', 'NOUN'),\n",
       "  ('olduğu', 'ADJ'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('kanserinde,', 'NOUN'),\n",
       "  ('erken', 'ADJ'),\n",
       "  ('tanı', 'NOUN'),\n",
       "  ('ile', 'CONJ'),\n",
       "  ('hastalığın', 'NOUN'),\n",
       "  ('ilk', 'ADJ'),\n",
       "  ('evrede', 'NOUN'),\n",
       "  ('teşhisi', 'NOUN'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('tedavisi', 'NOUN'),\n",
       "  ('mümkün!', 'NOUN')],\n",
       " [('Florence', 'NOUN'),\n",
       "  ('Nightingale', 'NOUN'),\n",
       "  (\"Hastanesi'nden\", 'NOUN'),\n",
       "  ('Ortopedi', 'NOUN'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('Travmatoloji', 'NOUN'),\n",
       "  ('Uzmanı', 'NOUN'),\n",
       "  ('Prof', 'NOUN'),\n",
       "  ('Dr', 'NOUN'),\n",
       "  ('Uğur', 'NOUN'),\n",
       "  ('Işıklar', 'NOUN'),\n",
       "  ('kemik', 'NOUN'),\n",
       "  ('yoğunluğunda', 'NOUN'),\n",
       "  ('azalma', 'NOUN'),\n",
       "  ('rahatsızlığındaki', 'NOUN'),\n",
       "  ('artışa', 'NOUN'),\n",
       "  ('dikkat', 'NOUN'),\n",
       "  ('çekiyor:', 'VERB'),\n",
       "  ('Ayağı', 'NOUN'),\n",
       "  ('halıya', 'NOUN'),\n",
       "  ('takılan', 'ADJ'),\n",
       "  ('kişinin', 'NOUN'),\n",
       "  ('omuzu,', 'NOUN'),\n",
       "  ('el', 'NOUN'),\n",
       "  ('bileği', 'NOUN'),\n",
       "  ('kırılabiliyor.', 'VERB'),\n",
       "  ('Sağlık', 'NOUN'),\n",
       "  ('Bakanı', 'NOUN'),\n",
       "  ('Akdağ,', 'NOUN'),\n",
       "  ('bakanlığının', 'NOUN'),\n",
       "  ('2013', 'NUM'),\n",
       "  ('yılı', 'NOUN'),\n",
       "  ('projelerini', 'NOUN'),\n",
       "  ('Anadolu', 'NOUN'),\n",
       "  ('Ajansı', 'NOUN'),\n",
       "  ('muhabirine', 'NOUN'),\n",
       "  ('anlattı.', 'VERB'),\n",
       "  ('Akciğer', 'NOUN'),\n",
       "  ('kanserinin', 'NOUN'),\n",
       "  ('görülme', 'NOUN'),\n",
       "  ('sıklığı,', 'NOUN'),\n",
       "  ('kadınlarda', 'NOUN'),\n",
       "  ('da', 'CONJ'),\n",
       "  ('giderek', 'ADV'),\n",
       "  ('artıyor.', 'VERB')],\n",
       " [('2030', 'NUM'),\n",
       "  ('yılında', 'NOUN'),\n",
       "  ('hastane', 'NOUN'),\n",
       "  ('yataklarının', 'NOUN'),\n",
       "  ('üçte', 'NOUN'),\n",
       "  ('biri', 'NOUN'),\n",
       "  ('kemik', 'NOUN'),\n",
       "  ('erimesi', 'NOUN'),\n",
       "  ('sebebiyle', 'ADV'),\n",
       "  ('kırık', 'ADJ'),\n",
       "  ('vakası', 'NOUN'),\n",
       "  ('yaşayan', 'ADJ'),\n",
       "  ('hastalardan', 'NOUN'),\n",
       "  ('oluşacak.', 'VERB'),\n",
       "  ('Yeni', 'ADJ'),\n",
       "  ('yılda', 'NOUN'),\n",
       "  ('ruh', 'NOUN'),\n",
       "  ('sağlığı', 'NOUN'),\n",
       "  ('hizmetleriyle', 'NOUN'),\n",
       "  ('ilgili', 'ADJ'),\n",
       "  ('yenilikler', 'NOUN'),\n",
       "  ('olacağını', 'ADJ'),\n",
       "  ('açıklayan', 'ADJ'),\n",
       "  ('Akdağ,', 'NOUN'),\n",
       "  ('Hastaların', 'NOUN'),\n",
       "  ('depo', 'NOUN'),\n",
       "  ('hastanelere', 'NOUN'),\n",
       "  ('götürüldüğü', 'ADJ'),\n",
       "  ('modelden', 'NOUN'),\n",
       "  ('Toplum', 'NOUN'),\n",
       "  ('Temelli', 'ADJ'),\n",
       "  ('Ruh', 'NOUN'),\n",
       "  ('Sağlığı', 'NOUN'),\n",
       "  (\"Modeli'ne\", 'NOUN'),\n",
       "  ('geçiyoruz', 'VERB'),\n",
       "  ('dedi.', 'VERB'),\n",
       "  ('Ağırlıkta', 'NOUN'),\n",
       "  ('sigaranın', 'NOUN'),\n",
       "  ('neden', 'NOUN'),\n",
       "  ('olduğu', 'ADJ'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('kanserinde,', 'NOUN'),\n",
       "  ('erken', 'ADJ'),\n",
       "  ('tanı', 'NOUN'),\n",
       "  ('ile', 'CONJ'),\n",
       "  ('hastalığın', 'NOUN'),\n",
       "  ('ilk', 'ADJ'),\n",
       "  ('evrede', 'NOUN'),\n",
       "  ('teşhisi', 'NOUN'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('tedavisi', 'NOUN'),\n",
       "  ('mümkün!', 'NOUN')],\n",
       " [('Kadınların', 'NOUN'),\n",
       "  ('bu', 'DET'),\n",
       "  ('konuda', 'NOUN'),\n",
       "  ('dezavantajı', 'NOUN'),\n",
       "  ('daha', 'ADV'),\n",
       "  ('yüksek.', 'NOUN'),\n",
       "  ('Şu', 'DET'),\n",
       "  ('anda', 'NOUN'),\n",
       "  (\"50'den\", 'NUM'),\n",
       "  ('fazla', 'ADJ'),\n",
       "  ('toplum', 'NOUN'),\n",
       "  ('temelli', 'ADJ'),\n",
       "  ('ruh', 'NOUN'),\n",
       "  ('sağlığı', 'NOUN'),\n",
       "  ('merkezinin', 'NOUN'),\n",
       "  ('bulunduğunu', 'ADJ'),\n",
       "  ('bu', 'DET'),\n",
       "  ('sayının', 'NOUN'),\n",
       "  (\"200'ün\", 'NUM'),\n",
       "  ('üzerine', 'ADV'),\n",
       "  ('çıkarılacağını', 'ADJ'),\n",
       "  ('belirten', 'ADJ'),\n",
       "  ('Akdağ,', 'PRON'),\n",
       "  ('şöyle', 'ADV'),\n",
       "  ('devam', 'NOUN'),\n",
       "  ('etti:', 'VERB'),\n",
       "  ('Bütün', 'ADJ'),\n",
       "  ('şehirlerimizde', 'NOUN'),\n",
       "  ('ruh', 'NOUN'),\n",
       "  ('sağlığı', 'NOUN'),\n",
       "  ('problemi', 'NOUN'),\n",
       "  ('olan', 'ADJ'),\n",
       "  ('hastalarımızı', 'NOUN'),\n",
       "  ('toplum', 'NOUN'),\n",
       "  ('içinde', 'ADV'),\n",
       "  ('tedavi', 'NOUN'),\n",
       "  ('edebileceğimiz,', 'VERB'),\n",
       "  ('ailesiyle', 'NOUN'),\n",
       "  ('ya', 'CONJ'),\n",
       "  ('da', 'CONJ'),\n",
       "  ('toplumla', 'NOUN'),\n",
       "  ('irtibatını', 'NOUN'),\n",
       "  ('daha', 'ADV'),\n",
       "  ('yakın', 'ADJ'),\n",
       "  ('tutup', 'ADV'),\n",
       "  ('rehabilitasyonunu', 'NOUN'),\n",
       "  ('sağlayabileceğimiz', 'ADJ'),\n",
       "  ('bir', 'DET'),\n",
       "  ('modele', 'NOUN'),\n",
       "  ('geçiyoruz.', 'VERB'),\n",
       "  ('Erken', 'ADJ'),\n",
       "  ('evrede', 'NOUN'),\n",
       "  ('tümörleri', 'NOUN'),\n",
       "  ('yakalayan', 'ADJ'),\n",
       "  ('düşük', 'ADJ'),\n",
       "  ('doz', 'NOUN'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('tomografisi,', 'NOUN'),\n",
       "  ('ne', 'ADJ'),\n",
       "  ('zaman', 'NOUN'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('kimlere', 'PRON'),\n",
       "  ('yapılmalı?', 'VERB')],\n",
       " [('Bu', 'DET'),\n",
       "  ('yüzden', 'ADV'),\n",
       "  ('güneşle', 'NOUN'),\n",
       "  ('temas', 'NOUN'),\n",
       "  ('etmeleri,', 'NOUN'),\n",
       "  ('fazla', 'ADJ'),\n",
       "  ('kilo', 'NOUN'),\n",
       "  ('almamaları', 'NOUN'),\n",
       "  ('ya', 'CONJ'),\n",
       "  ('da', 'CONJ'),\n",
       "  ('zayıf', 'ADJ'),\n",
       "  ('kalmamaları', 'NOUN'),\n",
       "  ('gerekiyor.', 'VERB'),\n",
       "  ('Bu', 'DET'),\n",
       "  ('hastaneleri', 'NOUN'),\n",
       "  ('2', 'NUM'),\n",
       "  ('3', 'NUM'),\n",
       "  ('yıl', 'NOUN'),\n",
       "  ('içerisinde', 'ADV'),\n",
       "  ('dönüştürmüş', 'ADJ'),\n",
       "  ('olacağız.', 'VERB'),\n",
       "  ('Akciğer', 'NOUN'),\n",
       "  ('kanseri,', 'NOUN'),\n",
       "  ('son', 'ADJ'),\n",
       "  ('yıllarda', 'NOUN'),\n",
       "  ('giderek', 'ADV'),\n",
       "  ('artıyor.', 'VERB')],\n",
       " [('Aşırı', 'ADV'),\n",
       "  ('zayıf', 'ADJ'),\n",
       "  ('kadınlarda', 'NOUN'),\n",
       "  ('kemik', 'NOUN'),\n",
       "  ('erimesi', 'NOUN'),\n",
       "  ('daha', 'ADV'),\n",
       "  ('yoğun', 'ADJ'),\n",
       "  ('görülüyor.', 'VERB'),\n",
       "  ('Yani', 'CONJ'),\n",
       "  ('yüzlerce', 'ADJ'),\n",
       "  ('kişinin', 'NOUN'),\n",
       "  ('bir', 'DET'),\n",
       "  ('arada', 'NOUN'),\n",
       "  ('toplandığı', 'ADJ'),\n",
       "  ('hastanelerden', 'NOUN'),\n",
       "  ('kişinin', 'NOUN'),\n",
       "  ('daha', 'ADV'),\n",
       "  ('çok', 'ADJ'),\n",
       "  ('kendi', 'PRON'),\n",
       "  ('ortamında', 'NOUN'),\n",
       "  ('tedavi', 'NOUN'),\n",
       "  ('gördüğü', 'ADJ'),\n",
       "  ('bir', 'DET'),\n",
       "  ('modele', 'NOUN'),\n",
       "  ('geçiyoruz.', 'VERB'),\n",
       "  ('Geçmişte', 'NOUN'),\n",
       "  ('60', 'NUM'),\n",
       "  ('yaşın', 'NOUN'),\n",
       "  ('üzerindeki', 'ADJ'),\n",
       "  ('erkeklerde', 'NOUN'),\n",
       "  ('görülen', 'ADJ'),\n",
       "  ('bu', 'DET'),\n",
       "  ('kanser', 'NOUN'),\n",
       "  ('türü,', 'NOUN'),\n",
       "  ('günümüzde', 'NOUN'),\n",
       "  ('kadınlar', 'NOUN'),\n",
       "  ('arasında', 'NOUN'),\n",
       "  ('da', 'CONJ'),\n",
       "  ('sıklığı', 'NOUN'),\n",
       "  ('artmakla', 'NOUN'),\n",
       "  ('birlikte', 'ADV'),\n",
       "  ('erkeklerde', 'NOUN'),\n",
       "  ('görülme', 'NOUN'),\n",
       "  ('yaşı', 'NOUN'),\n",
       "  ('da', 'CONJ'),\n",
       "  ('60’ın', 'NUM'),\n",
       "  ('altına', 'NOUN'),\n",
       "  ('indi.', 'VERB')],\n",
       " [('Kemikleri', 'NOUN'),\n",
       "  ('güçlendirmek', 'NOUN'),\n",
       "  ('için', 'ADP'),\n",
       "  ('egzersiz', 'NOUN'),\n",
       "  ('dediğimizde', 'ADJ'),\n",
       "  ('insanlar', 'NOUN'),\n",
       "  ('ne', 'PRON'),\n",
       "  ('yapacağım', 'VERB'),\n",
       "  ('diye', 'ADP'),\n",
       "  ('düşünüyor.', 'VERB'),\n",
       "  ('Ambulans', 'NOUN'),\n",
       "  ('sayısı', 'NOUN'),\n",
       "  (\"5'e\", 'NUM'),\n",
       "  ('çıkacak.', 'VERB'),\n",
       "  ('Neolife', 'NOUN'),\n",
       "  ('Tıp', 'NOUN'),\n",
       "  ('Merkezi’nden', 'NOUN'),\n",
       "  ('Göğüs', 'NOUN'),\n",
       "  ('Cerrahisi', 'NOUN'),\n",
       "  ('Uzmanı', 'NOUN'),\n",
       "  ('Doç', 'NOUN'),\n",
       "  ('Dr', 'NOUN'),\n",
       "  ('Gökhan', 'NOUN'),\n",
       "  ('Hacıibrahimoğlu,', 'NOUN'),\n",
       "  ('hızlı', 'ADJ'),\n",
       "  ('ilerleyen', 'ADJ'),\n",
       "  ('kanser', 'NOUN'),\n",
       "  ('türleri', 'NOUN'),\n",
       "  ('arasında', 'NOUN'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('kanserinin', 'NOUN'),\n",
       "  ('erkeklerde', 'NOUN'),\n",
       "  ('birinci,', 'ADJ'),\n",
       "  ('kadınlarda', 'NOUN'),\n",
       "  ('ise', 'CONJ'),\n",
       "  ('ikinci', 'ADJ'),\n",
       "  ('sıklıkta', 'NOUN'),\n",
       "  ('görüldüğünü', 'ADJ'),\n",
       "  ('belirtiyor.', 'VERB')],\n",
       " [('Oysa', 'NOUN'),\n",
       "  ('egzersiz', 'NOUN'),\n",
       "  ('bir', 'DET'),\n",
       "  ('harekettir.', 'NOUN'),\n",
       "  ('Yeni', 'ADJ'),\n",
       "  ('yılda', 'NOUN'),\n",
       "  ('ayrıca', 'ADV'),\n",
       "  ('hava', 'NOUN'),\n",
       "  ('ambulans', 'NOUN'),\n",
       "  ('filosunu', 'NOUN'),\n",
       "  ('güçlendireceklerini', 'ADJ'),\n",
       "  ('anlatan', 'ADJ'),\n",
       "  ('Bakan', 'NOUN'),\n",
       "  ('Akdağ,', 'PRON'),\n",
       "  ('3', 'NUM'),\n",
       "  ('yeni', 'ADJ'),\n",
       "  ('ambulans', 'NOUN'),\n",
       "  ('uçağın', 'NOUN'),\n",
       "  ('eklenmesiyle', 'NOUN'),\n",
       "  ('toplam', 'NOUN'),\n",
       "  ('sayının', 'NOUN'),\n",
       "  (\"5'e\", 'NUM'),\n",
       "  ('ulaşacağını', 'ADJ'),\n",
       "  ('belirtti.', 'VERB'),\n",
       "  ('Hacıibrahimoğlu;', 'NOUN'),\n",
       "  ('sigara,', 'NOUN'),\n",
       "  ('puro', 'NOUN'),\n",
       "  ('pipo', 'NOUN'),\n",
       "  ('kullanımı,', 'NOUN'),\n",
       "  ('pasif', 'ADJ'),\n",
       "  ('içicilik;', 'NOUN'),\n",
       "  ('sanayi', 'NOUN'),\n",
       "  ('gelişmesine', 'NOUN'),\n",
       "  ('bağlı', 'ADJ'),\n",
       "  ('olarak', 'ADP'),\n",
       "  ('kullanımı', 'NOUN'),\n",
       "  ('artan', 'ADJ'),\n",
       "  ('asbest,', 'NOUN'),\n",
       "  ('benzen,', 'NOUN'),\n",
       "  ('arsenik', 'NOUN'),\n",
       "  ('gibi', 'ADP'),\n",
       "  ('kanser', 'NOUN'),\n",
       "  ('yapıcı', 'ADJ'),\n",
       "  ('maddeler;', 'NOUN'),\n",
       "  ('krom,', 'NOUN'),\n",
       "  ('nikel,', 'NOUN'),\n",
       "  ('kadmiyum', 'NOUN'),\n",
       "  ('gibi', 'ADP'),\n",
       "  ('metaller,', 'NOUN'),\n",
       "  ('hava', 'NOUN'),\n",
       "  ('kirliliği,', 'NOUN'),\n",
       "  ('ultraviyole', 'NOUN'),\n",
       "  ('ışınları', 'NOUN'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('geçirilmiş', 'ADJ'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('hastalıklarının,', 'NOUN'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('kanserine', 'NOUN'),\n",
       "  ('yakalanma', 'NOUN'),\n",
       "  ('riskini', 'NOUN'),\n",
       "  ('artırdığına', 'ADJ'),\n",
       "  ('dikkat', 'NOUN'),\n",
       "  ('çekiyor.', 'VERB')],\n",
       " [('Egzersiz', 'NOUN'),\n",
       "  ('yapamıyorsanız', 'VERB'),\n",
       "  ('dans', 'NOUN'),\n",
       "  ('edin.', 'VERB'),\n",
       "  ('Yapılan', 'VERB'),\n",
       "  ('pek', 'ADV'),\n",
       "  ('çok', 'ADJ'),\n",
       "  ('çalışma', 'NOUN'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('kanserlerinin', 'NOUN'),\n",
       "  ('yüzde', 'NOUN'),\n",
       "  ('90', 'NUM'),\n",
       "  ('sigara', 'NOUN'),\n",
       "  ('kaynaklı', 'NOUN'),\n",
       "  ('olduğunu', 'VERB'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('sigaranın', 'NOUN'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('kanser', 'NOUN'),\n",
       "  ('riskini', 'NOUN'),\n",
       "  ('5', 'NUM'),\n",
       "  ('ila', 'CONJ'),\n",
       "  ('15', 'NUM'),\n",
       "  ('kat', 'NOUN'),\n",
       "  ('arasında', 'NOUN'),\n",
       "  ('artırdığını', 'NOUN'),\n",
       "  ('gösteriyor.', 'VERB')],\n",
       " [('Sadece', 'ADV'),\n",
       "  ('kadınlara', 'NOUN'),\n",
       "  ('değil,', 'VERB'),\n",
       "  ('erkeklere', 'NOUN'),\n",
       "  ('de', 'CONJ'),\n",
       "  ('bol', 'ADJ'),\n",
       "  ('bol', 'PRT'),\n",
       "  ('dans', 'NOUN'),\n",
       "  ('etmelerini', 'NOUN'),\n",
       "  ('öneriyorum.', 'VERB'),\n",
       "  ('Sigaraya', 'NOUN'),\n",
       "  ('başlama', 'NOUN'),\n",
       "  ('yaşı,', 'NOUN'),\n",
       "  ('toplam', 'NOUN'),\n",
       "  ('sigara', 'NOUN'),\n",
       "  ('içme', 'NOUN'),\n",
       "  ('süresi', 'NOUN'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('günde', 'NOUN'),\n",
       "  ('içilen', 'ADJ'),\n",
       "  ('sigara', 'NOUN'),\n",
       "  ('sayısı', 'NOUN'),\n",
       "  ('ile', 'CONJ'),\n",
       "  ('bu', 'DET'),\n",
       "  ('risk', 'NOUN'),\n",
       "  ('artarken', 'ADJ'),\n",
       "  ('sigaranın', 'NOUN'),\n",
       "  ('bırakılması', 'NOUN'),\n",
       "  ('ile', 'CONJ'),\n",
       "  ('azalıyor.', 'VERB')],\n",
       " [('Tango', 'NOUN'),\n",
       "  ('yapın,', 'VERB'),\n",
       "  ('göbek', 'NOUN'),\n",
       "  ('atın,', 'VERB'),\n",
       "  ('zıplayın.', 'VERB'),\n",
       "  ('Bazı', 'ADJ'),\n",
       "  ('meslekler', 'NOUN'),\n",
       "  ('de', 'CONJ'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('kanserini', 'NOUN'),\n",
       "  ('tetikliyor', 'VERB'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('yakalanma', 'NOUN'),\n",
       "  ('riskini', 'NOUN'),\n",
       "  ('artırıyor.', 'VERB')],\n",
       " [('Bu', 'DET'),\n",
       "  ('nedenle', 'NOUN'),\n",
       "  ('madenciler,', 'NOUN'),\n",
       "  ('tekstil,', 'NOUN'),\n",
       "  ('izolasyon', 'NOUN'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('tersane', 'NOUN'),\n",
       "  ('işçileri,', 'NOUN'),\n",
       "  ('petro-kimya,', 'NOUN'),\n",
       "  ('baca', 'NOUN'),\n",
       "  ('temizleyicileri,', 'NOUN'),\n",
       "  ('plastik', 'ADJ'),\n",
       "  ('sanayi', 'NOUN'),\n",
       "  ('işçileri,', 'NOUN'),\n",
       "  ('maden', 'ADJ'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('kaynak', 'NOUN'),\n",
       "  ('işçileri,', 'NOUN'),\n",
       "  ('çamaşır', 'NOUN'),\n",
       "  ('suyu', 'NOUN'),\n",
       "  ('üreticileri,', 'NOUN'),\n",
       "  ('cam', 'NOUN'),\n",
       "  ('seramik,', 'NOUN'),\n",
       "  ('muşamba', 'NOUN'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('batarya', 'NOUN'),\n",
       "  ('işçileri,', 'NOUN'),\n",
       "  ('boya,', 'NOUN'),\n",
       "  ('çelik,', 'NOUN'),\n",
       "  ('dökümhane', 'NOUN'),\n",
       "  ('işçilerinin', 'NOUN'),\n",
       "  ('düzenli', 'ADJ'),\n",
       "  ('aralıklarla', 'NOUN'),\n",
       "  ('kontrol', 'NOUN'),\n",
       "  ('testlerini', 'NOUN'),\n",
       "  ('yaptırmaları', 'NOUN'),\n",
       "  ('gerekiyor.', 'VERB')],\n",
       " [('Her', 'DET'),\n",
       "  ('kanser', 'NOUN'),\n",
       "  ('türünde', 'NOUN'),\n",
       "  ('olduğu', 'ADJ'),\n",
       "  ('gibi', 'ADP'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('kanserinde', 'NOUN'),\n",
       "  ('de', 'CONJ'),\n",
       "  ('erken', 'ADJ'),\n",
       "  ('tanı', 'NOUN'),\n",
       "  ('ile', 'CONJ'),\n",
       "  ('hastalığın', 'NOUN'),\n",
       "  ('ilk', 'ADJ'),\n",
       "  ('evrede', 'NOUN'),\n",
       "  ('teşhisi', 'NOUN'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('tam', 'ADJ'),\n",
       "  ('tedavisi', 'NOUN'),\n",
       "  ('mümkün.', 'NOUN')],\n",
       " [('Akciğer', 'NOUN'),\n",
       "  ('kanserinin', 'NOUN'),\n",
       "  ('teşhisinde;', 'NOUN'),\n",
       "  ('bilgisayarlı', 'NOUN'),\n",
       "  ('tomografi,', 'NOUN'),\n",
       "  ('manyetik', 'ADJ'),\n",
       "  ('rezonans', 'NOUN'),\n",
       "  ('görüntüleme,', 'NOUN'),\n",
       "  ('pozitron', 'NOUN'),\n",
       "  ('emisyon', 'NOUN'),\n",
       "  ('tomografisi,', 'NOUN'),\n",
       "  ('kan', 'NOUN'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('balgam', 'NOUN'),\n",
       "  ('testleri', 'NOUN'),\n",
       "  ('kullanılıyor.', 'VERB')],\n",
       " [('Neolife', 'NOUN'),\n",
       "  ('Tıp', 'NOUN'),\n",
       "  ('Merkezi’nden', 'NOUN'),\n",
       "  ('Radyoloji', 'NOUN'),\n",
       "  ('Uzmanı', 'NOUN'),\n",
       "  ('Dr', 'NOUN'),\n",
       "  ('Esra', 'NOUN'),\n",
       "  ('Çay,', 'NOUN'),\n",
       "  ('hastalığın', 'NOUN'),\n",
       "  ('erken', 'NOUN'),\n",
       "  ('dönemde', 'NOUN'),\n",
       "  ('belirti', 'NOUN'),\n",
       "  ('vermemesi', 'NOUN'),\n",
       "  ('nedeniyle', 'ADV'),\n",
       "  ('özellikle', 'ADV'),\n",
       "  ('yoğun', 'ADJ'),\n",
       "  ('sigara', 'NOUN'),\n",
       "  ('içenlerde', 'ADJ'),\n",
       "  ('tümörün', 'NOUN'),\n",
       "  ('tespitini', 'NOUN'),\n",
       "  ('sağlayan', 'ADJ'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('tedavinin', 'NOUN'),\n",
       "  ('başarısını', 'NOUN'),\n",
       "  ('artıran', 'ADJ'),\n",
       "  ('bilgisayarlı', 'NOUN'),\n",
       "  ('tomografinin', 'NOUN'),\n",
       "  ('erken', 'ADJ'),\n",
       "  ('evre', 'NOUN'),\n",
       "  ('teşhiste', 'NOUN'),\n",
       "  ('önemini', 'NOUN'),\n",
       "  ('vurguluyor', 'VERB'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('şöyle', 'ADV'),\n",
       "  ('devam', 'NOUN'),\n",
       "  ('ediyor:', 'VERB'),\n",
       "  ('Akciğer', 'NOUN'),\n",
       "  ('kanseri', 'NOUN'),\n",
       "  ('taramasında', 'VERB'),\n",
       "  ('tanısal', 'NOUN'),\n",
       "  ('bir', 'DET'),\n",
       "  ('yöntemin', 'NOUN'),\n",
       "  ('yeterli', 'ADJ'),\n",
       "  ('olabilmesi', 'NOUN'),\n",
       "  ('için', 'ADP'),\n",
       "  ('erken', 'ADJ'),\n",
       "  ('evrede', 'NOUN'),\n",
       "  ('tümörleri', 'NOUN'),\n",
       "  ('yakalamada', 'NOUN'),\n",
       "  ('duyarlılığı', 'NOUN'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('iyi', 'ADJ'),\n",
       "  ('huylu', 'ADJ'),\n",
       "  ('nodülleri', 'NOUN'),\n",
       "  ('saptamada', 'NOUN'),\n",
       "  ('seçiciliğinin', 'NOUN'),\n",
       "  ('yüksek', 'ADJ'),\n",
       "  ('olması', 'NOUN'),\n",
       "  ('gerekiyor.', 'VERB')],\n",
       " [('Radyasyona', 'NOUN'),\n",
       "  ('daha', 'ADV'),\n",
       "  ('az', 'ADJ'),\n",
       "  ('maruz', 'NOUN'),\n",
       "  ('kalınıyor.', 'VERB')],\n",
       " [('Bilgisayarlı', 'NOUN'),\n",
       "  ('tomografi', 'NOUN'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('nodüllerinin', 'NOUN'),\n",
       "  ('tespitinde,', 'NOUN'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('grafisinden', 'NOUN'),\n",
       "  ('daha', 'ADV'),\n",
       "  ('duyarlı', 'ADJ'),\n",
       "  ('bir', 'DET'),\n",
       "  ('inceleme', 'NOUN'),\n",
       "  ('yöntemidir.', 'NOUN')],\n",
       " [('Akciğer', 'NOUN'),\n",
       "  ('grafisi', 'NOUN'),\n",
       "  ('teknik', 'ADJ'),\n",
       "  ('özelliklerinden', 'NOUN'),\n",
       "  ('dolayı', 'NOUN'),\n",
       "  ('küçük', 'ADJ'),\n",
       "  ('çaplı', 'NOUN'),\n",
       "  ('tümörleri', 'NOUN'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('akciğerlerin', 'NOUN'),\n",
       "  ('hilus', 'NOUN'),\n",
       "  ('adı', 'NOUN'),\n",
       "  ('verilen', 'ADJ'),\n",
       "  ('kesimlerinde', 'NOUN'),\n",
       "  ('yerleşim', 'NOUN'),\n",
       "  ('gösteren', 'ADJ'),\n",
       "  ('ve', 'CONJ'),\n",
       "  ('orta', 'ADJ'),\n",
       "  ('hatta', 'CONJ'),\n",
       "  ('yakın', 'ADJ'),\n",
       "  ('tümörleri', 'NOUN'),\n",
       "  ('erken', 'ADJ'),\n",
       "  ('evrede', 'NOUN'),\n",
       "  ('saptayamıyor.', 'VERB')],\n",
       " [('Üstelik', 'ADV'),\n",
       "  ('düşük', 'ADJ'),\n",
       "  ('doz', 'NOUN'),\n",
       "  ('bilgisayarlı', 'NOUN'),\n",
       "  ('tomografi', 'NOUN'),\n",
       "  ('tekniği,', 'NOUN'),\n",
       "  ('normal', 'ADJ'),\n",
       "  ('bilgisayarlı', 'NOUN'),\n",
       "  ('tomografiye', 'NOUN'),\n",
       "  ('göre', 'ADV'),\n",
       "  ('hastaya', 'NOUN'),\n",
       "  ('daha', 'NOUN'),\n",
       "  ('az', 'ADJ'),\n",
       "  ('radyasyon', 'NOUN'),\n",
       "  ('veren,', 'ADJ'),\n",
       "  ('ancak', 'CONJ'),\n",
       "  ('teşhis', 'NOUN'),\n",
       "  ('için', 'ADP'),\n",
       "  ('yeterli', 'ADJ'),\n",
       "  ('kalitede', 'NOUN'),\n",
       "  ('görüntü', 'NOUN'),\n",
       "  ('alınmasını', 'NOUN'),\n",
       "  ('sağlayan', 'ADJ'),\n",
       "  ('bir', 'DET'),\n",
       "  ('yöntemdir.', 'VERB')],\n",
       " [('Amerikan', 'ADJ'),\n",
       "  ('Akciğer', 'NOUN'),\n",
       "  ('Derneği,', 'NOUN'),\n",
       "  ('30', 'NUM'),\n",
       "  ('yıl', 'VERB'),\n",
       "  ('boyunca', 'ADV'),\n",
       "  ('günde', 'NOUN'),\n",
       "  ('bir', 'DET'),\n",
       "  ('paket', 'NOUN'),\n",
       "  ('sigara', 'NOUN'),\n",
       "  ('içmiş', 'ADJ'),\n",
       "  ('kişiler,', 'NOUN'),\n",
       "  ('sigara', 'NOUN'),\n",
       "  ('içen', 'ADJ'),\n",
       "  ('ya', 'CONJ'),\n",
       "  ('da', 'CONJ'),\n",
       "  ('sigarayı', 'NOUN'),\n",
       "  ('bırakmış', 'ADJ'),\n",
       "  ('55', 'NUM'),\n",
       "  ('-', '.'),\n",
       "  ('74', 'NUM'),\n",
       "  ('yaş', 'NOUN'),\n",
       "  ('arası', 'NOUN'),\n",
       "  ('kişiler,', 'NOUN'),\n",
       "  ('sigarayı', 'NOUN'),\n",
       "  ('bırakmasının', 'NOUN'),\n",
       "  ('üzerinden', 'ADV'),\n",
       "  ('15', 'NUM'),\n",
       "  ('yıldan', 'NOUN'),\n",
       "  ('daha', 'ADV'),\n",
       "  ('az', 'ADJ'),\n",
       "  ('zaman', 'NOUN'),\n",
       "  ('geçmiş', 'ADJ'),\n",
       "  ('kişilerde', 'NOUN'),\n",
       "  ('tarama', 'NOUN'),\n",
       "  ('amaçlı', 'ADJ'),\n",
       "  ('olarak', 'ADP'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('grafisi', 'NOUN'),\n",
       "  ('yerine,', 'NOUN'),\n",
       "  ('düşük', 'ADJ'),\n",
       "  ('doz', 'NOUN'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('tomografisini', 'NOUN'),\n",
       "  ('öneriyor.', 'VERB')],\n",
       " [('Yüksek', 'ADJ'),\n",
       "  ('risk', 'NOUN'),\n",
       "  ('taşıyanların', 'ADJ'),\n",
       "  ('düşük', 'ADJ'),\n",
       "  ('doz', 'NOUN'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('tomografisi', 'NOUN'),\n",
       "  ('ile', 'CONJ'),\n",
       "  ('taranması', 'NOUN'),\n",
       "  ('hem', 'CONJ'),\n",
       "  ('akciğer', 'NOUN'),\n",
       "  ('kanserinin', 'NOUN'),\n",
       "  ('erken', 'ADJ'),\n",
       "  ('teşhisini', 'NOUN'),\n",
       "  ('sağlıyor,', 'VERB'),\n",
       "  ('hem', 'CONJ'),\n",
       "  ('de', 'CONJ'),\n",
       "  ('radyasyon', 'NOUN'),\n",
       "  ('ile', 'CONJ'),\n",
       "  ('ilişkili', 'NOUN'),\n",
       "  ('riskleri', 'NOUN'),\n",
       "  ('minimize', 'NOUN'),\n",
       "  ('ediyor.', 'VERB')]]"
      ]
     },
     "execution_count": 165,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sentences = getter.sentences; sentences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('İster', 'CONJ'),\n",
       " ('tango', 'NOUN'),\n",
       " ('yap', 'VERB'),\n",
       " ('ister', 'CONJ'),\n",
       " ('göbek', 'NOUN'),\n",
       " ('at!', 'VERB'),\n",
       " ('Dünya', 'NOUN'),\n",
       " ('Sağlık', 'NOUN'),\n",
       " (\"Örgütü'nün\", 'NOUN'),\n",
       " ('rakamlarına', 'NOUN'),\n",
       " ('göre', 'ADV'),\n",
       " (\"2000'de\", 'NUM'),\n",
       " ('66', 'NUM'),\n",
       " ('olan', 'ADJ'),\n",
       " ('ortalama', 'ADJ'),\n",
       " ('ömür', 'NOUN'),\n",
       " (\"2009'da\", 'NUM'),\n",
       " (\"68'e\", 'NUM'),\n",
       " ('yükseldi.', 'VERB'),\n",
       " ('Akıl', 'NOUN'),\n",
       " ('hastaneleri', 'NOUN'),\n",
       " ('tarih', 'NOUN'),\n",
       " ('olacak.', 'VERB'),\n",
       " ('Akciğer', 'NOUN'),\n",
       " ('tomografisi', 'NOUN'),\n",
       " ('hayat', 'NOUN'),\n",
       " ('kurtarıyor.', 'VERB')]"
      ]
     },
     "execution_count": 166,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sentences[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {},
   "outputs": [],
   "source": [
    "agg_func = lambda s: [(w, t) for w, t in zip(s[\"word\"].values.tolist(),\n",
    "                                                           s[\"tag\"].values.tolist())]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('İster', 'CONJ'),\n",
       " ('tango', 'NOUN'),\n",
       " ('yap', 'VERB'),\n",
       " ('ister', 'CONJ'),\n",
       " ('göbek', 'NOUN'),\n",
       " ('at!', 'VERB'),\n",
       " ('Dünya', 'NOUN'),\n",
       " ('Sağlık', 'NOUN'),\n",
       " (\"Örgütü'nün\", 'NOUN'),\n",
       " ('rakamlarına', 'NOUN'),\n",
       " ('göre', 'ADV'),\n",
       " (\"2000'de\", 'NUM'),\n",
       " ('66', 'NUM'),\n",
       " ('olan', 'ADJ'),\n",
       " ('ortalama', 'ADJ'),\n",
       " ('ömür', 'NOUN'),\n",
       " (\"2009'da\", 'NUM'),\n",
       " (\"68'e\", 'NUM'),\n",
       " ('yükseldi.', 'VERB'),\n",
       " ('Akıl', 'NOUN'),\n",
       " ('hastaneleri', 'NOUN'),\n",
       " ('tarih', 'NOUN'),\n",
       " ('olacak.', 'VERB'),\n",
       " ('Akciğer', 'NOUN'),\n",
       " ('tomografisi', 'NOUN'),\n",
       " ('hayat', 'NOUN'),\n",
       " ('kurtarıyor.', 'VERB')]"
      ]
     },
     "execution_count": 168,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.groupby(\"sentence_no\").apply(agg_func)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {},
   "outputs": [],
   "source": [
    "max_len = 75\n",
    "max_len_char = 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "metadata": {},
   "outputs": [],
   "source": [
    "word2idx = {w: i + 2 for i, w in enumerate(words)}\n",
    "word2idx[\"UNK\"] = 1\n",
    "word2idx[\"PAD\"] = 0\n",
    "idx2word = {i: w for w, i in word2idx.items()}\n",
    "tag2idx = {t: i + 1 for i, t in enumerate(tags)}\n",
    "tag2idx[\"PAD\"] = 0\n",
    "idx2tag = {i: w for w, i in tag2idx.items()}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "108\n",
      "4\n"
     ]
    }
   ],
   "source": [
    "print(word2idx[\"ömür\"])\n",
    "print(tag2idx[\"NOUN\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.preprocessing.sequence import pad_sequences\n",
    "X_word = [[word2idx[w[0]] for w in s] for s in sentences]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_word = pad_sequences(maxlen=max_len, sequences=X_word, value=word2idx[\"PAD\"], padding='post', truncating='post')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "72\n"
     ]
    }
   ],
   "source": [
    "chars = set([w_i for w in words for w_i in w])\n",
    "n_chars = len(chars)\n",
    "print(n_chars)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {},
   "outputs": [],
   "source": [
    "char2idx = {c: i + 2 for i, c in enumerate(chars)}\n",
    "char2idx[\"UNK\"] = 1\n",
    "char2idx[\"PAD\"] = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_char = []\n",
    "for sentence in sentences:\n",
    "    sent_seq = []\n",
    "    for i in range(max_len):\n",
    "        word_seq = []\n",
    "        for j in range(max_len_char):\n",
    "            try:\n",
    "                word_seq.append(char2idx.get(sentence[i][0][j]))\n",
    "            except:\n",
    "                word_seq.append(char2idx.get(\"PAD\"))\n",
    "        sent_seq.append(word_seq)\n",
    "    X_char.append(np.array(sent_seq))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {},
   "outputs": [],
   "source": [
    "y = [[tag2idx[w[1]] for w in s] for s in sentences]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "metadata": {},
   "outputs": [],
   "source": [
    "y = pad_sequences(maxlen=max_len, sequences=y, value=tag2idx[\"PAD\"], padding='post', truncating='post')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "X_word_tr, X_word_te, y_tr, y_te = train_test_split(X_word, y, test_size=0.1, random_state=2018)\n",
    "X_char_tr, X_char_te, _, _ = train_test_split(X_char, y, test_size=0.1, random_state=2018)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import Model, Input\n",
    "from keras.layers import LSTM, Embedding, Dense, TimeDistributed, Dropout, Conv1D\n",
    "from keras.layers import Bidirectional, concatenate, SpatialDropout1D, GlobalMaxPooling1D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "metadata": {},
   "outputs": [],
   "source": [
    "# input and embedding for words\n",
    "word_in = Input(shape=(max_len,))\n",
    "emb_word = Embedding(input_dim=n_words + 2, output_dim=20,\n",
    "                     input_length=max_len, mask_zero=True)(word_in)\n",
    "\n",
    "# input and embeddings for characters\n",
    "char_in = Input(shape=(max_len, max_len_char,))\n",
    "emb_char = TimeDistributed(Embedding(input_dim=n_chars + 2, output_dim=10,\n",
    "                           input_length=max_len_char, mask_zero=True))(char_in)\n",
    "# character LSTM to get word encodings by characters\n",
    "char_enc = TimeDistributed(LSTM(units=20, return_sequences=False,\n",
    "                                recurrent_dropout=0.3))(emb_char)\n",
    "\n",
    "# main LSTM\n",
    "x = concatenate([emb_word, char_enc])\n",
    "x = SpatialDropout1D(0.1)(x)\n",
    "main_lstm = Bidirectional(LSTM(units=50, return_sequences=True,\n",
    "                               recurrent_dropout=0.2))(x)\n",
    "out = TimeDistributed(Dense(n_tags + 1, activation=\"sigmoid\"))(main_lstm)\n",
    "\n",
    "model = Model([word_in, char_in], out)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(optimizer=\"adam\", loss=\"sparse_categorical_crossentropy\", metrics=[\"acc\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_5\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_10 (InputLayer)           (None, 75, 10)       0                                            \n",
      "__________________________________________________________________________________________________\n",
      "input_9 (InputLayer)            (None, 75)           0                                            \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_13 (TimeDistri (None, 75, 10, 10)   740         input_10[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "embedding_9 (Embedding)         (None, 75, 20)       10380       input_9[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_14 (TimeDistri (None, 75, 20)       2480        time_distributed_13[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_5 (Concatenate)     (None, 75, 40)       0           embedding_9[0][0]                \n",
      "                                                                 time_distributed_14[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "spatial_dropout1d_5 (SpatialDro (None, 75, 40)       0           concatenate_5[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bidirectional_5 (Bidirectional) (None, 75, 100)      36400       spatial_dropout1d_5[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_15 (TimeDistri (None, 75, 12)       1212        bidirectional_5[0][0]            \n",
      "==================================================================================================\n",
      "Total params: 51,212\n",
      "Trainable params: 51,212\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/merve/anaconda3/lib/python3.7/site-packages/tensorflow_core/python/framework/indexed_slices.py:433: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.\n",
      "  \"Converting sparse IndexedSlices to a dense Tensor of unknown shape. \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 17 samples, validate on 2 samples\n",
      "Epoch 1/10\n",
      "17/17 [==============================] - 20s 1s/step - loss: 1.0465 - acc: 0.1210 - val_loss: 1.5376 - val_acc: 0.4409\n",
      "Epoch 2/10\n",
      "17/17 [==============================] - 2s 101ms/step - loss: 1.0443 - acc: 0.4786 - val_loss: 1.5342 - val_acc: 0.6022\n",
      "Epoch 3/10\n",
      "17/17 [==============================] - 2s 100ms/step - loss: 1.0418 - acc: 0.5773 - val_loss: 1.5304 - val_acc: 0.6452\n",
      "Epoch 4/10\n",
      "17/17 [==============================] - 2s 95ms/step - loss: 1.0395 - acc: 0.5885 - val_loss: 1.5263 - val_acc: 0.6452\n",
      "Epoch 5/10\n",
      "17/17 [==============================] - 2s 116ms/step - loss: 1.0363 - acc: 0.5903 - val_loss: 1.5216 - val_acc: 0.6452\n",
      "Epoch 6/10\n",
      "17/17 [==============================] - 2s 123ms/step - loss: 1.0330 - acc: 0.5903 - val_loss: 1.5162 - val_acc: 0.6452\n",
      "Epoch 7/10\n",
      "17/17 [==============================] - 2s 107ms/step - loss: 1.0292 - acc: 0.5903 - val_loss: 1.5101 - val_acc: 0.6452\n",
      "Epoch 8/10\n",
      "17/17 [==============================] - 2s 115ms/step - loss: 1.0246 - acc: 0.5903 - val_loss: 1.5030 - val_acc: 0.6452\n",
      "Epoch 9/10\n",
      "17/17 [==============================] - 2s 101ms/step - loss: 1.0196 - acc: 0.5903 - val_loss: 1.4948 - val_acc: 0.6452\n",
      "Epoch 10/10\n",
      "17/17 [==============================] - 2s 98ms/step - loss: 1.0136 - acc: 0.5903 - val_loss: 1.4852 - val_acc: 0.6452\n"
     ]
    }
   ],
   "source": [
    "history = model.fit([X_word_tr,\n",
    "                     np.array(X_char_tr).reshape((len(X_char_tr), max_len, max_len_char))],\n",
    "                    np.array(y_tr).reshape(len(y_tr), max_len, 1),\n",
    "                    batch_size=32, epochs=10, validation_split=0.1, verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {},
   "outputs": [],
   "source": [
    "hist = pd.DataFrame(history.history)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x864 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.style.use(\"ggplot\")\n",
    "plt.figure(figsize=(12,12))\n",
    "plt.plot(hist[\"acc\"])\n",
    "plt.plot(hist[\"val_acc\"])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = model.predict([X_word_te,\n",
    "                        np.array(X_char_te).reshape((len(X_char_te),\n",
    "                                                     max_len, max_len_char))])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Word           ||True ||Pred\n",
      "==============================\n",
      "Akciğer        : NOUN  NOUN\n",
      "kanserinin     : NOUN  NOUN\n",
      "teşhisinde;    : NOUN  NOUN\n",
      "bilgisayarlı   : NOUN  NOUN\n",
      "tomografi,     : NOUN  NOUN\n",
      "manyetik       : ADJ   NOUN\n",
      "rezonans       : NOUN  NOUN\n",
      "görüntüleme,   : NOUN  NOUN\n",
      "pozitron       : NOUN  NOUN\n",
      "emisyon        : NOUN  NOUN\n",
      "tomografisi,   : NOUN  NOUN\n",
      "kan            : NOUN  NOUN\n",
      "ve             : CONJ  NOUN\n",
      "balgam         : NOUN  NOUN\n",
      "testleri       : NOUN  NOUN\n",
      "kullanılıyor.  : VERB  NOUN\n"
     ]
    }
   ],
   "source": [
    "i = 1\n",
    "p = np.argmax(y_pred[i], axis=-1)\n",
    "print(\"{:15}||{:5}||{}\".format(\"Word\", \"True\", \"Pred\"))\n",
    "print(30 * \"=\")\n",
    "for w, t, pred in zip(X_word_te[i], y_te[i], p):\n",
    "    if w != 0:\n",
    "        print(\"{:15}: {:5} {}\".format(idx2word[w], idx2tag[t], idx2tag[pred]))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}