probabilistic_language_modeling/assessing_corpus_quality.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Assessing Corpus Quality\n",
    "### In this notebook, we'll learn about assessing corpus quality and potentially correcting problems. \n",
    "## Potential Problem areas\n",
    "1. Unexpected characters\n",
    "1. Improperly joined words\n",
    "1. Loanwords\n",
    "### We will consider each of these in turn.\n",
    "### Datasets: We'll be using the freely available CLTK datasets of the Latin Library and Perseus Latin texts, but the techniques in this notebook will apply to other languages as well.\n",
    "#### set up the notebook"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### A few open source imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from collections import Counter, defaultdict\n",
    "import re\n",
    "from bisect import bisect\n",
    "import statistics\n",
    "\n",
    "from cltk.corpus.readers import get_corpus_reader\n",
    "from cltk.stem.latin.j_v import JVReplacer\n",
    "from tqdm import tqdm\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.ticker import MaxNLocator\n",
    "plt.style.use('fivethirtyeight')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Add parent directory to path so we can access our common code"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os,sys,inspect\n",
    "currentdir = os.path.dirname(os.path.abspath(inspect.getfile(inspect.currentframe())))\n",
    "parentdir = os.path.dirname(currentdir)\n",
    "sys.path.insert(0,parentdir) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from mlyoucanuse.corpus_analysis_fun import (get_word_lengths,\n",
    "                                             get_samples_for_lengths,\n",
    "                                             get_char_counts)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## The Latin Library Corpus"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "latin_library_reader = get_corpus_reader(corpus_name='latin_text_latin_library', language='latin')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## The Perseus Latin Corpus"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "perseus_latin_reader = get_corpus_reader(corpus_name='latin_text_perseus', language='latin')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "### 1. Unexpected Characters\n",
    "#### Let's have a look at the characters: Grab all the words, check if they are alphabetical, and count them"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 2141/2141 [14:50<00:00,  1.77files/s]    | 1/2141 [00:00<06:09,  5.79files/s]\n",
      "  0%|          | 0/293 [00:00<?, ?files/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Latin Library Corpus: Number of characters: 75,451,171\n",
      "Latin Library character frequency distribution: Counter({'e': 8576123, 'i': 8573678, 't': 6306262, 'a': 6258185, 'u': 6106664, 's': 5695715, 'r': 4717114, 'n': 4700797, 'o': 4184861, 'm': 3968984, 'c': 2826383, 'd': 2323221, 'l': 2220037, 'p': 2046252, 'b': 1052786, 'q': 1000896, 'v': 764958, 'g': 761427, 'f': 676516, 'h': 536389, 'x': 341117, 'I': 205845, 'A': 153068, 'S': 134701, 'C': 130344, 'D': 119576, 'P': 107924, 'E': 90935, 'N': 74718, 'T': 73036, 'M': 69371, 'R': 64627, 'V': 64109, 'H': 58901, 'Q': 57106, 'y': 56645, 'L': 56146, 'X': 42245, 'O': 41873, 'U': 38028, 'G': 35341, 'B': 28814, 'F': 26066, 'j': 19535, 'æ': 8938, 'z': 6607, 'k': 5944, 'J': 5182, 'K': 2833, 'w': 2767, 'Y': 2480, 'Z': 1969, 'ν': 1809, 'W': 1672, 'τ': 1613, 'α': 1509, 'ο': 1445, 'ε': 1064, 'ι': 1042, 'ê': 882, 'ς': 782, 'ρ': 770, 'â': 743, 'π': 712, 'κ': 684, 'μ': 667, 'σ': 638, 'λ': 553, 'δ': 527, 'ὶ': 333, 'ω': 322, 'ί': 313, 'υ': 302, 'γ': 296, 'έ': 272, 'η': 264, 'ἐ': 263, 'ά': 254, 'θ': 247, 'φ': 233, 'ῶ': 233, 'ë': 228, 'ό': 226, 'œ': 225, 'ὰ': 222, 'ὲ': 216, 'à': 211, 'ù': 208, 'ἀ': 208, 'ò': 193, 'ὸ': 183, 'Æ': 159, 'ὐ': 159, 'í': 156, 'ῖ': 148, 'χ': 146, 'β': 146, 'ῦ': 140, 'Ç': 139, 'ύ': 139, 'ô': 138, 'î': 130, 'ἰ': 124, 'ὴ': 103, 'ή': 96, 'é': 91, 'û': 73, 'ξ': 71, 'ῆ': 68, 'Î': 64, 'Œ': 61, 'ἔ': 61, 'ζ': 58, 'ὺ': 57, 'Ï': 56, 'ὑ': 55, 'ώ': 55, 'Ò': 54, 'ὅ': 54, 'ἄ': 52, 'ῷ': 49, 'ὁ': 49, 'ï': 47, 'è': 43, 'ἂ': 42, 'ó': 40, 'ä': 40, 'ü': 39, 'Ì': 37, 'É': 37, 'ἴ': 36, 'ψ': 35, 'ἡ': 35, 'Þ': 35, 'Ü': 33, 'ῳ': 33, 'ἱ': 31, 'ὔ': 31, 'ὡ': 28, 'ἢ': 28, 'À': 27, 'Ī': 27, 'ᾶ': 27, 'ὼ': 27, 'ὀ': 26, 'ὕ': 25, 'ῃ': 23, 'ἶ': 21, 'ὖ': 21, 'ἑ': 20, 'Ω': 20, 'È': 19, 'ç': 19, 'Ἀ': 19, 'ã': 18, 'ª': 17, 'ì': 17, 'Â': 17, 'þ': 17, 'Í': 16, 'ÿ': 16, 'Ê': 15, 'ἕ': 15, 'ῇ': 14, 'ß': 13, 'ἤ': 13, 'ὄ': 12, 'ἅ': 12, 'ἁ': 11, 'Ú': 10, 'ñ': 10, 'š': 10, 'ῥ': 10, 'ἷ': 10, 'ὥ': 10, 'á': 9, 'ᾳ': 9, 'Δ': 9, 'Π': 9, 'Ô': 8, 'ἦ': 8, 'Κ': 8, 'Θ': 8, 'Û': 7, 'ἵ': 7, 'ἠ': 7, 'Φ': 7, 'ᾷ': 7, 'Τ': 7, 'ð': 7, 'Ã': 6, 'ﬁ': 6, 'ö': 6, 'Ù': 6, 'ﬂ': 6, 'Õ': 6, 'Å': 6, 'Ε': 6, 'ἧ': 6, 'ú': 5, 'ί': 5, 'ƒ': 5, 'Ὀ': 5, 'Μ': 5, 'ἥ': 5, 'ὠ': 5, 'Ἰ': 5, 'Σ': 5, 'Λ': 5, 'ῄ': 5, 'ό': 5, 'Á': 4, 'Ö': 4, 'å': 4, 'Ø': 4, 'Ð': 4, 'ὧ': 4, 'Ὁ': 4, 'Ä': 4, 'Ó': 3, 'ʻ': 3, 'ž': 3, 'Ž': 3, 'ῴ': 3, 'Ἡ': 3, 'ὗ': 3, 'Ῥ': 3, 'ᾄ': 3, 'ὃ': 3, 'Ο': 3, 'ἃ': 3, 'ὦ': 3, 'Ñ': 3, 'Ë': 3, 'Υ': 3, 'Α': 3, 'Ι': 3, 'ˇ': 2, 'Ÿ': 2, 'ý': 2, 'Ἑ': 2, 'ἳ': 2, 'ὤ': 2, 'Γ': 2, 'Ἔ': 2, 'Ý': 2, 'έ': 2, 'Ἤ': 2, 'ũ': 1, 'ę': 1, 'ˆ': 1, 'õ': 1, 'Ἠ': 1, 'ᾴ': 1, 'Ὅ': 1, 'ᾤ': 1, 'ᾅ': 1, 'ἣ': 1, 'Ἥ': 1, 'ϋ': 1, 'ᾧ': 1, 'ῂ': 1, 'ἆ': 1, 'Ὦ': 1, 'Ὥ': 1, 'Ἕ': 1, 'Ἐ': 1, 'ύ': 1, 'ά': 1, 'Β': 1, 'Η': 1, 'Ψ': 1})\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 293/293 [02:41<00:00,  6.91files/s]1%|          | 2/293 [00:05<15:47,  3.25s/files]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Perseus Latin Corpus: Number of characters: 22,570,613\n",
      "Perseus Latin character frequency distribution: Counter({'e': 2573626, 'i': 2540681, 'a': 1950909, 't': 1859409, 'u': 1826139, 's': 1713549, 'r': 1424121, 'n': 1384612, 'm': 1239489, 'o': 1230708, 'c': 856902, 'l': 695765, 'd': 680521, 'p': 593449, 'q': 322142, 'b': 314761, 'v': 258914, 'g': 232105, 'f': 208419, 'h': 168682, 'x': 104819, 'A': 35456, 'C': 33845, 'I': 32417, 'S': 29175, 'P': 25424, 'y': 22051, 'N': 17196, 'M': 17033, 'T': 16179, 'E': 15970, 'H': 13570, 'V': 13223, 'L': 12281, 'Q': 12187, 'D': 11757, 'R': 11549, 'G': 9311, 'O': 6775, 'F': 6029, 'B': 5992, 'X': 5248, 'U': 2926, 'é': 2180, 'í': 1803, 'ν': 1785, 'α': 1656, 'ο': 1615, 'j': 1518, 'á': 1477, 'σ': 1402, 'ó': 1286, 'ι': 1285, 'τ': 1281, 'z': 1267, 'ú': 1170, 'ε': 1152, 'ρ': 1015, 'π': 852, 'λ': 748, 'μ': 748, 'κ': 671, 'Z': 641, 'K': 573, 'δ': 501, 'υ': 460, 'η': 369, 'γ': 355, 'ί': 337, 'χ': 334, 'ς': 315, 'ω': 315, 'θ': 301, 'φ': 295, 'ἀ': 290, 'ë': 266, 'έ': 264, 'ό': 248, 'k': 241, 'ά': 226, 'ἐ': 209, 'ὸ': 151, 'ύ': 151, 'ὶ': 147, 'ῶ': 136, 'ὐ': 133, 'ↄ': 132, 'β': 129, 'ῖ': 111, 'ί': 110, 'ή': 109, 'ὰ': 101, 'w': 101, 'ό': 96, 'ῦ': 96, 'ἰ': 80, 'ὴ': 76, 'ώ': 76, 'ὑ': 74, 'ξ': 72, 'έ': 70, 'ἔ': 68, 'ἄ': 68, 'ῆ': 65, 'ζ': 62, 'Y': 60, 'ï': 51, 'ῳ': 47, 'û': 46, 'ὲ': 45, 'ά': 44, 'ψ': 38, 'Ↄ': 37, 'ύ': 34, 'ὁ': 34, 'ῃ': 33, 'ὺ': 32, 'î': 31, 'Ἀ': 29, 'ἴ': 28, 'ἱ': 26, 'а': 25, 'е': 24, 'ή': 23, 'ῤ': 23, 'Π': 21, 'J': 21, 'ʼ': 19, 'ἡ': 19, 'Κ': 19, 'ῷ': 18, 'ὀ': 18, 'ʽ': 18, 'ὅ': 18, 'ᾶ': 17, 'ᾳ': 17, 'Τ': 17, 'ὔ': 16, 'ώ': 14, 'ῇ': 14, 'ὄ': 14, 'ἶ': 13, 'W': 13, 'ὃ': 13, 'ἑ': 13, 'Α': 13, 'Μ': 13, 'Σ': 13, 'É': 13, 'Í': 12, 'о': 12, 'ē': 12, 'Δ': 10, 'ὕ': 10, 'Á': 10, 'ὼ': 9, 'ἠ': 9, 'Θ': 9, 'Λ': 9, 'ü': 9, 'ý': 9, 'ἷ': 8, 'Φ': 8, 'Ι': 8, 'ἕ': 7, 'ἅ': 7, 'ἤ': 7, 'Ω': 7, 'ὠ': 7, 'Ἡ': 7, 'ἢ': 7, 'Ν': 7, 'м': 7, 'Χ': 6, 'Ε': 6, 'ἁ': 6, 'ἥ': 5, 'Ἄ': 5, 'Ο': 5, 'ἂ': 5, 'ὗ': 4, 'Ὀ': 4, 'Ó': 4, 'ὢ': 4, 'ö': 4, 'ᾧ': 3, 'ᾷ': 3, 'Ῥ': 3, 'ὦ': 3, 'ἦ': 3, 'Γ': 3, 'ἵ': 3, 'ὖ': 3, 'à': 3, 'æ': 3, 'Β': 2, 'Ψ': 2, 'ὤ': 2, 'ὥ': 2, 'Ἐ': 2, 'Ш': 2, 'Ὄ': 1, 'ἲ': 1, 'Ἑ': 1, 'Ξ': 1, 'Ϡ': 1, 'ᾠ': 1, 'ἆ': 1, 'œ': 1, 'ϊ': 1, 'ΐ': 1, 'ἧ': 1, 'ἓ': 1, 'ὧ': 1, 'Ἰ': 1, 'Ἕ': 1, 'ὡ': 1, 'Ζ': 1, 'ἳ': 1, 'С': 1, 'Ἆ': 1, 'ϝ': 1, 'ῥ': 1, 'Å': 1, 'ϋ': 1, 'Ἤ': 1, 'Ὦ': 1, 'Î': 1, 'ů': 1, 'Υ': 1, 'Ρ': 1, 'Ἁ': 1})\n"
     ]
    }
   ],
   "source": [
    "latin_lib_char_freq = get_char_counts(latin_library_reader)\n",
    "print(f'Latin Library Corpus: Number of characters: {sum(latin_lib_char_freq.values()):,}')\n",
    "print('Latin Library character frequency distribution:', latin_lib_char_freq)\n",
    "#####################\n",
    "perseus_latin_char_freq = get_char_counts(perseus_latin_reader)\n",
    "print(f'Perseus Latin Corpus: Number of characters: {sum(perseus_latin_char_freq.values()):,}')\n",
    "print('Perseus Latin character frequency distribution:', perseus_latin_char_freq)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## We can see some obvious dirtiness:\n",
    "1. There are two characters 'œ' and 'æ' that represent dipthongs which should be split into two characters.\n",
    "1. There are some Greek letters in a Greek font, and some accented vowels foreign to Greek and Latin."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Let's consider these items the start of a bread crumb trail towards something bad and let's look further. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. OE and AE \n",
    "#### Classical Latin wrote the o and e separately (as has today again become the general practice), but the ligature was used by medieval and early modern writings, in part because the diphthongal sound had, by Late Latin, merged into the sound [e].\n",
    "#### https://en.wikipedia.org/wiki/%C5%92\n",
    "\n",
    "#### Æ (minuscule: æ) is a grapheme named æsc or ash, formed from the letters a and e, originally a ligature representing the Latin diphthong ae. It has been promoted to the full status of a letter in the alphabets of some languages, including Danish, Norwegian, Icelandic, and Faroese. \n",
    "#### https://en.wikipedia.org/wiki/%C3%86\n",
    "\n",
    "#### We have written an AEOEReplacer in the style of the CLTK JVReplacer and will use this to clean the corpora in other notebooks."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Greek Letters, Greek words; Foreign characters\n",
    "### Filtering out these Greek words and obvious characters is easy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "0it [00:00, ?it/s]136it [00:00, 1350.96it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "foreign_characters ['φ', 'α', 'ν', 'τ', 'σ', 'ί', 'ς', 'ἔ', 'λ', 'β', 'ε', 'π', 'ο', 'ρ', 'ύ', 'ά', 'κ', 'ὶ', 'μ', 'ῖ', 'ι', 'ή', 'δ', 'ἱ', 'ό', 'υ', 'ῳ', 'ῷ', 'χ', 'ἄ', 'ῃ', 'ἀ', 'ω', 'θ', 'ζ', 'ὐ', 'ὰ', 'η', 'γ', 'ὸ', 'ἷ', 'ἐ', 'ἶ', 'έ', 'ὺ', 'ὲ', 'ὴ', 'ῆ', 'ἴ', 'ὁ', 'ῶ', 'ῦ', 'ώ', 'ἰ', 'ῇ', 'ὀ', 'ξ', 'ʽ', 'ᾶ', 'ὃ', 'ʼ', 'ᾧ', 'ὑ', 'ἡ', 'ἑ', 'ὼ', 'ὗ', 'ᾷ', 'ᾳ', 'ἠ', 'ὅ', 'ἕ', 'ἅ', 'ὔ', 'ώ', 'ύ', 'ά', 'Ἀ', 'ί', 'ë', 'έ', 'ἥ', 'ό', 'Δ', 'ὕ', 'Κ', 'Α', 'Ὄ', 'Μ', 'Φ', 'Β', 'ή', 'Τ', 'ἲ', 'ῤ', 'Χ', 'Θ', 'Ψ', 'Λ', 'Ἑ', 'Ἄ', 'Ι', 'ψ', 'Π', 'ἤ', 'Ο', 'Σ', 'Ῥ', 'Ὀ', 'Ξ', 'Ϡ', 'Ε', 'Ω', 'ὄ', 'ὦ', 'ὤ', 'ᾠ', 'ὥ', 'ὠ', 'ἆ', 'ü', 'é', 'ↄ', 'ú', 'í', 'á', 'ó', 'Í', 'ý', 'œ', 'Ó', 'ϊ', 'ΐ', 'Ↄ', 'Ἡ', 'ἧ', 'É', 'ἂ', 'ὢ', 'ἦ', 'ἢ', 'ἁ', 'ἓ', 'Γ', 'ἵ', 'ὖ', 'ὧ', 'Ἐ', 'Ἰ', 'Ἕ', 'ὡ', 'Ν', 'Ζ', 'ἳ', 'à', 'ï', 'а', 'е', 'м', 'Ш', 'о', 'С', 'î', 'Á', 'Ἆ', 'ē', 'ϝ', 'ῥ', 'Å', 'ϋ', 'æ', 'Ἤ', 'Ὦ', 'Î', 'ů', 'Υ', 'Ρ', 'ö', 'û', 'Ἁ']\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "16455728it [15:05, 18170.19it/s] [00:00, 2421.71it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number just latin words: 16,450,347\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "foreign_characters = [ tmp for tmp in \n",
    "                        perseus_latin_char_freq.keys()\n",
    "                        if not re.match('[a-zA-Z]', tmp) ]\n",
    "print('foreign_characters', foreign_characters)\n",
    "\n",
    "just_latin_words = []\n",
    "latin_library_reader = get_corpus_reader(corpus_name='latin_text_latin_library', language='latin')\n",
    "for word in tqdm(latin_library_reader.words()):\n",
    "    if not re.match('[{}]'.format(''.join(foreign_characters)), word):\n",
    "        just_latin_words.append(word)\n",
    "\n",
    "print(f'Number just latin words: {len(just_latin_words):,}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### This is easily extended to \"Show me all the Greek in Cicero\", \"Show me Plautus without Greek\", etc.\n",
    "#### But other problems lurk: such as \n",
    "## Acceptable Characters may represent foreign words transliterated:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'kai' in just_latin_words # kai means 'and' in Greek"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### We will deal with the complex problem of detecting and correcting transliterated loadwords in another notebook"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Simple Heuristic for detecting improperly joined words\n",
    "### Almost always when a sequence of lower case letters is followed by a capital letter without a space intervening it is an indicator of improperly joined words."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      "  0%|          | 0/16450347 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "naviIndica\n",
      "quaeGoa\n",
      "seuMadagascar\n",
      "digitiIndici\n",
      "aureaMozambici\n",
      "anniDecembri\n",
      "quiCafres\n",
      "suntCafres\n",
      "osCafres\n",
      "estBazaini\n",
      "cantaturPraefatio\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 16450347/16450347 [00:24<00:00, 659738.78it/s]%|          | 114161/16450347 [00:00<00:28, 575202.76it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of camel cased words 516\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "camelCase_regex = '[a-z]+[A-Z][a-z]+'\n",
    "\n",
    "for word in just_latin_words[:10000]:\n",
    "    if re.match(camelCase_regex, word):\n",
    "        print (word)\n",
    "\n",
    "camel_case_count = 0\n",
    "\n",
    "for word in tqdm(just_latin_words):\n",
    "    if re.match(camelCase_regex, word):\n",
    "        camel_case_count +=1 \n",
    "\n",
    "print(f'Number of camel cased words {camel_case_count:,}')        \n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## camelCased words are easily corrected\n",
    "### e.g.:\n",
    "#### from cltk.util.matrix_corpus_fun import split_camel\n",
    "at: https://github.com/cltk/cltk/blob/master/cltk/utils/matrix_corpus_fun.py#L213"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Detecting Improperly Joined Words In General\n",
    "* Errors in data transmission, \n",
    "* Improperly OCRed text\n",
    "* botched formating (e.g. blindly joining lines and dropping newline characters that are also meant to encode word separations)\n",
    "\n",
    "#### Can all cause words to become improperly joined in a corpus.\n",
    "#### To help spot candidates of improperly joined words, we need to first examine some sample corpora to determine typical word lengths."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 2141/2141 [7:52:48<00:00,  1.33files/s]     /2141 [00:00<06:48,  5.24files/s]\n"
     ]
    }
   ],
   "source": [
    "latin_lib_lengths = get_word_lengths(latin_library_reader)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "top_label, side_label, bottom_label = ( 'Word Lengths', 'Number of words',\n",
    "                                       'Raw Word Lengths of Latin Library Corpus')\n",
    "fig = plt.figure(figsize=(7,7))\n",
    "plt.xlabel(top_label)\n",
    "plt.ylabel(side_label)\n",
    "plt.title(bottom_label)\n",
    "ax = fig.gca()\n",
    "ax.xaxis.set_major_locator(MaxNLocator(integer=True))\n",
    "data = list(latin_lib_lengths.items())\n",
    "data.sort(key=lambda k: k[0])\n",
    "indices, values = zip(*data)\n",
    "idx= bisect(indices, 28) \n",
    "# expert testimony, reasonably longest words:\n",
    "# undēquinquāgēsimōrum = 24 letters\n",
    "# conscrībillātūrōrum\n",
    "# percrēbrēscēbāminī\n",
    "indices, values = zip(*data[:idx]) # reasonably longest word in latin \n",
    "indices, values = zip(*data[:16]) # Nothing shows up on the graph beyond this range\n",
    "plt.bar(indices, values)\n",
    "plt.tight_layout(pad=0.5, w_pad=20, h_pad=0.5)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total words 13,137,256\n",
      "Word lengths: Counter({5: 1736309, 2: 1608422, 3: 1605589, 4: 1601042, 6: 1600574, 7: 1466665, 8: 1207092, 9: 882205, 10: 615554, 11: 371324, 12: 192327, 13: 94100, 1: 87346, 14: 41324, 15: 17094, 16: 6702, 17: 2162, 18: 575, 19: 354, 20: 289, 21: 86, 22: 45, 23: 38, 25: 12, 26: 7, 24: 6, 28: 6, 30: 1, 27: 1, 48: 1, 32: 1, 36: 1, 29: 1, 39: 1})\n",
      "Mean Latin word length: 5.74, standard deviation: 2.76\n",
      "98% cutoff 11 letters\n"
     ]
    }
   ],
   "source": [
    "total_words = sum(latin_lib_lengths.values())\n",
    "print(f'Total words {total_words:,}')\n",
    "print('Word lengths:', latin_lib_lengths)\n",
    "all_lens = [] \n",
    "for key in latin_lib_lengths:\n",
    "    all_lens += [key] * mycounter[key]\n",
    "mean_word_len, stdev_word_length = round(statistics.mean(all_lens), 3), round(statistics.stdev(all_lens), 3)\n",
    "print(f'Mean Latin word length: {round(mean_word_len, 2)}, standard deviation: {round(stdev_word_length, 2)}')\n",
    "print(f'98% cutoff {round(mean_word_len + (2* stdev_word_length))} letters')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 293/293 [02:50<00:00,  6.41files/s]     | 1/293 [00:04<21:21,  4.39s/files]\n"
     ]
    }
   ],
   "source": [
    "perseus_latin_lengths = get_word_lengths(perseus_latin_reader) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "top_label, side_label, bottom_label = ('Word Lengths', 'Number of words',\n",
    "                                       'Raw Word Lengths of Perseus Latin(CLTK)')\n",
    "fig = plt.figure(2, figsize=(7,7))\n",
    "ax = fig.gca()\n",
    "ax.xaxis.set_major_locator(MaxNLocator(integer=True))\n",
    "plt.xlabel(top_label)\n",
    "plt.ylabel(side_label)\n",
    "plt.title(bottom_label )\n",
    "data = list(perseus_latin_lengths.items())\n",
    "data.sort(key=lambda k: k[0])\n",
    "indices, values = zip(*data)\n",
    "idx = bisect(indices, 28) \n",
    "indices, values = zip(*data[:16]) # reasonably Longest word in latin \n",
    "# print(indices)\n",
    "# print(values)\n",
    "plt.bar(indices, values)\n",
    "plt.tight_layout(pad=0.5, w_pad=20, h_pad=0.5)\n",
    "plt.show();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total words 3,932,338\n",
      "Word Lengths Counter({5: 554740, 6: 524202, 4: 491842, 7: 456598, 3: 453098, 2: 449547, 8: 349433, 9: 242573, 10: 166776, 11: 97863, 12: 52377, 1: 33003, 13: 28622, 14: 15548, 15: 8358, 16: 4370, 17: 1942, 18: 864, 19: 287, 20: 146, 21: 64, 22: 41, 23: 16, 24: 10, 25: 7, 28: 3, 38: 2, 26: 2, 31: 1, 33: 1, 32: 1, 34: 1})\n",
      "Mean Latin word length: 5.74, standard deviation: 2.73\n",
      "98% cutoff 11 letters\n"
     ]
    }
   ],
   "source": [
    "total_words = sum(perseus_latin_lengths.values())\n",
    "print(f'Total words {total_words:,}')\n",
    "print('Word Lengths', perseus_latin_lengths)\n",
    "all_lens = [] \n",
    "for key in perseus_latin_lengths:\n",
    "    all_lens += [key] * mycounter[key]\n",
    "mean_word_len, stdev_word_length = round(statistics.mean(all_lens), 3), round(statistics.stdev(all_lens), 3)\n",
    "print(f'Mean Latin word length: {round(mean_word_len, 2)}, standard deviation: {round(stdev_word_length, 2)}')\n",
    "print(f'98% cutoff {round(mean_word_len + (2* stdev_word_length))} letters')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Based on these numbers we can see that the long words in the Latin corpora are exceptional, and we can take a reasonable stance that any word in a corpus 12 letters or longer is a candidate for examination to determine whether or not it's an improperly joined word. Let's take a closer look at their representations."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Breaking long words in two when possible, to do this we'll need to \n",
    "### Build a language model that will help us find word boundaries and vet candidates for un-merging, this is the task in another notebook `make_trie_language_model.ipynb`.\n",
    "### In another notebook, `detecting_and_correcting_loanwords.ipynb` we will take on the complex problem of transliterated loanwords."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## That's all for now folks!"
   ]
  }
 ],
 "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": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}