quality_of_tagging.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "quality_of_tagging.ipynb",
      "provenance": [],
      "authorship_tag": "ABX9TyP03//DeAAKtrLLOpBgQ1Ks",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/overfit-ir/persian-twitter-ner/blob/master/quality_of_tagging.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lW2HKpoPSQiB",
        "outputId": "f6bd2e75-f43c-4140-faeb-47e759fc39a4",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "!git clone https://github.com/overfit-ir/persian-twitter-ner.git"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Cloning into 'persian-twitter-ner'...\n",
            "remote: Enumerating objects: 99, done.\u001b[K\n",
            "remote: Counting objects: 100% (99/99), done.\u001b[K\n",
            "remote: Compressing objects: 100% (73/73), done.\u001b[K\n",
            "remote: Total 1906 (delta 51), reused 61 (delta 25), pack-reused 1807\u001b[K\n",
            "Receiving objects: 100% (1906/1906), 19.64 MiB | 18.74 MiB/s, done.\n",
            "Resolving deltas: 100% (1398/1398), done.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vFHtniDtmCZk"
      },
      "source": [
        "import os"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oK5pc5NmS_9y"
      },
      "source": [
        "tweets1 = []\n",
        "tweet = []\n",
        "with open('persian-twitter-ner/twitter_data/kappa_test_data/test_data1.txt', 'r') as file:\n",
        "    for line_number, line in enumerate(file.readlines()):\n",
        "        if line != '\\n':\n",
        "            tweet.append(line.replace('\\n', '').split('\\t'))\n",
        "        else:\n",
        "            tweets1.append(tweet)\n",
        "            tweet = []"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2pph4Vnal8T5"
      },
      "source": [
        "tweets2 = []\n",
        "tweet = []\n",
        "with open('persian-twitter-ner/twitter_data/kappa_test_data/test_data12.txt', 'r') as file:\n",
        "    for line_number, line in enumerate(file.readlines()):\n",
        "        if line != '\\n':\n",
        "            tweet.append(line.replace('\\n', '').split('\\t'))\n",
        "        else:\n",
        "            tweets2.append(tweet)\n",
        "            tweet = []"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "awRNPvWomQJA"
      },
      "source": [
        "tags1 = [x[1] for tweet in tweets1 for x in tweet ]\n",
        "tags2 = [x[1] for tweet in tweets2 for x in tweet ]"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hozBRLlJmu8z"
      },
      "source": [
        "diff = [0 if x==y else 1 for x,y in zip(tags1, tags2)]"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u2u77lNZnC4t",
        "outputId": "cdfc4bfe-b986-4903-a574-d2155e7db7c8",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "sum(diff)"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "76"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "02_iS8wQnEMH",
        "outputId": "a18b42aa-9522-4794-d5d5-37d40c920b2a",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "len([x  for x in tags1 if x != 'O'])"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "983"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-QA9X2DKnJLm",
        "outputId": "fc15fe88-665c-45bf-ad18-175ea997a771",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "len(tags1)"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "9674"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uLQFXE8ytq67"
      },
      "source": [
        "### KAPPA"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9jtOES9Kucky"
      },
      "source": [
        "from sklearn.metrics import confusion_matrix, cohen_kappa_score\n",
        "import pandas as pd"
      ],
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c2NnaEdavIiR",
        "outputId": "83a48bd7-474a-4972-8eb2-b659d978b4e0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 826
        }
      },
      "source": [
        "import seaborn as sn\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "array = confusion_matrix(tags1, tags2,['B-PER', 'I-PER', 'B-ORG', 'B-LOC', 'I-ORG', 'B-NAT', 'B-EVE',\n",
        "       'B-POG', 'I-NAT', 'I-EVE', 'I-LOC', 'I-POG'])\n",
        "df_cm = pd.DataFrame(array, index = ['B-PER', 'I-PER', 'B-ORG', 'B-LOC', 'I-ORG', 'B-NAT', 'B-EVE',\n",
        "       'B-POG', 'I-NAT', 'I-EVE', 'I-LOC', 'I-POG'],\n",
        "                  columns = ['B-PER', 'I-PER', 'B-ORG', 'B-LOC', 'I-ORG', 'B-NAT', 'B-EVE',\n",
        "       'B-POG', 'I-NAT', 'I-EVE', 'I-LOC', 'I-POG'])\n",
        "plt.figure(figsize = (20,14))\n",
        "sn.heatmap(df_cm, annot=True, cmap='Blues')"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f5cbfdea410>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x1008 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BzEfyggGueBA",
        "outputId": "0c7a8ada-44c6-414f-aa5a-aa4884886f63",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "cohen_kappa_score(tags1, tags2, ['B-PER', 'I-PER', 'B-ORG', 'B-LOC', 'I-ORG', 'B-NAT', 'B-EVE',\n",
        "       'B-POG', 'I-NAT', 'I-EVE', 'I-LOC', 'I-POG'])"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.984105343759293"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 23
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iviv8KiNyXHD",
        "outputId": "dd9e8318-ced3-4ef7-dd3a-2881f8b3bd1b",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "cohen_kappa_score(tags1, tags2, ['B-PER', 'I-PER', 'B-ORG', 'O', 'B-LOC', 'I-ORG', 'B-NAT', 'B-EVE',\n",
        "       'B-POG', 'I-NAT', 'I-EVE', 'I-LOC', 'I-POG'])"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.9577135146174927"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tLGKKcWGyeQm"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}