Topic_Modeling_LSA.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Topic Modeling-LSA.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "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"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/diem-ai/topic-modeling/blob/master/Topic_Modeling_LSA.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f28PG4o7x4SB",
        "colab_type": "text"
      },
      "source": [
        "#### Introduction\n",
        "\n",
        "<p>In another notebook Topic Modeling-LDA.ipynb, we examine the idea of latent spaces and how we\n",
        "could use Latent Dirichlet Allocation to create a topic space. LDA is not the only method to create latent spaces. In this notebook, we are going to use  Latent Semantic Analysis (LSA) to accomplish the same task. LSA is another mathematical technique to decompose a matrix into sub-matrices</p>\n",
        "\n",
        "#### Project tasks:\n",
        "- Cleaning the dataset & Lemmatization (done in notebook model_preparation.ipynb )\n",
        "- Creat a dictionay from processed data (done in notebook model_preparation.ipynb)\n",
        "- Create Corpus and LSA Model with bag of words\n",
        "- Create Coprpus and LSA with TF-IDF\n",
        "- Test 2 models with unseen data\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sC9PMZHNeVTx",
        "colab_type": "text"
      },
      "source": [
        "#### Google Colab Setup"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Olt7Xq5byPKU",
        "colab_type": "code",
        "outputId": "7016f2ee-3300-4a17-e3b7-dbcbbb809129",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "from google.colab import drive\n",
        "# This will prompt for authorization.\n",
        "# authorization code: 4/OwErfUj6QceGXhIGx_RWv0MKclb9rilw8UsJnZqFbSez-QS8zQ399JU\n",
        "drive.mount('/content/drive')\n"
      ],
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "q-X20vwCyxDa",
        "colab_type": "code",
        "outputId": "8c5b9a17-aeca-41ab-977a-a196ec6d3e38",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 213
        }
      },
      "source": [
        "!pip install PyDrive\n",
        "\n",
        "from pydrive.auth import GoogleAuth\n",
        "from pydrive.drive import GoogleDrive\n",
        "from google.colab import auth\n",
        "from oauth2client.client import GoogleCredentials\n",
        "\n",
        "auth.authenticate_user()\n",
        "gauth = GoogleAuth()\n",
        "gauth.credentials = GoogleCredentials.get_application_default()\n",
        "drive = GoogleDrive(gauth)\n",
        "\n"
      ],
      "execution_count": 47,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: PyDrive in /usr/local/lib/python3.6/dist-packages (1.3.1)\n",
            "Requirement already satisfied: google-api-python-client>=1.2 in /usr/local/lib/python3.6/dist-packages (from PyDrive) (1.6.7)\n",
            "Requirement already satisfied: oauth2client>=4.0.0 in /usr/local/lib/python3.6/dist-packages (from PyDrive) (4.1.3)\n",
            "Requirement already satisfied: PyYAML>=3.0 in /usr/local/lib/python3.6/dist-packages (from PyDrive) (3.13)\n",
            "Requirement already satisfied: httplib2<1dev,>=0.9.2 in /usr/local/lib/python3.6/dist-packages (from google-api-python-client>=1.2->PyDrive) (0.11.3)\n",
            "Requirement already satisfied: uritemplate<4dev,>=3.0.0 in /usr/local/lib/python3.6/dist-packages (from google-api-python-client>=1.2->PyDrive) (3.0.0)\n",
            "Requirement already satisfied: six<2dev,>=1.6.1 in /usr/local/lib/python3.6/dist-packages (from google-api-python-client>=1.2->PyDrive) (1.12.0)\n",
            "Requirement already satisfied: rsa>=3.1.4 in /usr/local/lib/python3.6/dist-packages (from oauth2client>=4.0.0->PyDrive) (4.0)\n",
            "Requirement already satisfied: pyasn1>=0.1.7 in /usr/local/lib/python3.6/dist-packages (from oauth2client>=4.0.0->PyDrive) (0.4.5)\n",
            "Requirement already satisfied: pyasn1-modules>=0.0.5 in /usr/local/lib/python3.6/dist-packages (from oauth2client>=4.0.0->PyDrive) (0.2.5)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FeNsxd-MgJXf",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#import accessory_functions.py from Colab\n",
        "#https://drive.google.com/open?id=1S7URZIBq4zMh5QWv0qXPHv4ixhgHWN_y\n",
        "\n",
        "my_module = drive.CreateFile({'id':'1S7URZIBq4zMh5QWv0qXPHv4ixhgHWN_y'})\n",
        "my_module.GetContentFile('accessory_functions.py')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YXVkDzd8x4SG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import numpy as np\n",
        "import string\n",
        "import pandas as pd\n",
        "import unidecode\n",
        "\n",
        "import gensim\n",
        "import gensim.corpora as corpora\n",
        "from gensim.utils import simple_preprocess\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "from accessory_functions import read_pickle_file\n",
        "\n",
        "\n",
        "%matplotlib inline\n",
        "# Make all my plots 538 Style\n",
        "plt.style.use('fivethirtyeight')\n",
        "\n",
        "import warnings\n",
        "warnings.simplefilter('ignore')\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dJhgmYj30bym",
        "colab_type": "text"
      },
      "source": [
        "<p>Data Path & Model parameters</p>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "i8qQjhJd0ZdS",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "datapath = '/content/drive/My Drive/data/'\n",
        "n_topics = 50\n",
        "iterations = 50"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lgx_Dhqm0NaK",
        "colab_type": "text"
      },
      "source": [
        "<p> Loading data</p>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JQSpF8_J0SLr",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "processed_docs = read_pickle_file(datapath + 'processed_docs.pkl')\n",
        "bow = read_pickle_file(datapath + 'bow.pkl')\n",
        "tfidf = read_pickle_file(datapath + 'tfidf.pkl')\n",
        "dictionary = read_pickle_file(datapath + 'dictionary.pkl')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4fsS1J-X095M",
        "colab_type": "code",
        "outputId": "89771626-205f-40c4-badf-2ad82ba83940",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "print(processed_docs[1])\n",
        "\n"
      ],
      "execution_count": 52,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "['trade', 'flow', 'traditional', 'economic', 'measure', 'reveal', 'true', 'cost', 'tariff', 'washington', 'friday', 'hike', 'duty', 'billion', 'chinese', 'good', 'side', 'intimate', 'survive', 'blow', 'may', 'slightly', 'dent', 'gdp', 'shift', 'supply', 'chain', 'though', 'show', 'extent', 'long', 'term', 'loss']\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aV0eGmD71s7f",
        "colab_type": "code",
        "outputId": "857cc0a1-f80c-4e38-bdae-fed0fb42f252",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 634
        }
      },
      "source": [
        "doc = bow[1]\n",
        "\n",
        "print(doc)\n",
        "\n",
        "[(dictionary[id], count) for id, count in doc]"
      ],
      "execution_count": 53,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[(4, 1), (10, 1), (33, 1), (38, 1), (39, 1), (40, 1), (41, 1), (42, 1), (43, 1), (44, 1), (45, 1), (46, 1), (47, 1), (48, 1), (49, 1), (50, 1), (51, 1), (52, 1), (53, 1), (54, 1), (55, 1), (56, 1), (57, 1), (58, 1), (59, 1), (60, 1), (61, 1), (62, 1), (63, 1), (64, 1), (65, 1), (66, 1), (67, 1)]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('billion', 1),\n",
              " ('friday', 1),\n",
              " ('trade', 1),\n",
              " ('blow', 1),\n",
              " ('chain', 1),\n",
              " ('chinese', 1),\n",
              " ('cost', 1),\n",
              " ('dent', 1),\n",
              " ('duty', 1),\n",
              " ('economic', 1),\n",
              " ('extent', 1),\n",
              " ('flow', 1),\n",
              " ('gdp', 1),\n",
              " ('good', 1),\n",
              " ('hike', 1),\n",
              " ('intimate', 1),\n",
              " ('long', 1),\n",
              " ('loss', 1),\n",
              " ('may', 1),\n",
              " ('measure', 1),\n",
              " ('reveal', 1),\n",
              " ('shift', 1),\n",
              " ('show', 1),\n",
              " ('side', 1),\n",
              " ('slightly', 1),\n",
              " ('supply', 1),\n",
              " ('survive', 1),\n",
              " ('tariff', 1),\n",
              " ('term', 1),\n",
              " ('though', 1),\n",
              " ('traditional', 1),\n",
              " ('true', 1),\n",
              " ('washington', 1)]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 53
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PUKbys3g17gG",
        "colab_type": "code",
        "outputId": "55189cc7-2dd6-444e-d995-960ede31404e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 634
        }
      },
      "source": [
        "doc = tfidf[1]\n",
        "\n",
        "print(doc)\n",
        "\n",
        "[(dictionary[id], freq) for id, freq in doc]"
      ],
      "execution_count": 54,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[(4, 0.05775843375562911), (10, 0.12310661992926017), (33, 0.10439212162929286), (38, 0.1961303786255056), (39, 0.17310433437273764), (40, 0.10148900465133513), (41, 0.1199009983425769), (42, 0.23725627027918395), (43, 0.21614153934185534), (44, 0.1071217734433688), (45, 0.28633219743654287), (46, 0.1961303786255056), (47, 0.1961303786255056), (48, 0.1091949636509986), (49, 0.1923966831387611), (50, 0.28633219743654287), (51, 0.11962217757715381), (52, 0.1524456522665324), (53, 0.06964385147961394), (54, 0.17420790459371643), (55, 0.19813974994500413), (56, 0.1765129312952131), (57, 0.12161692894792232), (58, 0.1829493640280444), (59, 0.21010281715361415), (60, 0.18157469132600734), (61, 0.21010281715361415), (62, 0.16023546606560787), (63, 0.13550592942070178), (64, 0.11798968502713748), (65, 0.19813974994500413), (66, 0.20487183805983358), (67, 0.14595088124716785)]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('billion', 0.05775843375562911),\n",
              " ('friday', 0.12310661992926017),\n",
              " ('trade', 0.10439212162929286),\n",
              " ('blow', 0.1961303786255056),\n",
              " ('chain', 0.17310433437273764),\n",
              " ('chinese', 0.10148900465133513),\n",
              " ('cost', 0.1199009983425769),\n",
              " ('dent', 0.23725627027918395),\n",
              " ('duty', 0.21614153934185534),\n",
              " ('economic', 0.1071217734433688),\n",
              " ('extent', 0.28633219743654287),\n",
              " ('flow', 0.1961303786255056),\n",
              " ('gdp', 0.1961303786255056),\n",
              " ('good', 0.1091949636509986),\n",
              " ('hike', 0.1923966831387611),\n",
              " ('intimate', 0.28633219743654287),\n",
              " ('long', 0.11962217757715381),\n",
              " ('loss', 0.1524456522665324),\n",
              " ('may', 0.06964385147961394),\n",
              " ('measure', 0.17420790459371643),\n",
              " ('reveal', 0.19813974994500413),\n",
              " ('shift', 0.1765129312952131),\n",
              " ('show', 0.12161692894792232),\n",
              " ('side', 0.1829493640280444),\n",
              " ('slightly', 0.21010281715361415),\n",
              " ('supply', 0.18157469132600734),\n",
              " ('survive', 0.21010281715361415),\n",
              " ('tariff', 0.16023546606560787),\n",
              " ('term', 0.13550592942070178),\n",
              " ('though', 0.11798968502713748),\n",
              " ('traditional', 0.19813974994500413),\n",
              " ('true', 0.20487183805983358),\n",
              " ('washington', 0.14595088124716785)]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 54
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JjI_ypGK2oXU",
        "colab_type": "text"
      },
      "source": [
        "<p>LSA with Bag-of-Word</p>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wuXDkkcf21Pp",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "lsi_bow = gensim.models.LsiModel(bow      \n",
        "                                 , num_topics=n_topics\n",
        "                                       , id2word=dictionary)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FwyCgiRS3a45",
        "colab_type": "text"
      },
      "source": [
        "<p>Print top 10 popular topics</p>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jfr-0QJd3OSq",
        "colab_type": "code",
        "outputId": "6da661bb-c291-41b5-d83d-975e428f0806",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 213
        }
      },
      "source": [
        "topics = lsi_bow.print_topics(num_topics=5, num_words=10)\n",
        "\n",
        "for idx, topic in topics:\n",
        "  print(\"topic: {}\\n {}\".format(idx, topic))\n",
        "  \n",
        "#[print(\"topic: {}\\n {}\".format(idx, topic)) for idx, topic in topics]"
      ],
      "execution_count": 56,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "topic: 0\n",
            " 0.297*\"billion\" + 0.213*\"president\" + 0.211*\"trump\" + 0.187*\"may\" + 0.174*\"year\" + 0.163*\"donald\" + 0.156*\"new\" + 0.155*\"could\" + 0.155*\"company\" + 0.146*\"bank\"\n",
            "topic: 1\n",
            " -0.516*\"trump\" + -0.478*\"president\" + -0.408*\"donald\" + 0.280*\"billion\" + -0.149*\"say\" + 0.123*\"company\" + 0.102*\"year\" + 0.099*\"investor\" + -0.096*\"house\" + 0.082*\"bank\"\n",
            "topic: 2\n",
            " -0.516*\"billion\" + 0.401*\"bank\" + 0.264*\"may\" + 0.211*\"european\" + -0.172*\"company\" + 0.144*\"union\" + 0.144*\"britain\" + 0.143*\"minister\" + -0.135*\"trump\" + 0.122*\"prime\"\n",
            "topic: 3\n",
            " -0.540*\"bank\" + 0.435*\"may\" + -0.190*\"year\" + 0.171*\"minister\" + -0.166*\"chief\" + -0.155*\"executive\" + 0.152*\"prime\" + 0.151*\"union\" + 0.150*\"european\" + 0.143*\"britain\"\n",
            "topic: 4\n",
            " 0.515*\"china\" + 0.249*\"market\" + -0.229*\"chief\" + -0.217*\"executive\" + 0.211*\"could\" + -0.209*\"billion\" + 0.180*\"chinese\" + 0.166*\"beijing\" + -0.164*\"may\" + 0.161*\"state\"\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "btvJgf_0x4Up",
        "colab_type": "text"
      },
      "source": [
        "#### LSA model with TF-IDF\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OFQqDYq3ySFC",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "lsi_tfidf = gensim.models.LsiModel(tfidf\n",
        "                                       , num_topics=n_topics\n",
        "                                       , id2word=dictionary)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lgPWLBarx4VM",
        "colab_type": "text"
      },
      "source": [
        "<p>Print top 5 topics</p>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7y6wB93Xx4VN",
        "colab_type": "code",
        "outputId": "b2896c51-ef88-42cf-a1d4-0659bcbca451",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 300
        }
      },
      "source": [
        "#pprint(lda_model.print_topics())\n",
        "for idx, topic in lsi_tfidf.print_topics(num_topics=5, num_words=10):\n",
        "    print('\\nTopic: {}\\nWords: {}'.format(idx, topic))\n"
      ],
      "execution_count": 58,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Topic: 0\n",
            "Words: 0.148*\"trump\" + 0.143*\"president\" + 0.126*\"donald\" + 0.124*\"billion\" + 0.112*\"may\" + 0.111*\"bank\" + 0.108*\"say\" + 0.106*\"could\" + 0.102*\"year\" + 0.102*\"china\"\n",
            "\n",
            "Topic: 1\n",
            "Words: -0.347*\"trump\" + -0.312*\"donald\" + -0.308*\"president\" + -0.173*\"say\" + -0.149*\"verified\" + -0.144*\"twitter\" + -0.144*\"house\" + -0.141*\"statement\" + -0.120*\"post\" + -0.119*\"account\"\n",
            "\n",
            "Topic: 2\n",
            "Words: -0.376*\"verified\" + -0.341*\"twitter\" + -0.339*\"statement\" + -0.317*\"account\" + -0.294*\"post\" + -0.265*\"follow\" + -0.182*\"personal\" + 0.108*\"election\" + 0.091*\"minister\" + 0.081*\"house\"\n",
            "\n",
            "Topic: 3\n",
            "Words: 0.236*\"european\" + 0.221*\"minister\" + 0.216*\"union\" + 0.210*\"brexit\" + 0.207*\"britain\" + 0.200*\"prime\" + 0.192*\"theresa\" + 0.156*\"verified\" + 0.144*\"may\" + 0.139*\"statement\"\n",
            "\n",
            "Topic: 4\n",
            "Words: 0.246*\"china\" + -0.215*\"fox\" + -0.186*\"murdoch\" + 0.157*\"trade\" + -0.152*\"disney\" + -0.150*\"twenty\" + -0.142*\"century\" + -0.142*\"sky\" + -0.136*\"rupert\" + 0.132*\"chinese\"\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZyVpFYFlziV8",
        "colab_type": "text"
      },
      "source": [
        "<p>Test model with unseen data</p>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u-FzWk9jzno5",
        "colab_type": "code",
        "outputId": "7658ad7a-570f-43fd-efc2-8b52da3caa09",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 213
        }
      },
      "source": [
        "def sortbyvalue(item):\n",
        "  \"\"\"\n",
        "  @item is a tuple (key,val)\n",
        "  \"\"\"\n",
        "  return item[1]\n",
        "\n",
        "\n",
        "\n",
        "text = 'Uber Technologies lackluster stock-market debut is a warning for other tech unicorns'\n",
        "\n",
        "unseen_doc = dictionary.doc2bow(text.split())\n",
        "\n",
        "vector = lsi_bow[unseen_doc]\n",
        "# sort vector descending by score\n",
        "vector.sort(key=sortbyvalue, reverse=True)\n",
        "\n",
        "#print top 5 topics\n",
        "for idx, score in vector[:5]:\n",
        "  print(\"topic:{}  score:{} \\n {}\".format(idx, score, lsi_bow.print_topic(topicno=idx, topn=10)))\n",
        "\n"
      ],
      "execution_count": 59,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "topic:7  score:0.08944203321384314 \n",
            " -0.402*\"year\" + 0.384*\"market\" + -0.289*\"state\" + -0.265*\"china\" + 0.263*\"investor\" + 0.224*\"new\" + 0.174*\"may\" + -0.171*\"united\" + 0.134*\"big\" + 0.128*\"share\"\n",
            "topic:23  score:0.08572696990351683 \n",
            " 0.236*\"would\" + -0.213*\"market\" + 0.201*\"economic\" + 0.198*\"trade\" + -0.167*\"deal\" + 0.167*\"public\" + -0.160*\"business\" + 0.159*\"percent\" + -0.152*\"election\" + -0.149*\"like\"\n",
            "topic:4  score:0.06558887598748853 \n",
            " 0.515*\"china\" + 0.249*\"market\" + -0.229*\"chief\" + -0.217*\"executive\" + 0.211*\"could\" + -0.209*\"billion\" + 0.180*\"chinese\" + 0.166*\"beijing\" + -0.164*\"may\" + 0.161*\"state\"\n",
            "topic:21  score:0.06183928438486232 \n",
            " 0.445*\"would\" + -0.261*\"percent\" + 0.239*\"investor\" + 0.214*\"like\" + -0.149*\"price\" + 0.142*\"chinese\" + -0.140*\"say\" + 0.138*\"bank\" + 0.137*\"hong\" + 0.137*\"kong\"\n",
            "topic:49  score:0.05487636452245267 \n",
            " -0.213*\"percent\" + 0.203*\"government\" + -0.197*\"court\" + 0.160*\"price\" + -0.159*\"tuesday\" + -0.159*\"well\" + -0.143*\"group\" + 0.142*\"look\" + 0.140*\"giant\" + 0.133*\"firm\"\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rfjSPpiY299P",
        "colab_type": "code",
        "outputId": "81be037c-cb8e-47b8-ccb7-7713aaeff760",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 213
        }
      },
      "source": [
        "text = 'Uber Technologies lackluster stock-market debut is a warning for other tech unicorns'\n",
        "\n",
        "unseen_doc = dictionary.doc2bow(text.split())\n",
        "\n",
        "vector = lsi_tfidf[unseen_doc]\n",
        "\n",
        "vector.sort(key = sortbyvalue)\n",
        "\n",
        "for idx, score in vector[:5]:\n",
        "  print(\"topic:{}  score:{} \\n {}\".format(idx, score, lsi_tfidf.print_topic(topicno=idx, topn=10)))\n"
      ],
      "execution_count": 60,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "topic:14  score:-0.07928652879263831 \n",
            " -0.215*\"hong\" + -0.215*\"kong\" + -0.152*\"house\" + -0.143*\"tesla\" + -0.142*\"representative\" + -0.137*\"musk\" + 0.120*\"percent\" + -0.111*\"public\" + -0.107*\"elon\" + 0.106*\"quarter\"\n",
            "topic:7  score:-0.07058474855568372 \n",
            " -0.220*\"court\" + -0.192*\"supreme\" + 0.166*\"fox\" + -0.166*\"kavanaugh\" + 0.164*\"mueller\" + -0.161*\"brett\" + 0.159*\"robert\" + 0.155*\"counsel\" + 0.154*\"bank\" + 0.149*\"murdoch\"\n",
            "topic:3  score:-0.07052834479840216 \n",
            " 0.236*\"european\" + 0.221*\"minister\" + 0.216*\"union\" + 0.210*\"brexit\" + 0.207*\"britain\" + 0.200*\"prime\" + 0.192*\"theresa\" + 0.156*\"verified\" + 0.144*\"may\" + 0.139*\"statement\"\n",
            "topic:11  score:-0.06236072498176796 \n",
            " 0.259*\"saudi\" + -0.208*\"house\" + 0.153*\"arabia\" + -0.134*\"white\" + 0.119*\"prince\" + -0.118*\"representative\" + 0.117*\"oil\" + 0.117*\"facebook\" + 0.114*\"court\" + -0.113*\"tax\"\n",
            "topic:28  score:-0.060062131617195424 \n",
            " 0.194*\"white\" + -0.149*\"market\" + -0.141*\"amazon\" + 0.121*\"house\" + 0.118*\"euro\" + 0.114*\"percent\" + -0.111*\"presidential\" + -0.103*\"republican\" + -0.101*\"korea\" + 0.098*\"italy\"\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JYKtDBlS3ULv",
        "colab_type": "text"
      },
      "source": [
        "<p>With the same text, both models give the different results</p>"
      ]
    }
  ]
}