{
  "id": 79112,
  "title": "Can we use googletrans？",
  "url": "/competitions/quora-insincere-questions-classification/discussion/79112",
  "author_name": "",
  "post_date": "2019-01-31T02:55:25.232424200Z",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Looking at those augmentation techniques, one way we can augment the text is to make two translations.  I'm about to try googletrans. But I'm not sure if we can use it, or to what extend if we can. \n1. Use a separate kernel to generate text data from the data in this competition, then add into the submission kernel.\n2. Generate and use it in the same kernel if I can make it in 2h.\n3. Not allow to use it at all.</p>\n\n<p>Which can be the case?</p>",
  "messages": [
    {
      "id": "463997",
      "postDate": "01/31/2019 02:55:25",
      "content": "<p>Looking at those augmentation techniques, one way we can augment the text is to make two translations.  I'm about to try googletrans. But I'm not sure if we can use it, or to what extend if we can. \n1. Use a separate kernel to generate text data from the data in this competition, then add into the submission kernel.\n2. Generate and use it in the same kernel if I can make it in 2h.\n3. Not allow to use it at all.</p>\n\n<p>Which can be the case?</p>",
      "rawMarkdown": "Looking at those augmentation techniques, one way we can augment the text is to make two translations.  I'm about to try googletrans. But I'm not sure if we can use it, or to what extend if we can. \n1. Use a separate kernel to generate text data from the data in this competition, then add into the submission kernel.\n2. Generate and use it in the same kernel if I can make it in 2h.\n3. Not allow to use it at all.\n\nWhich can be the case?",
      "votes": null
    },
    {
      "id": "464048",
      "postDate": "01/31/2019 05:53:05",
      "content": "<ol>\n<li>No internet access</li>\n<li>No external data allowed (you can't upload extra file)</li>\n<li>2 hours</li>\n</ol>\n\n<p>That's pretty much the restriction. So if you can generate text data locally and compress it as string/dictionary/json/whatever you would be fine. It won't make much different than the hottest pre-processing kernel if you can manually define a variable containing the text data.</p>\n\n<p>BUT, that's playing on the edge.</p>",
      "rawMarkdown": "1. No internet access\n2. No external data allowed (you can't upload extra file)\n3. 2 hours\n\nThat's pretty much the restriction. So if you can generate text data locally and compress it as string/dictionary/json/whatever you would be fine. It won't make much different than the hottest pre-processing kernel if you can manually define a variable containing the text data.\n\nBUT, that's playing on the edge.",
      "votes": null
    },
    {
      "id": "464049",
      "postDate": "01/31/2019 05:55:10",
      "content": "<p>2</p>",
      "rawMarkdown": "2",
      "votes": null
    },
    {
      "id": "464533",
      "postDate": "02/01/2019 03:31:01",
      "content": "<p>Googletrans relies on translate.google.com so you can't use it. You'll get an html error. As far as I know, all of the libraries that have translation make use of google translate. </p>",
      "rawMarkdown": "Googletrans relies on translate.google.com so you can't use it. You'll get an html error. As far as I know, all of the libraries that have translation make use of google translate.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 464048,
      "author_name": "jihangz",
      "author_url": "",
      "post_date": "01/31/2019 05:53:05",
      "content": "<ol>\n<li>No internet access</li>\n<li>No external data allowed (you can't upload extra file)</li>\n<li>2 hours</li>\n</ol>\n\n<p>That's pretty much the restriction. So if you can generate text data locally and compress it as string/dictionary/json/whatever you would be fine. It won't make much different than the hottest pre-processing kernel if you can manually define a variable containing the text data.</p>\n\n<p>BUT, that's playing on the edge.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 464049,
      "author_name": "xiaobai1123q",
      "author_url": "",
      "post_date": "01/31/2019 05:55:10",
      "content": "<p>2</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 464533,
      "author_name": "julius6",
      "author_url": "",
      "post_date": "02/01/2019 03:31:01",
      "content": "<p>Googletrans relies on translate.google.com so you can't use it. You'll get an html error. As far as I know, all of the libraries that have translation make use of google translate. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "463997": "Looking at those augmentation techniques, one way we can augment the text is to make two translations.  I'm about to try googletrans. But I'm not sure if we can use it, or to what extend if we can. \n1. Use a separate kernel to generate text data from the data in this competition, then add into the submission kernel.\n2. Generate and use it in the same kernel if I can make it in 2h.\n3. Not allow to use it at all.\n\nWhich can be the case?",
    "464048": "1. No internet access\n2. No external data allowed (you can't upload extra file)\n3. 2 hours\n\nThat's pretty much the restriction. So if you can generate text data locally and compress it as string/dictionary/json/whatever you would be fine. It won't make much different than the hottest pre-processing kernel if you can manually define a variable containing the text data.\n\nBUT, that's playing on the edge.",
    "464049": "2",
    "464533": "Googletrans relies on translate.google.com so you can't use it. You'll get an html error. As far as I know, all of the libraries that have translation make use of google translate."
  },
  "source": "meta"
}