{
  "id": 76852,
  "title": "Why don't we use stemming while using DL models?",
  "url": "/competitions/quora-insincere-questions-classification/discussion/76852",
  "author_name": "",
  "post_date": "2019-01-07T09:41:42.604537600Z",
  "votes": 2,
  "comment_count": 6,
  "views": 0,
  "content": "<p>Just as a curiosity. why don't we think of using stemming as preprocessing while using RNN based models? Is there some reason for it? </p>",
  "messages": [
    {
      "id": "451572",
      "postDate": "01/07/2019 09:41:42",
      "content": "<p>Just as a curiosity. why don't we think of using stemming as preprocessing while using RNN based models? Is there some reason for it? </p>",
      "rawMarkdown": "Just as a curiosity. why don't we think of using stemming as preprocessing while using RNN based models? Is there some reason for it?",
      "votes": null
    },
    {
      "id": "451574",
      "postDate": "01/07/2019 09:43:46",
      "content": "<p>almost everyone is using word embeddings and when you stem, you wont be able to find the embedding for the stemmed word. thats what i think the reason is :)</p>",
      "rawMarkdown": "almost everyone is using word embeddings and when you stem, you wont be able to find the embedding for the stemmed word. thats what i think the reason is :)",
      "votes": null
    },
    {
      "id": "451581",
      "postDate": "01/07/2019 09:51:06",
      "content": "<p>I think it's up to you, you can use it :)</p>",
      "rawMarkdown": "I think it's up to you, you can use it :)",
      "votes": null
    },
    {
      "id": "451587",
      "postDate": "01/07/2019 09:59:13",
      "content": "<p>That seems like the reason but what if we stem the word in the embedding too while processing the embedding?</p>",
      "rawMarkdown": "That seems like the reason but what if we stem the word in the embedding too while processing the embedding?",
      "votes": null
    },
    {
      "id": "451633",
      "postDate": "01/07/2019 12:00:15",
      "content": "<p>Try it out? In the end it doesn't matter much if you find the respective embedding.</p>",
      "rawMarkdown": "Try it out? In the end it doesn't matter much if you find the respective embedding.",
      "votes": null
    },
    {
      "id": "1005782",
      "postDate": "09/10/2020 18:33:36",
      "content": "<p>Did you get the answer <a href=\"https://www.kaggle.com/mlwhiz\" target=\"_blank\">@mlwhiz</a>. Please share if you got because there are similar queries I have regarding stemming and lemmatization</p>",
      "rawMarkdown": "Did you get the answer @mlwhiz. Please share if you got because there are similar queries I have regarding stemming and lemmatization",
      "votes": null
    },
    {
      "id": "1016366",
      "postDate": "09/18/2020 22:16:43",
      "content": "<p>As Abhishek said: Almost everyone is using word embeddings and when you stem, <strong>you wont be able to find the embedding for the stemmed word</strong>. thats what i think the reason is :)</p>",
      "rawMarkdown": "As Abhishek said: Almost everyone is using word embeddings and when you stem, **you wont be able to find the embedding for the stemmed word**. thats what i think the reason is :)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1005782,
      "author_name": "ranand60",
      "author_url": "",
      "post_date": "09/10/2020 18:33:36",
      "content": "<p>Did you get the answer <a href=\"https://www.kaggle.com/mlwhiz\" target=\"_blank\">@mlwhiz</a>. Please share if you got because there are similar queries I have regarding stemming and lemmatization</p>",
      "votes": null,
      "replies": [
        {
          "id": 1016366,
          "author_name": "mlwhiz",
          "author_url": "",
          "post_date": "09/18/2020 22:16:43",
          "content": "<p>As Abhishek said: Almost everyone is using word embeddings and when you stem, <strong>you wont be able to find the embedding for the stemmed word</strong>. thats what i think the reason is :)</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 451574,
      "author_name": "abhishek",
      "author_url": "",
      "post_date": "01/07/2019 09:43:46",
      "content": "<p>almost everyone is using word embeddings and when you stem, you wont be able to find the embedding for the stemmed word. thats what i think the reason is :)</p>",
      "votes": null,
      "replies": [
        {
          "id": 451587,
          "author_name": "mlwhiz",
          "author_url": "",
          "post_date": "01/07/2019 09:59:13",
          "content": "<p>That seems like the reason but what if we stem the word in the embedding too while processing the embedding?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 451633,
          "author_name": "philippsinger",
          "author_url": "",
          "post_date": "01/07/2019 12:00:15",
          "content": "<p>Try it out? In the end it doesn't matter much if you find the respective embedding.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 451581,
      "author_name": "hireme",
      "author_url": "",
      "post_date": "01/07/2019 09:51:06",
      "content": "<p>I think it's up to you, you can use it :)</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "451572": "Just as a curiosity. why don't we think of using stemming as preprocessing while using RNN based models? Is there some reason for it?",
    "451574": "almost everyone is using word embeddings and when you stem, you wont be able to find the embedding for the stemmed word. thats what i think the reason is :)",
    "451581": "I think it's up to you, you can use it :)",
    "451587": "That seems like the reason but what if we stem the word in the embedding too while processing the embedding?",
    "451633": "Try it out? In the end it doesn't matter much if you find the respective embedding.",
    "1005782": "Did you get the answer @mlwhiz. Please share if you got because there are similar queries I have regarding stemming and lemmatization",
    "1016366": "As Abhishek said: Almost everyone is using word embeddings and when you stem, **you wont be able to find the embedding for the stemmed word**. thats what i think the reason is :)"
  },
  "source": "meta"
}