{
  "id": 5812,
  "title": "I don't really understand the Top 5 Words benchmark",
  "url": "/competitions/facebook-recruiting-iii-keyword-extraction/discussion/5812",
  "author_name": "AlexM",
  "post_date": "2013-09-19T14:33:38.710000",
  "votes": 0,
  "comment_count": 0,
  "views": 981,
  "content": "<p>I don't really understand what the&nbsp;<strong>Top 5 Words&nbsp;</strong>benchmark is doing. &nbsp;The description is&nbsp;</p>\n<p><em>selecting top 5 word count in descending order after removing a select few stop words and letting python's Counter.most_common() sort ties</em></p>\n<p>&nbsp;</p>\n<p>Where were the top 5 words taken from? The training set? If so from the titles or bodies or tags? And then were the top 5 words used as the predicted tags for all examples in the test set? &nbsp;I don't really understand the description</p>",
  "messages": [
    {
      "id": 31188,
      "postDate": "2013-09-19T14:33:38.710Z",
      "content": "<p>I don't really understand what the&nbsp;<strong>Top 5 Words&nbsp;</strong>benchmark is doing. &nbsp;The description is&nbsp;</p>\n<p><em>selecting top 5 word count in descending order after removing a select few stop words and letting python's Counter.most_common() sort ties</em></p>\n<p>&nbsp;</p>\n<p>Where were the top 5 words taken from? The training set? If so from the titles or bodies or tags? And then were the top 5 words used as the predicted tags for all examples in the test set? &nbsp;I don't really understand the description</p>"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "31188": ""
  }
}