{
  "id": 3703,
  "title": "Friends network not very useful !!",
  "url": "/competitions/event-recommendation-engine-challenge/discussion/3703",
  "author_name": "imam123",
  "post_date": "2013-01-27T07:00:41.837000",
  "votes": 0,
  "comment_count": 0,
  "views": 911,
  "content": "<p>Hello Kagglers,</p>\r\n<p>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; It seems collaborative filtering is useless here. I also tried to check if friends of the user in test.csv have either said yes or maybe for the test &nbsp;event (to be recommended ) and this too seems to be going no where as out 700-800\r\n friends in some cases I only got 1 or 2 yes/maybe (event_attendees) for the event in question, very little information there. I also tried to look at second degree network (friends of friends) but this information is missing. I searched for friends in user_friends\r\n (first column) and they are not there. I will now be looking into cold start problem of recommender systems.&nbsp;</p>\r\n<p>&nbsp; &nbsp; &nbsp; &nbsp; I was wondering what approach other participants are taking .</p>\r\n<p>&nbsp;</p>\r\n<p>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp;</p>",
  "messages": [
    {
      "id": 19735,
      "postDate": "2013-01-27T07:00:41.837Z",
      "content": "<p>Hello Kagglers,</p>\r\n<p>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; It seems collaborative filtering is useless here. I also tried to check if friends of the user in test.csv have either said yes or maybe for the test &nbsp;event (to be recommended ) and this too seems to be going no where as out 700-800\r\n friends in some cases I only got 1 or 2 yes/maybe (event_attendees) for the event in question, very little information there. I also tried to look at second degree network (friends of friends) but this information is missing. I searched for friends in user_friends\r\n (first column) and they are not there. I will now be looking into cold start problem of recommender systems.&nbsp;</p>\r\n<p>&nbsp; &nbsp; &nbsp; &nbsp; I was wondering what approach other participants are taking .</p>\r\n<p>&nbsp;</p>\r\n<p>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp;</p>"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "19735": ""
  }
}