{
  "fetched_at": "2026-10-20T07:02:25Z",
  "source_url": "https://www.kaggle.com/competitions/cse244-ml-for-nlp-hw1/data",
  "pages": [
    {
      "name": "rules",
      "content": "This is a page where you can include rules that participants must accept before joining. You may wish to include rules like:\n\n - Don't cheat!\n - Apply yourself!\n - Have fun!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "data-description",
      "content": "### File descriptions\n- hw1_train.csv - the training set\n- hw1_test.csv - the test set\n- sampleSubmission.csv - a sample submission file in the correct format\n\n### Field descriptions\n\n- ID - A unique ID associated with every utterance.\n- UTTERANCE\n- CORE RELATIONS",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "The goal of this homework is to train deep learning models to determine knowledge graph relations (in this case, according to the Freebase schema) that are invoked in user utterances to a conversational system. You will be given a training set of utterances paired with a set of relations, that you can use to train multi-layer perceptron (MLP) models. Here is an example utterance from the dataset:\n\n> Show me movies directed by Woody Allen recently.\n\nThere are two relations that are invoked by this utterance:\n\n\n- *movie.directed_by*\n- *movie.initial_release_date*\n\nThe task is to train deep learning models and output the associated set of relations when given a new utterance.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "Submissions will be evaluated based on their mean F-1 score. The F score, commonly used in information retrieval, measures accuracy using the precision p and recall r. Precision is the ratio of true positives (tp) to all predicted positives (tp + fp). Recall is the ratio of true positives to all actual positives (tp + fn). The F-1 score is given by\n\n$$ \\frac{pr}{p+r}  where  p=\\frac{tp}{tp+fp},  r=\\frac{tp}{tp+fn}$$\n\nNote that the F-1 score weights recall higher than precision. The mean F-1 score is formed by averaging the individual F-1 scores for each row in the test set.\n\n### Submission File\n\nYou need to submit your predictions in the format as below:\n\n```\nID,CORE RELATIONS\n0,rel1 rel2\n1,rel1\n2,rel3\n3,re1 rel3 rel4\netc.\n```",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}