{
  "fetched_at": "2026-10-19T11:49:51Z",
  "source_url": "https://www.kaggle.com/competitions/alta-2018-challenge/data",
  "pages": [
    {
      "name": "rules",
      "content": "This competition has two categories:\n\n* All members of the **student category** team must be university students. The team cannot have members that are full-time employed or have completed a PhD.\n*  Any other teams will fall into the **open category**.\n\nThe results to submit must be produced automatically.\n\nYou are free to use any resources, as long as these resources do not contain the actual classification labels of the test data.\n\nThe organisers reserve the right not to award the prize if there are no submissions that improve the results of the Naive Bayes baseline in the private partition of the test set.\n\nTo register, send an email to Diego Molla ([shared.task@alta.asn.au](shared.task@alta.asn.au)) with the following information:\n\n*  Name of your team;\n*  Name and current studies, including the current year, degree name, and university (e.g. \"third year of the Bachelor of IT at Macquarie University\") of each team member.\n\nThe winning team will be awarded a prize of $350.\n\nThe results of this competition will be presented at the [2018 ALTA workshop][1] in Dunedin on December 2018. To claim the prize, the winning team must present their work at the workshop. All participants are encouraged to prepare a poster and/or submit a paper describing their work, and present it in the 2018 ALTA workshop.\n\n\n  [1]: http://alta2018.alta.asn.au/",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "data-description",
      "content": "<h2>File descriptions</h2>\n<ul>\n<li><strong>patents.zip</strong> - The text files with the patents</li>\n<li><strong>train_data.csv</strong> - the training set</li>\n<li><strong>test_data.csv</strong> - the test set</li>\n<li><strong>sampleSubmission.csv</strong> - a sample submission file in the correct format</li>\n</ul>\n<h2>Data fields</h2>\n<ul>\n<li><strong>id</strong> - an anonymous id unique to a given patent; the corresponding patent is a text file with the name id.txt</li>\n<li><strong>first_ipc_mark_section</strong> - the main section of the first IPC mark (from A to H)</li>\n</ul>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "Please read the task description at the [ALTA 2018 shared task website](http://alta.asn.au/events/sharedtask2018/index.html).\n\nFeel free to use the code of the Naive Bayes baseline to get started! The code is available from the \"Kernels\" tab.\n\nTo register please send an email to Diego Molla (shared.task@alta.asn.au) with the following information:\n\n* Name of your team;\n* Name and current studies, including the current year, degree name, and university (e.g. \"third year of the Bachelor of IT at Macquarie University\") of each team member.\n\n**Submission deadline: 22 October 2018**\n\nSubmissions are now closed but you can still submit late submissions and compare your results with those of the systems that participated.\n\n**If you use the data from this shared task, please cite:**\n\nDiego Molla, Dilesha Seneviratne. Overview of the 2018 ALTA Shared Task: Classifying Patent Applications. *Proceedings ALTA 2018*, Dunedin.\nhttps://aclweb.org/anthology/papers/U/U18/U18-1011/",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "<p>The evaluation metric for this competition is  the <strong>micro-averaged F1-Score</strong>. The F1 score measures accuracy using the statistics 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 micro-averaged computation of the true positives is the sum of all true positives across all classes. Similarly for the micro-averaged computations of the false positives and the false negatives. The micro-averaged F1 score is given by:</p>\n<p>\\[ F1 = 2\\frac{p \\cdot r}{p+r}\\ \\ \\mathrm{where}\\ \\ p = \\frac{\\sum_{k\\in C} tp_k}{\\sum_{k\\in C} tp_k+\\sum_{k\\in C} fp_k},\\ \\ r = \\frac{\\sum_{k\\in C} tp_k}{\\sum_{k\\in C} tp_k+\\sum_{k\\in C} fn_k}\\]</p>\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}