)]}'
{
  "commit": "8bacebb64a582d71eee5005a2de377fb1027da69",
  "tree": "82d0cb70ae7ca471f2dd268d395a199c91ad80ae",
  "parents": [
    "1e9f79174b3cb0b5983ab321db0c9e34d0f781ce"
  ],
  "author": {
    "name": "Marc Kupietz",
    "email": "kupietz@ids-mannheim.de",
    "time": "Thu Sep 03 14:26:36 2026 +0200"
  },
  "committer": {
    "name": "Marc Kupietz",
    "email": "kupietz@ids-mannheim.de",
    "time": "Thu Sep 03 14:26:36 2026 +0200"
  },
  "message": "Compute logDice as defined by Rychly, comparable to other tools\n\nlogDice was computed as 14 + log2(2 * O / (w * O1 + O2)), multiplying\nthe node frequency by the window size, while the documentation cites\nRychly (2008), who defines it as 14 + log2(2 * O / (O1 + O2)).\n\nThe window size factor does not belong here. It is correct in the\nexpected frequency E, where w * O1 counts the positions in which a\ncollocate can occur, and the other scores take the window into account\nthrough E alone. The Dice coefficient, however, relates the\nco-occurrence frequency to how often the two words occur at all, so its\ndenominator sums marginal frequencies. Adding w * O1, a count of window\npositions, to O2, a count of word tokens, mixes units, and it makes the\ncoefficient asymmetric: for Grund/triftiger it gave 0.64, for\ntriftiger/Grund 3.96, from the same counts.\n\nThe practical effect was that a frequent node made w * O1 dominate the\ndenominator - for Grund and triftiger in a 5+5 window, O2 contributed\n0.006% of it - so that rare collocates were penalised. triftiger scored\n0.64, below Berlin at 2.03, although Berlin co-occurs with Grund less\noften than chance predicts (pmi -1.74). They are now 3.96 and 3.84.\n\nScores rise by up to log2(w), so up to 3.32 for the default context of\n5 left and 5 right, which makes collocationAnalysis() somewhat more\npermissive when recursing with the default thresholdScore \u003d \"logDice\"\nand threshold \u003d 2. Results computed with a total window size of 1 are\nunaffected, the two formulas being identical there, which is why the\nlight verb construction table in the Readme stays as it is, while the\ncollocationScoreQuery example, which uses the default window, was rerun.\n\nCo-Authored-By: Claude Opus 5 \u003cnoreply@anthropic.com\u003e\nChange-Id: I190fd65d0c53517df74c26c007a6440d974f700c\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "49b43199a602c0251c4115befa2ff8a69e8b4e1c",
      "old_mode": 33188,
      "old_path": "NEWS.md",
      "new_id": "efa0a7ee4d9ae608790888e517e8d5c4a97380f3",
      "new_mode": 33188,
      "new_path": "NEWS.md"
    },
    {
      "type": "modify",
      "old_id": "741e69fa85a96c243e5f2e8c33490bb66059a3f9",
      "old_mode": 33188,
      "old_path": "R/association-scores.R",
      "new_id": "fec7991937b1939625f91cd01ac453ef5b4e10a3",
      "new_mode": 33188,
      "new_path": "R/association-scores.R"
    },
    {
      "type": "modify",
      "old_id": "c4923590bdb2946f13c380db03c0df6225298450",
      "old_mode": 33188,
      "old_path": "Readme.md",
      "new_id": "21de106968519300bb5a15791fa193e83d0683c2",
      "new_mode": 33188,
      "new_path": "Readme.md"
    },
    {
      "type": "modify",
      "old_id": "74a2864998014a0d6898955284df7c87cf6538b5",
      "old_mode": 33188,
      "old_path": "man/association-score-functions.Rd",
      "new_id": "78b5ec4d3e153519f5289e0a08611d16fd5a1ec6",
      "new_mode": 33188,
      "new_path": "man/association-score-functions.Rd"
    },
    {
      "type": "modify",
      "old_id": "f7228d39f4d4d22b19bfd8a24a5d0975ee8d15dd",
      "old_mode": 33188,
      "old_path": "tests/testthat/test-association-score-functions.R",
      "new_id": "3df08c7cfed38291936f9f0b74166e0f9df3344c",
      "new_mode": 33188,
      "new_path": "tests/testthat/test-association-score-functions.R"
    }
  ]
}