)]}'
{
  "commit": "c38d8fb443384a6febac1508c3fea5b24ecb810e",
  "tree": "a8d0c8a98843e5dc112e7f297a3033d64ed2fc20",
  "parents": [
    "d4a7b306b2ea21b18e65ac211ca358d8d9f501a5"
  ],
  "author": {
    "name": "Marc Kupietz",
    "email": "kupietz@ids-mannheim.de",
    "time": "Fri Sep 04 17:51:58 2026 +0200"
  },
  "committer": {
    "name": "Marc Kupietz",
    "email": "kupietz@ids-mannheim.de",
    "time": "Fri Sep 04 17:57:58 2026 +0200"
  },
  "message": "Scale the whole log-likelihood table with the window size\n\nca_ll\u0027s contingency table multiplied only its row total by the window\nsize, leaving cells that do not add up to one sample: the row was\ncounted in window positions while the column and the sample size were\ncounted in corpus tokens.\n\nFollowing Evert (2004), which the function cites, the sample consists of\nco-occurrence tokens rather than corpus tokens. With a window of\nwindow_size positions, every occurrence of either word takes part in\nthat many pairs, so the sample size and both marginals scale with it.\n\nThe expected co-occurrence frequency comes out as window_size * O1 * O2\n/ N either way, which is why pmi, mi2 and mi3 never depended on this and\nwhy it went unnoticed for so long. Only ll needs the rest of the table,\nand there the values differ, by 0.1% for a frequent collocate up to well\nover 100% for a frequent node in a wide window. The ll column of the\ncollocationScoreQuery example in the Readme was recomputed accordingly.\nThe light verb construction table is unaffected, its total window size\nbeing 1, where the two formulations are identical.\n\nIt also removes a pathology rather than papering over it: N -\nwindow_size * O1 could turn negative, at which point the score became\nNaN through a log of a negative number. window_size * (N - O1) cannot,\nso the warning added earlier in this development version is dropped\nagain - it treated a symptom of exactly this inconsistency.\n\nReported-by: Tim Feldmüller \u003chttps://github.com/feldmueller\u003e\n\nCo-Authored-By: Claude Opus 5 \u003cnoreply@anthropic.com\u003e\nChange-Id: I8b912c5bd870a0c9913d019043b2447f951f1891\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "e1c4df8d0f3882dd5e6a41653e5b3fec0e9be468",
      "old_mode": 33188,
      "old_path": "NEWS.md",
      "new_id": "43a233c45e35ec1f22226b6a2453d836a9e4b502",
      "new_mode": 33188,
      "new_path": "NEWS.md"
    },
    {
      "type": "modify",
      "old_id": "7c5b081a5ee56e3490bf24ec8b3351963e1173ef",
      "old_mode": 33188,
      "old_path": "R/association-scores.R",
      "new_id": "21a569be3f1211d52c1fc6087e56af939b77063b",
      "new_mode": 33188,
      "new_path": "R/association-scores.R"
    },
    {
      "type": "modify",
      "old_id": "166cb6c24137ac70191f29e667566c7cea562e6f",
      "old_mode": 33188,
      "old_path": "Readme.md",
      "new_id": "d38a7a9fc28073e5ce0a53fcc227eb63ecb2d975",
      "new_mode": 33188,
      "new_path": "Readme.md"
    },
    {
      "type": "modify",
      "old_id": "64f712f7b769be10bbdcabc8d269e0f27144c729",
      "old_mode": 33188,
      "old_path": "tests/testthat/test-association-score-functions.R",
      "new_id": "0b14d6ec97e84845d4141d7ccff00211a3915d92",
      "new_mode": 33188,
      "new_path": "tests/testthat/test-association-score-functions.R"
    }
  ]
}