{
  "id": 680256,
  "title": "eBL text divergence",
  "url": "/competitions/deep-past-initiative-machine-translation/discussion/680256",
  "author_name": "",
  "post_date": "2026-03-06T19:35:14.748527500Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>It's a little script for the eBL text divergence, it is not perfect - it may be seen as a brutal force attack to the eBL dictionary, but it works.</p>\n<p><strong><a href=\"https://www.kaggle.com/datasets/davidochoacorrales/ebl-mimation-restorer\" target=\"_blank\">the tool</a></strong></p>\n<p>I build this readme as explanation for the code with Google Gemini - can have errors.</p>\n<hr>\n<h1>eBL to OARE Mimation Restorer</h1>\n<p>This script provides an automated solution for converting Standard Babylonian (eBL) lexemes into Old Assyrian (OARE) lexemes by restoring historical mimation.</p>\n<h2>Overview</h2>\n<p>In the context of Assyriology, Old Assyrian (OA) preserves the proto-semitic mimation, whereas Standard Babylonian (SB) lost it during the post-Old Babylonian period. This tool helps bridge the gap between these two linguistic conventions:</p>\n<ul>\n<li><strong>eBL format:</strong> Uses SB conventions, often including Roman numeral suffixes (e.g., \"šakānu I\").</li>\n<li><strong>OARE format:</strong> Adheres to OA conventions, which include the characteristic final \"-m\" (e.g., \"šakānum\").</li>\n</ul>\n<p>This converter was developed based on an empirical analysis of 585 verified pairs with a 92.8% precision rate regarding the consonant skeleton.</p>\n<h2>Conversion Logic</h2>\n<p>The script employs a three-tier priority system to handle various grammatical patterns:</p>\n<table>\n<thead>\n<tr>\n<th>Tier</th>\n<th>Strategy</th>\n<th>Description</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td><strong>Tier 1</strong></td>\n<td>Simple Addition</td>\n<td>Appends \"-m\" to standard endings (covers ~87% of cases).</td>\n</tr>\n<tr>\n<td><strong>Tier 2</strong></td>\n<td>III-weak Reversal</td>\n<td>Reverses contractions in verbs (e.g., -û → -ā'um).</td>\n</tr>\n<tr>\n<td><strong>Tier 3</strong></td>\n<td>Passthrough</td>\n<td>Retains words that already possess mimation.</td>\n</tr>\n</tbody>\n</table>\n<h2>Normalization</h2>\n<p>To ensure high-fidelity matching, the script includes a <code>normalize_for_matching</code> utility. This function handles diacritic mapping (e.g., <code>ḫ</code> → <code>h</code>, <code>š</code> → <code>sz</code>) and removes parenthetical data or alternative formatting, creating a uniform basis for comparison.</p>\n<h2>Usage Example</h2>\n<p>The script is designed to process individual strings. It produces candidate forms that can then be validated against the OARE dataset.</p>\n<pre><code>from mimationRestorer import ebl_to_oare\n\n# Simple conversion\nprint(ebl_to_oare(\"šakānu I\"))\n# Output: [('šakānum', 'nominative mimation -u→-um', 1)]\n\n# Verb III-weak conversion\nprint(ebl_to_oare(\"qabû II\"))\n# Output: [(\"qabā'um\", 'III-weak: -û → -āʾum (inf) or -ium (adj)', 2), ...]\n</code></pre>\n<h2>Credits</h2>\n<ul>\n<li><strong>Author:</strong> David ok.</li>\n<li><strong>Date:</strong> 2026-02-26.</li>\n</ul>",
  "messages": [
    {
      "id": "3417994",
      "postDate": "03/06/2026 19:35:14",
      "content": "<p>It's a little script for the eBL text divergence, it is not perfect - it may be seen as a brutal force attack to the eBL dictionary, but it works.</p>\n<p><strong><a href=\"https://www.kaggle.com/datasets/davidochoacorrales/ebl-mimation-restorer\" target=\"_blank\">the tool</a></strong></p>\n<p>I build this readme as explanation for the code with Google Gemini - can have errors.</p>\n<hr>\n<h1>eBL to OARE Mimation Restorer</h1>\n<p>This script provides an automated solution for converting Standard Babylonian (eBL) lexemes into Old Assyrian (OARE) lexemes by restoring historical mimation.</p>\n<h2>Overview</h2>\n<p>In the context of Assyriology, Old Assyrian (OA) preserves the proto-semitic mimation, whereas Standard Babylonian (SB) lost it during the post-Old Babylonian period. This tool helps bridge the gap between these two linguistic conventions:</p>\n<ul>\n<li><strong>eBL format:</strong> Uses SB conventions, often including Roman numeral suffixes (e.g., \"šakānu I\").</li>\n<li><strong>OARE format:</strong> Adheres to OA conventions, which include the characteristic final \"-m\" (e.g., \"šakānum\").</li>\n</ul>\n<p>This converter was developed based on an empirical analysis of 585 verified pairs with a 92.8% precision rate regarding the consonant skeleton.</p>\n<h2>Conversion Logic</h2>\n<p>The script employs a three-tier priority system to handle various grammatical patterns:</p>\n<table>\n<thead>\n<tr>\n<th>Tier</th>\n<th>Strategy</th>\n<th>Description</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td><strong>Tier 1</strong></td>\n<td>Simple Addition</td>\n<td>Appends \"-m\" to standard endings (covers ~87% of cases).</td>\n</tr>\n<tr>\n<td><strong>Tier 2</strong></td>\n<td>III-weak Reversal</td>\n<td>Reverses contractions in verbs (e.g., -û → -ā'um).</td>\n</tr>\n<tr>\n<td><strong>Tier 3</strong></td>\n<td>Passthrough</td>\n<td>Retains words that already possess mimation.</td>\n</tr>\n</tbody>\n</table>\n<h2>Normalization</h2>\n<p>To ensure high-fidelity matching, the script includes a <code>normalize_for_matching</code> utility. This function handles diacritic mapping (e.g., <code>ḫ</code> → <code>h</code>, <code>š</code> → <code>sz</code>) and removes parenthetical data or alternative formatting, creating a uniform basis for comparison.</p>\n<h2>Usage Example</h2>\n<p>The script is designed to process individual strings. It produces candidate forms that can then be validated against the OARE dataset.</p>\n<pre><code>from mimationRestorer import ebl_to_oare\n\n# Simple conversion\nprint(ebl_to_oare(\"šakānu I\"))\n# Output: [('šakānum', 'nominative mimation -u→-um', 1)]\n\n# Verb III-weak conversion\nprint(ebl_to_oare(\"qabû II\"))\n# Output: [(\"qabā'um\", 'III-weak: -û → -āʾum (inf) or -ium (adj)', 2), ...]\n</code></pre>\n<h2>Credits</h2>\n<ul>\n<li><strong>Author:</strong> David ok.</li>\n<li><strong>Date:</strong> 2026-02-26.</li>\n</ul>",
      "rawMarkdown": "It's a little script for the eBL text divergence, it is not perfect - it may be seen as a brutal force attack to the eBL dictionary, but it works.\n\n\n**[the tool](https://www.kaggle.com/datasets/davidochoacorrales/ebl-mimation-restorer)**\n\n\nI build this readme as explanation for the code with Google Gemini - can have errors.\n\n---\n\n# eBL to OARE Mimation Restorer\n\nThis script provides an automated solution for converting Standard Babylonian (eBL) lexemes into Old Assyrian (OARE) lexemes by restoring historical mimation.\n\n## Overview\n\nIn the context of Assyriology, Old Assyrian (OA) preserves the proto-semitic mimation, whereas Standard Babylonian (SB) lost it during the post-Old Babylonian period. This tool helps bridge the gap between these two linguistic conventions:\n\n* **eBL format:** Uses SB conventions, often including Roman numeral suffixes (e.g., \"šakānu I\").\n* **OARE format:** Adheres to OA conventions, which include the characteristic final \"-m\" (e.g., \"šakānum\").\n\nThis converter was developed based on an empirical analysis of 585 verified pairs with a 92.8% precision rate regarding the consonant skeleton.\n\n## Conversion Logic\n\nThe script employs a three-tier priority system to handle various grammatical patterns:\n\n| Tier | Strategy | Description |\n| --- | --- | --- |\n| **Tier 1** | Simple Addition | Appends \"-m\" to standard endings (covers ~87% of cases). |\n| **Tier 2** | III-weak Reversal | Reverses contractions in verbs (e.g., -û → -ā'um). |\n| **Tier 3** | Passthrough | Retains words that already possess mimation. |\n\n## Normalization\n\nTo ensure high-fidelity matching, the script includes a `normalize_for_matching` utility. This function handles diacritic mapping (e.g., `ḫ` → `h`, `š` → `sz`) and removes parenthetical data or alternative formatting, creating a uniform basis for comparison.\n\n## Usage Example\n\nThe script is designed to process individual strings. It produces candidate forms that can then be validated against the OARE dataset.\n\n```python\nfrom mimationRestorer import ebl_to_oare\n\n# Simple conversion\nprint(ebl_to_oare(\"šakānu I\"))\n# Output: [('šakānum', 'nominative mimation -u→-um', 1)]\n\n# Verb III-weak conversion\nprint(ebl_to_oare(\"qabû II\"))\n# Output: [(\"qabā'um\", 'III-weak: -û → -āʾum (inf) or -ium (adj)', 2), ...]\n\n```\n\n## Credits\n\n* **Author:** David ok.\n* **Date:** 2026-02-26.",
      "votes": null
    },
    {
      "id": "3418149",
      "postDate": "03/07/2026 09:22:41",
      "content": "<blockquote>\n  <p>š → sz</p>\n</blockquote>\n<p>If this is not a typo, I'd advise to reverse the direction to  \"sz → š\" as a standard ASCII fallback to DPC format substitution and leave it be otherwise.</p>",
      "rawMarkdown": "> š → sz\n\nIf this is not a typo, I'd advise to reverse the direction to  \"sz → š\" as a standard ASCII fallback to DPC format substitution and leave it be otherwise.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3418149,
      "author_name": "angantyr",
      "author_url": "",
      "post_date": "03/07/2026 09:22:41",
      "content": "<blockquote>\n  <p>š → sz</p>\n</blockquote>\n<p>If this is not a typo, I'd advise to reverse the direction to  \"sz → š\" as a standard ASCII fallback to DPC format substitution and leave it be otherwise.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3417994": "It's a little script for the eBL text divergence, it is not perfect - it may be seen as a brutal force attack to the eBL dictionary, but it works.\n\n\n**[the tool](https://www.kaggle.com/datasets/davidochoacorrales/ebl-mimation-restorer)**\n\n\nI build this readme as explanation for the code with Google Gemini - can have errors.\n\n---\n\n# eBL to OARE Mimation Restorer\n\nThis script provides an automated solution for converting Standard Babylonian (eBL) lexemes into Old Assyrian (OARE) lexemes by restoring historical mimation.\n\n## Overview\n\nIn the context of Assyriology, Old Assyrian (OA) preserves the proto-semitic mimation, whereas Standard Babylonian (SB) lost it during the post-Old Babylonian period. This tool helps bridge the gap between these two linguistic conventions:\n\n* **eBL format:** Uses SB conventions, often including Roman numeral suffixes (e.g., \"šakānu I\").\n* **OARE format:** Adheres to OA conventions, which include the characteristic final \"-m\" (e.g., \"šakānum\").\n\nThis converter was developed based on an empirical analysis of 585 verified pairs with a 92.8% precision rate regarding the consonant skeleton.\n\n## Conversion Logic\n\nThe script employs a three-tier priority system to handle various grammatical patterns:\n\n| Tier | Strategy | Description |\n| --- | --- | --- |\n| **Tier 1** | Simple Addition | Appends \"-m\" to standard endings (covers ~87% of cases). |\n| **Tier 2** | III-weak Reversal | Reverses contractions in verbs (e.g., -û → -ā'um). |\n| **Tier 3** | Passthrough | Retains words that already possess mimation. |\n\n## Normalization\n\nTo ensure high-fidelity matching, the script includes a `normalize_for_matching` utility. This function handles diacritic mapping (e.g., `ḫ` → `h`, `š` → `sz`) and removes parenthetical data or alternative formatting, creating a uniform basis for comparison.\n\n## Usage Example\n\nThe script is designed to process individual strings. It produces candidate forms that can then be validated against the OARE dataset.\n\n```python\nfrom mimationRestorer import ebl_to_oare\n\n# Simple conversion\nprint(ebl_to_oare(\"šakānu I\"))\n# Output: [('šakānum', 'nominative mimation -u→-um', 1)]\n\n# Verb III-weak conversion\nprint(ebl_to_oare(\"qabû II\"))\n# Output: [(\"qabā'um\", 'III-weak: -û → -āʾum (inf) or -ium (adj)', 2), ...]\n\n```\n\n## Credits\n\n* **Author:** David ok.\n* **Date:** 2026-02-26.",
    "3418149": "> š → sz\n\nIf this is not a typo, I'd advise to reverse the direction to  \"sz → š\" as a standard ASCII fallback to DPC format substitution and leave it be otherwise."
  },
  "source": "meta"
}