{
  "id": 671402,
  "title": "Difficulty understanding additional data creation suggestion",
  "url": "/competitions/deep-past-initiative-machine-translation/discussion/671402",
  "author_name": "",
  "post_date": "2026-02-01T16:58:05.308083700Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>In the Data tab of this competition, under the Suggested Workflow for Building Additional Training Data, step 3 mentions to break the Akkadian and English texts into \"sentences and align them pairwise. This sentence-level mapping is the most useful format…\", that part confuses me.</p>\n<p>I'm still rather unexperienced with machine learning, so I don't quite understand what is meant by \"pairwise\" alignment. </p>\n<p>If someone can clarify that would be much appreciated.</p>",
  "messages": [
    {
      "id": "3400503",
      "postDate": "02/01/2026 16:58:05",
      "content": "<p>In the Data tab of this competition, under the Suggested Workflow for Building Additional Training Data, step 3 mentions to break the Akkadian and English texts into \"sentences and align them pairwise. This sentence-level mapping is the most useful format…\", that part confuses me.</p>\n<p>I'm still rather unexperienced with machine learning, so I don't quite understand what is meant by \"pairwise\" alignment. </p>\n<p>If someone can clarify that would be much appreciated.</p>",
      "rawMarkdown": "In the Data tab of this competition, under the Suggested Workflow for Building Additional Training Data, step 3 mentions to break the Akkadian and English texts into \"sentences and align them pairwise. This sentence-level mapping is the most useful format...\", that part confuses me.\n\nI'm still rather unexperienced with machine learning, so I don't quite understand what is meant by \"pairwise\" alignment. \n\nIf someone can clarify that would be much appreciated.",
      "votes": null
    },
    {
      "id": "3400551",
      "postDate": "02/01/2026 18:40:26",
      "content": "<p>It likely suggests preparing Old Assyrian-English pairs of sentences. Take a look at the exampels at \"Sentences_Oare_FirstWord_LinNum.csv\" in the Data Section, e.g., <code>text_uuid = f352e4ba-65ff-43bf-88ce-8e3e073943be</code>.</p>\n<p>There are 8 entries, each with its own entry in the <code>translation</code> column. By aligning <code>text_uuid</code> with <code>oare_id</code> in the <code>train.csv</code> you will get the full transliteration text from the <code>transliteration</code> column</p>\n<blockquote>\n  <p>'ša 2 ma-na KÙ.BABBAR ṣa-ru-pá-am i-ṣé-er en-na-nim DUMU a-zu-a e-la-ma i-šu ITU.1.KAM ša ke-na-tim li-mu-um ša qá-tí a-šùr-SIG₅ DUMU a-bar-sí-sí-im ḫa-muš-tum ša a-šur-mì-tí DUMU a-ta-a 0.66666 ma-na.TA i-na ša-tim i-ša-qal šu-ma lá iš-qúl 1.5 GÍN.TA ṣí-ib-tám a-na 1 ma-na-im i ITU.1.KAM ú-ṣa-áb IGI e-na-a-nim IGI la-qé-pì-im a-na 11 (TÚG)ku-ta-ni a-mur-IŠTAR ù a-la-ḫi-im a-na e-la-ma a-šùr-la-ma-sí a-na ITU.5.KAM ú-kà-an šu-ma lá uk-ta-in-šu-nu 0.33333 ma-na 5 GÍN.TA KÙ.BABBAR ṣa-ru-pá-am a-na e-lá-ma a-šùr-la-ma-sí i-ša-qal ITU.1.KAM ṣí-ip-um li-mu-um PUZUR₄-(d)MU IGI a-bu-sa-lim DUMU i-dí-ìr-ra IGI i-dí-a-šùr DUMU pu-ur-sí-sí-im'</p>\n</blockquote>\n<p>The <code>first_word_number</code>, <code>first_word_spelling</code> and the <code>translation</code> from the <code>Sentences_Oare_FirstWord_LinNum.csv</code> tell you the number and the spelling of the word beginning the new sentence:</p>\n<pre><code>1 ša Ennanim s. Azua owes Elamma 2 minas refined silver\n12 ITU.1.KAM Month: ša kēnātim. \n15 li-mu-um Eponymate: after Aššur-damiq s. Abarsisin. \n21 ḫa-muš-tum Ḫamuštum: Aššur-imittī s. Attaya. \n27 ma-na.TA He will pay 2/3 minas in one year. \n31 šu-ma If he does not pay, he will add interest at the rate of 1 1/2 shekels per mina per month. \n43 IGI Witnesses: Ennanim, Lā-qēpim. \n76 IGI Witnesses: Abu-šalim s. Idari, Idī-Aššur s. pu-ur-sí-sì-im.\n</code></pre>\n<h2>Taking translations from the <code>Sentences_Oare_FirstWord_LinNum.csv</code></h2>\n<p>This can help you split the transliteration into the 'left side' of the Old Assyrian-English pair:</p>\n<table>\n<thead>\n<tr>\n<th>Old Assyrian</th>\n<th>English</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>ša 2 ma-na KÙ.BABBAR ṣa-ru-pá-am i-ṣé-er en-na-nim DUMU a-zu-a e-la-ma i-šu</td>\n<td>Ennanim s. Azua owes Elamma 2 minas refined silver</td>\n</tr>\n<tr>\n<td>ITU.1.KAM ša ke-na-tim</td>\n<td>Month: ša kēnātim</td>\n</tr>\n<tr>\n<td>li-mu-um ša qá-tí a-šùr-SIG₅ DUMU a-bar-sí-sí-im</td>\n<td>Eponymate: after Aššur-damiq s. Abarsisin</td>\n</tr>\n<tr>\n<td>ḫa-muš-tum ša a-šur-mì-tí DUMU a-ta-a 0.66666</td>\n<td>Ḫamuštum: Aššur-imittī s. Attaya</td>\n</tr>\n<tr>\n<td>ma-na.TA i-na ša-tim i-ša-qal</td>\n<td>He will pay 2/3 minas in one year</td>\n</tr>\n<tr>\n<td>šu-ma lá iš-qúl 1.5 GÍN.TA ṣí-ib-tám a-na 1 ma-na-im i ITU.1.KAM ú-ṣa-áb</td>\n<td>If he does not pay, he will add interest at the rate of 1 1/2 shekels per mina per month</td>\n</tr>\n<tr>\n<td>IGI e-na-a-nim IGI la-qé-pì-im a-na 11 (TÚG)ku-ta-ni a-mur-IŠTAR ù a-la-ḫi-im a-na e-la-ma a-šùr-la-ma-sí a-na ITU.5.KAM ú-kà-an šu-ma lá uk-ta-in-šu-nu 0.33333 ma-na 5 GÍN.TA KÙ.BABBAR ṣa-ru-pá-am a-na e-lá-ma a-šùr-la-ma-sí i-ša-qal ITU.1.KAM ṣí-ip-um li-mu-um PUZUR₄-(d)MU</td>\n<td>Witnesses: Ennanim, Lā-qēpim</td>\n</tr>\n<tr>\n<td>IGI a-bu-sa-lim DUMU i-dí-ìr-ra IGI i-dí-a-šùr DUMU pu-ur-sí-sí-im</td>\n<td>Witnesses: Abu-šalim s. Idari, Idī-Aššur s. pu-ur-sí-sì-im</td>\n</tr>\n</tbody>\n</table>\n<h2>Taking translations from the <code>train.csv</code></h2>\n<p>Notice that the last translations from the <code>Sentences_Oare_FirstWord_LinNum.csv</code> are not complete. This happens sometimes, so it's better to double check with other translation sources. <code>train.csv</code> is one such source, though the phrasing and formatting differs sometimes:</p>\n<table>\n<thead>\n<tr>\n<th>Old Assyrian</th>\n<th>English</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>ša 2 ma-na KÙ.BABBAR ṣa-ru-pá-am i-ṣé-er en-na-nim DUMU a-zu-a e-la-ma i-šu</td>\n<td>That Ennānum, son of Azua, owes 2 minas of refined silver to Elamma.</td>\n</tr>\n<tr>\n<td>ITU.1.KAM ša ke-na-tim</td>\n<td>Month III</td>\n</tr>\n<tr>\n<td>li-mu-um ša qá-tí a-šùr-SIG₅ DUMU a-bar-sí-sí-im</td>\n<td>eponymy of the successor of Aššur-damiq, son of Abarsisum</td>\n</tr>\n<tr>\n<td>ḫa-muš-tum ša a-šur-mì-tí DUMU a-ta-a 0.66666</td>\n<td>week eponymy of Aššur-imittī, son of Atāya</td>\n</tr>\n<tr>\n<td>ma-na.TA i-na ša-tim i-ša-qal</td>\n<td>He will pay 0.6666 mina per year</td>\n</tr>\n<tr>\n<td>šu-ma lá iš-qúl 1.5 GÍN.TA ṣí-ib-tám a-na 1 ma-na-im i ITU.1.KAM ú-ṣa-áb</td>\n<td>if he does not pay he will add an interest of 1.5 shekel per mina per month</td>\n</tr>\n<tr>\n<td>IGI e-na-a-nim IGI la-qé-pì-im a-na 11 (TÚG)ku-ta-ni a-mur-IŠTAR ù a-la-ḫi-im a-na e-la-ma a-šùr-la-ma-sí a-na ITU.5.KAM ú-kà-an šu-ma lá uk-ta-in-šu-nu 0.33333 ma-na 5 GÍN.TA KÙ.BABBAR ṣa-ru-pá-am a-na e-lá-ma a-šùr-la-ma-sí i-ša-qal ITU.1.KAM ṣí-ip-um li-mu-um PUZUR₄-(d)MU</td>\n<td>In the presence of En(n)ānum, of Lā-qēpum. For 11 -textiles Aššur-lamassī will confirm to Elamma (the liability of) Āmur-Ištar and Ali-ahum within 5 months. If he does not confirm their liability Aššur-lamassī will pay to Elamma 25 shekels of silver apiece. Month VII, eponymy of Puzur-Nirah.</td>\n</tr>\n<tr>\n<td>IGI a-bu-sa-lim DUMU i-dí-ìr-ra IGI i-dí-a-šùr DUMU pu-ur-sí-sí-im</td>\n<td>In the presence of Abu-šalim son of Iddin-Erra, of Iddin-Aššur, son of Bursa / isum</td>\n</tr>\n</tbody>\n</table>\n<hr>\n<p>Look into past discussions, especially into <a href=\"https://www.kaggle.com/competitions/deep-past-initiative-machine-translation/discussion/668402\" target=\"_blank\">the round-up</a> posted by <a href=\"https://www.kaggle.com/jackvd\" target=\"_blank\">@jackvd</a> </p>\n<p>Hope this helps and good luck!</p>",
      "rawMarkdown": "It likely suggests preparing Old Assyrian-English pairs of sentences. Take a look at the exampels at \"Sentences_Oare_FirstWord_LinNum.csv\" in the Data Section, e.g., `text_uuid = f352e4ba-65ff-43bf-88ce-8e3e073943be`.\n\nThere are 8 entries, each with its own entry in the `translation` column. By aligning `text_uuid` with `oare_id` in the `train.csv` you will get the full transliteration text from the `transliteration` column\n\n> 'ša 2 ma-na KÙ.BABBAR ṣa-ru-pá-am i-ṣé-er en-na-nim DUMU a-zu-a e-la-ma i-šu ITU.1.KAM ša ke-na-tim li-mu-um ša qá-tí a-šùr-SIG₅ DUMU a-bar-sí-sí-im ḫa-muš-tum ša a-šur-mì-tí DUMU a-ta-a 0.66666 ma-na.TA i-na ša-tim i-ša-qal šu-ma lá iš-qúl 1.5 GÍN.TA ṣí-ib-tám a-na 1 ma-na-im i ITU.1.KAM ú-ṣa-áb IGI e-na-a-nim IGI la-qé-pì-im a-na 11 (TÚG)ku-ta-ni a-mur-IŠTAR ù a-la-ḫi-im a-na e-la-ma a-šùr-la-ma-sí a-na ITU.5.KAM ú-kà-an šu-ma lá uk-ta-in-šu-nu 0.33333 ma-na 5 GÍN.TA KÙ.BABBAR ṣa-ru-pá-am a-na e-lá-ma a-šùr-la-ma-sí i-ša-qal ITU.1.KAM ṣí-ip-um li-mu-um PUZUR₄-(d)MU IGI a-bu-sa-lim DUMU i-dí-ìr-ra IGI i-dí-a-šùr DUMU pu-ur-sí-sí-im'\n\nThe `first_word_number`, `first_word_spelling` and the `translation` from the `Sentences_Oare_FirstWord_LinNum.csv` tell you the number and the spelling of the word beginning the new sentence:\n\n```\n1 ša Ennanim s. Azua owes Elamma 2 minas refined silver\n12 ITU.1.KAM Month: ša kēnātim. \n15 li-mu-um Eponymate: after Aššur-damiq s. Abarsisin. \n21 ḫa-muš-tum Ḫamuštum: Aššur-imittī s. Attaya. \n27 ma-na.TA He will pay 2/3 minas in one year. \n31 šu-ma If he does not pay, he will add interest at the rate of 1 1/2 shekels per mina per month. \n43 IGI Witnesses: Ennanim, Lā-qēpim. \n76 IGI Witnesses: Abu-šalim s. Idari, Idī-Aššur s. pu-ur-sí-sì-im.\n```\n\n## Taking translations from the `Sentences_Oare_FirstWord_LinNum.csv`\nThis can help you split the transliteration into the 'left side' of the Old Assyrian-English pair:\n\n| Old Assyrian | English |\n| --- | --- |\n| ša 2 ma-na KÙ.BABBAR ṣa-ru-pá-am i-ṣé-er en-na-nim DUMU a-zu-a e-la-ma i-šu | Ennanim s. Azua owes Elamma 2 minas refined silver |\n| ITU.1.KAM ša ke-na-tim | Month: ša kēnātim |\n| li-mu-um ša qá-tí a-šùr-SIG₅ DUMU a-bar-sí-sí-im | Eponymate: after Aššur-damiq s. Abarsisin |\n| ḫa-muš-tum ša a-šur-mì-tí DUMU a-ta-a 0.66666 | Ḫamuštum: Aššur-imittī s. Attaya |\n| ma-na.TA i-na ša-tim i-ša-qal | He will pay 2/3 minas in one year |\n| šu-ma lá iš-qúl 1.5 GÍN.TA ṣí-ib-tám a-na 1 ma-na-im i ITU.1.KAM ú-ṣa-áb | If he does not pay, he will add interest at the rate of 1 1/2 shekels per mina per month |\n| IGI e-na-a-nim IGI la-qé-pì-im a-na 11 (TÚG)ku-ta-ni a-mur-IŠTAR ù a-la-ḫi-im a-na e-la-ma a-šùr-la-ma-sí a-na ITU.5.KAM ú-kà-an šu-ma lá uk-ta-in-šu-nu 0.33333 ma-na 5 GÍN.TA KÙ.BABBAR ṣa-ru-pá-am a-na e-lá-ma a-šùr-la-ma-sí i-ša-qal ITU.1.KAM ṣí-ip-um li-mu-um PUZUR₄-(d)MU | Witnesses: Ennanim, Lā-qēpim |\n| IGI a-bu-sa-lim DUMU i-dí-ìr-ra IGI i-dí-a-šùr DUMU pu-ur-sí-sí-im | Witnesses: Abu-šalim s. Idari, Idī-Aššur s. pu-ur-sí-sì-im |\n\n## Taking translations from the `train.csv`\n\nNotice that the last translations from the `Sentences_Oare_FirstWord_LinNum.csv` are not complete. This happens sometimes, so it's better to double check with other translation sources. `train.csv` is one such source, though the phrasing and formatting differs sometimes:\n\n| Old Assyrian | English |\n| --- | --- |\n| ša 2 ma-na KÙ.BABBAR ṣa-ru-pá-am i-ṣé-er en-na-nim DUMU a-zu-a e-la-ma i-šu | That Ennānum, son of Azua, owes 2 minas of refined silver to Elamma. |\n| ITU.1.KAM ša ke-na-tim | Month III |\n| li-mu-um ša qá-tí a-šùr-SIG₅ DUMU a-bar-sí-sí-im | eponymy of the successor of Aššur-damiq, son of Abarsisum |\n| ḫa-muš-tum ša a-šur-mì-tí DUMU a-ta-a 0.66666 | week eponymy of Aššur-imittī, son of Atāya |\n| ma-na.TA i-na ša-tim i-ša-qal | He will pay 0.6666 mina per year |\n| šu-ma lá iš-qúl 1.5 GÍN.TA ṣí-ib-tám a-na 1 ma-na-im i ITU.1.KAM ú-ṣa-áb | if he does not pay he will add an interest of 1.5 shekel per mina per month |\n| IGI e-na-a-nim IGI la-qé-pì-im a-na 11 (TÚG)ku-ta-ni a-mur-IŠTAR ù a-la-ḫi-im a-na e-la-ma a-šùr-la-ma-sí a-na ITU.5.KAM ú-kà-an šu-ma lá uk-ta-in-šu-nu 0.33333 ma-na 5 GÍN.TA KÙ.BABBAR ṣa-ru-pá-am a-na e-lá-ma a-šùr-la-ma-sí i-ša-qal ITU.1.KAM ṣí-ip-um li-mu-um PUZUR₄-(d)MU | In the presence of En(n)ānum, of Lā-qēpum. For 11 -textiles Aššur-lamassī will confirm to Elamma (the liability of) Āmur-Ištar and Ali-ahum within 5 months. If he does not confirm their liability Aššur-lamassī will pay to Elamma 25 shekels of silver apiece. Month VII, eponymy of Puzur-Nirah. |\n| IGI a-bu-sa-lim DUMU i-dí-ìr-ra IGI i-dí-a-šùr DUMU pu-ur-sí-sí-im | In the presence of Abu-šalim son of Iddin-Erra, of Iddin-Aššur, son of Bursa / isum |\n\n<hr>\n\nLook into past discussions, especially into [the round-up](https://www.kaggle.com/competitions/deep-past-initiative-machine-translation/discussion/668402) posted by @jackvd \n\nHope this helps and good luck!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3400551,
      "author_name": "angantyr",
      "author_url": "",
      "post_date": "02/01/2026 18:40:26",
      "content": "<p>It likely suggests preparing Old Assyrian-English pairs of sentences. Take a look at the exampels at \"Sentences_Oare_FirstWord_LinNum.csv\" in the Data Section, e.g., <code>text_uuid = f352e4ba-65ff-43bf-88ce-8e3e073943be</code>.</p>\n<p>There are 8 entries, each with its own entry in the <code>translation</code> column. By aligning <code>text_uuid</code> with <code>oare_id</code> in the <code>train.csv</code> you will get the full transliteration text from the <code>transliteration</code> column</p>\n<blockquote>\n  <p>'ša 2 ma-na KÙ.BABBAR ṣa-ru-pá-am i-ṣé-er en-na-nim DUMU a-zu-a e-la-ma i-šu ITU.1.KAM ša ke-na-tim li-mu-um ša qá-tí a-šùr-SIG₅ DUMU a-bar-sí-sí-im ḫa-muš-tum ša a-šur-mì-tí DUMU a-ta-a 0.66666 ma-na.TA i-na ša-tim i-ša-qal šu-ma lá iš-qúl 1.5 GÍN.TA ṣí-ib-tám a-na 1 ma-na-im i ITU.1.KAM ú-ṣa-áb IGI e-na-a-nim IGI la-qé-pì-im a-na 11 (TÚG)ku-ta-ni a-mur-IŠTAR ù a-la-ḫi-im a-na e-la-ma a-šùr-la-ma-sí a-na ITU.5.KAM ú-kà-an šu-ma lá uk-ta-in-šu-nu 0.33333 ma-na 5 GÍN.TA KÙ.BABBAR ṣa-ru-pá-am a-na e-lá-ma a-šùr-la-ma-sí i-ša-qal ITU.1.KAM ṣí-ip-um li-mu-um PUZUR₄-(d)MU IGI a-bu-sa-lim DUMU i-dí-ìr-ra IGI i-dí-a-šùr DUMU pu-ur-sí-sí-im'</p>\n</blockquote>\n<p>The <code>first_word_number</code>, <code>first_word_spelling</code> and the <code>translation</code> from the <code>Sentences_Oare_FirstWord_LinNum.csv</code> tell you the number and the spelling of the word beginning the new sentence:</p>\n<pre><code>1 ša Ennanim s. Azua owes Elamma 2 minas refined silver\n12 ITU.1.KAM Month: ša kēnātim. \n15 li-mu-um Eponymate: after Aššur-damiq s. Abarsisin. \n21 ḫa-muš-tum Ḫamuštum: Aššur-imittī s. Attaya. \n27 ma-na.TA He will pay 2/3 minas in one year. \n31 šu-ma If he does not pay, he will add interest at the rate of 1 1/2 shekels per mina per month. \n43 IGI Witnesses: Ennanim, Lā-qēpim. \n76 IGI Witnesses: Abu-šalim s. Idari, Idī-Aššur s. pu-ur-sí-sì-im.\n</code></pre>\n<h2>Taking translations from the <code>Sentences_Oare_FirstWord_LinNum.csv</code></h2>\n<p>This can help you split the transliteration into the 'left side' of the Old Assyrian-English pair:</p>\n<table>\n<thead>\n<tr>\n<th>Old Assyrian</th>\n<th>English</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>ša 2 ma-na KÙ.BABBAR ṣa-ru-pá-am i-ṣé-er en-na-nim DUMU a-zu-a e-la-ma i-šu</td>\n<td>Ennanim s. Azua owes Elamma 2 minas refined silver</td>\n</tr>\n<tr>\n<td>ITU.1.KAM ša ke-na-tim</td>\n<td>Month: ša kēnātim</td>\n</tr>\n<tr>\n<td>li-mu-um ša qá-tí a-šùr-SIG₅ DUMU a-bar-sí-sí-im</td>\n<td>Eponymate: after Aššur-damiq s. Abarsisin</td>\n</tr>\n<tr>\n<td>ḫa-muš-tum ša a-šur-mì-tí DUMU a-ta-a 0.66666</td>\n<td>Ḫamuštum: Aššur-imittī s. Attaya</td>\n</tr>\n<tr>\n<td>ma-na.TA i-na ša-tim i-ša-qal</td>\n<td>He will pay 2/3 minas in one year</td>\n</tr>\n<tr>\n<td>šu-ma lá iš-qúl 1.5 GÍN.TA ṣí-ib-tám a-na 1 ma-na-im i ITU.1.KAM ú-ṣa-áb</td>\n<td>If he does not pay, he will add interest at the rate of 1 1/2 shekels per mina per month</td>\n</tr>\n<tr>\n<td>IGI e-na-a-nim IGI la-qé-pì-im a-na 11 (TÚG)ku-ta-ni a-mur-IŠTAR ù a-la-ḫi-im a-na e-la-ma a-šùr-la-ma-sí a-na ITU.5.KAM ú-kà-an šu-ma lá uk-ta-in-šu-nu 0.33333 ma-na 5 GÍN.TA KÙ.BABBAR ṣa-ru-pá-am a-na e-lá-ma a-šùr-la-ma-sí i-ša-qal ITU.1.KAM ṣí-ip-um li-mu-um PUZUR₄-(d)MU</td>\n<td>Witnesses: Ennanim, Lā-qēpim</td>\n</tr>\n<tr>\n<td>IGI a-bu-sa-lim DUMU i-dí-ìr-ra IGI i-dí-a-šùr DUMU pu-ur-sí-sí-im</td>\n<td>Witnesses: Abu-šalim s. Idari, Idī-Aššur s. pu-ur-sí-sì-im</td>\n</tr>\n</tbody>\n</table>\n<h2>Taking translations from the <code>train.csv</code></h2>\n<p>Notice that the last translations from the <code>Sentences_Oare_FirstWord_LinNum.csv</code> are not complete. This happens sometimes, so it's better to double check with other translation sources. <code>train.csv</code> is one such source, though the phrasing and formatting differs sometimes:</p>\n<table>\n<thead>\n<tr>\n<th>Old Assyrian</th>\n<th>English</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>ša 2 ma-na KÙ.BABBAR ṣa-ru-pá-am i-ṣé-er en-na-nim DUMU a-zu-a e-la-ma i-šu</td>\n<td>That Ennānum, son of Azua, owes 2 minas of refined silver to Elamma.</td>\n</tr>\n<tr>\n<td>ITU.1.KAM ša ke-na-tim</td>\n<td>Month III</td>\n</tr>\n<tr>\n<td>li-mu-um ša qá-tí a-šùr-SIG₅ DUMU a-bar-sí-sí-im</td>\n<td>eponymy of the successor of Aššur-damiq, son of Abarsisum</td>\n</tr>\n<tr>\n<td>ḫa-muš-tum ša a-šur-mì-tí DUMU a-ta-a 0.66666</td>\n<td>week eponymy of Aššur-imittī, son of Atāya</td>\n</tr>\n<tr>\n<td>ma-na.TA i-na ša-tim i-ša-qal</td>\n<td>He will pay 0.6666 mina per year</td>\n</tr>\n<tr>\n<td>šu-ma lá iš-qúl 1.5 GÍN.TA ṣí-ib-tám a-na 1 ma-na-im i ITU.1.KAM ú-ṣa-áb</td>\n<td>if he does not pay he will add an interest of 1.5 shekel per mina per month</td>\n</tr>\n<tr>\n<td>IGI e-na-a-nim IGI la-qé-pì-im a-na 11 (TÚG)ku-ta-ni a-mur-IŠTAR ù a-la-ḫi-im a-na e-la-ma a-šùr-la-ma-sí a-na ITU.5.KAM ú-kà-an šu-ma lá uk-ta-in-šu-nu 0.33333 ma-na 5 GÍN.TA KÙ.BABBAR ṣa-ru-pá-am a-na e-lá-ma a-šùr-la-ma-sí i-ša-qal ITU.1.KAM ṣí-ip-um li-mu-um PUZUR₄-(d)MU</td>\n<td>In the presence of En(n)ānum, of Lā-qēpum. For 11 -textiles Aššur-lamassī will confirm to Elamma (the liability of) Āmur-Ištar and Ali-ahum within 5 months. If he does not confirm their liability Aššur-lamassī will pay to Elamma 25 shekels of silver apiece. Month VII, eponymy of Puzur-Nirah.</td>\n</tr>\n<tr>\n<td>IGI a-bu-sa-lim DUMU i-dí-ìr-ra IGI i-dí-a-šùr DUMU pu-ur-sí-sí-im</td>\n<td>In the presence of Abu-šalim son of Iddin-Erra, of Iddin-Aššur, son of Bursa / isum</td>\n</tr>\n</tbody>\n</table>\n<hr>\n<p>Look into past discussions, especially into <a href=\"https://www.kaggle.com/competitions/deep-past-initiative-machine-translation/discussion/668402\" target=\"_blank\">the round-up</a> posted by <a href=\"https://www.kaggle.com/jackvd\" target=\"_blank\">@jackvd</a> </p>\n<p>Hope this helps and good luck!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3400503": "In the Data tab of this competition, under the Suggested Workflow for Building Additional Training Data, step 3 mentions to break the Akkadian and English texts into \"sentences and align them pairwise. This sentence-level mapping is the most useful format...\", that part confuses me.\n\nI'm still rather unexperienced with machine learning, so I don't quite understand what is meant by \"pairwise\" alignment. \n\nIf someone can clarify that would be much appreciated.",
    "3400551": "It likely suggests preparing Old Assyrian-English pairs of sentences. Take a look at the exampels at \"Sentences_Oare_FirstWord_LinNum.csv\" in the Data Section, e.g., `text_uuid = f352e4ba-65ff-43bf-88ce-8e3e073943be`.\n\nThere are 8 entries, each with its own entry in the `translation` column. By aligning `text_uuid` with `oare_id` in the `train.csv` you will get the full transliteration text from the `transliteration` column\n\n> 'ša 2 ma-na KÙ.BABBAR ṣa-ru-pá-am i-ṣé-er en-na-nim DUMU a-zu-a e-la-ma i-šu ITU.1.KAM ša ke-na-tim li-mu-um ša qá-tí a-šùr-SIG₅ DUMU a-bar-sí-sí-im ḫa-muš-tum ša a-šur-mì-tí DUMU a-ta-a 0.66666 ma-na.TA i-na ša-tim i-ša-qal šu-ma lá iš-qúl 1.5 GÍN.TA ṣí-ib-tám a-na 1 ma-na-im i ITU.1.KAM ú-ṣa-áb IGI e-na-a-nim IGI la-qé-pì-im a-na 11 (TÚG)ku-ta-ni a-mur-IŠTAR ù a-la-ḫi-im a-na e-la-ma a-šùr-la-ma-sí a-na ITU.5.KAM ú-kà-an šu-ma lá uk-ta-in-šu-nu 0.33333 ma-na 5 GÍN.TA KÙ.BABBAR ṣa-ru-pá-am a-na e-lá-ma a-šùr-la-ma-sí i-ša-qal ITU.1.KAM ṣí-ip-um li-mu-um PUZUR₄-(d)MU IGI a-bu-sa-lim DUMU i-dí-ìr-ra IGI i-dí-a-šùr DUMU pu-ur-sí-sí-im'\n\nThe `first_word_number`, `first_word_spelling` and the `translation` from the `Sentences_Oare_FirstWord_LinNum.csv` tell you the number and the spelling of the word beginning the new sentence:\n\n```\n1 ša Ennanim s. Azua owes Elamma 2 minas refined silver\n12 ITU.1.KAM Month: ša kēnātim. \n15 li-mu-um Eponymate: after Aššur-damiq s. Abarsisin. \n21 ḫa-muš-tum Ḫamuštum: Aššur-imittī s. Attaya. \n27 ma-na.TA He will pay 2/3 minas in one year. \n31 šu-ma If he does not pay, he will add interest at the rate of 1 1/2 shekels per mina per month. \n43 IGI Witnesses: Ennanim, Lā-qēpim. \n76 IGI Witnesses: Abu-šalim s. Idari, Idī-Aššur s. pu-ur-sí-sì-im.\n```\n\n## Taking translations from the `Sentences_Oare_FirstWord_LinNum.csv`\nThis can help you split the transliteration into the 'left side' of the Old Assyrian-English pair:\n\n| Old Assyrian | English |\n| --- | --- |\n| ša 2 ma-na KÙ.BABBAR ṣa-ru-pá-am i-ṣé-er en-na-nim DUMU a-zu-a e-la-ma i-šu | Ennanim s. Azua owes Elamma 2 minas refined silver |\n| ITU.1.KAM ša ke-na-tim | Month: ša kēnātim |\n| li-mu-um ša qá-tí a-šùr-SIG₅ DUMU a-bar-sí-sí-im | Eponymate: after Aššur-damiq s. Abarsisin |\n| ḫa-muš-tum ša a-šur-mì-tí DUMU a-ta-a 0.66666 | Ḫamuštum: Aššur-imittī s. Attaya |\n| ma-na.TA i-na ša-tim i-ša-qal | He will pay 2/3 minas in one year |\n| šu-ma lá iš-qúl 1.5 GÍN.TA ṣí-ib-tám a-na 1 ma-na-im i ITU.1.KAM ú-ṣa-áb | If he does not pay, he will add interest at the rate of 1 1/2 shekels per mina per month |\n| IGI e-na-a-nim IGI la-qé-pì-im a-na 11 (TÚG)ku-ta-ni a-mur-IŠTAR ù a-la-ḫi-im a-na e-la-ma a-šùr-la-ma-sí a-na ITU.5.KAM ú-kà-an šu-ma lá uk-ta-in-šu-nu 0.33333 ma-na 5 GÍN.TA KÙ.BABBAR ṣa-ru-pá-am a-na e-lá-ma a-šùr-la-ma-sí i-ša-qal ITU.1.KAM ṣí-ip-um li-mu-um PUZUR₄-(d)MU | Witnesses: Ennanim, Lā-qēpim |\n| IGI a-bu-sa-lim DUMU i-dí-ìr-ra IGI i-dí-a-šùr DUMU pu-ur-sí-sí-im | Witnesses: Abu-šalim s. Idari, Idī-Aššur s. pu-ur-sí-sì-im |\n\n## Taking translations from the `train.csv`\n\nNotice that the last translations from the `Sentences_Oare_FirstWord_LinNum.csv` are not complete. This happens sometimes, so it's better to double check with other translation sources. `train.csv` is one such source, though the phrasing and formatting differs sometimes:\n\n| Old Assyrian | English |\n| --- | --- |\n| ša 2 ma-na KÙ.BABBAR ṣa-ru-pá-am i-ṣé-er en-na-nim DUMU a-zu-a e-la-ma i-šu | That Ennānum, son of Azua, owes 2 minas of refined silver to Elamma. |\n| ITU.1.KAM ša ke-na-tim | Month III |\n| li-mu-um ša qá-tí a-šùr-SIG₅ DUMU a-bar-sí-sí-im | eponymy of the successor of Aššur-damiq, son of Abarsisum |\n| ḫa-muš-tum ša a-šur-mì-tí DUMU a-ta-a 0.66666 | week eponymy of Aššur-imittī, son of Atāya |\n| ma-na.TA i-na ša-tim i-ša-qal | He will pay 0.6666 mina per year |\n| šu-ma lá iš-qúl 1.5 GÍN.TA ṣí-ib-tám a-na 1 ma-na-im i ITU.1.KAM ú-ṣa-áb | if he does not pay he will add an interest of 1.5 shekel per mina per month |\n| IGI e-na-a-nim IGI la-qé-pì-im a-na 11 (TÚG)ku-ta-ni a-mur-IŠTAR ù a-la-ḫi-im a-na e-la-ma a-šùr-la-ma-sí a-na ITU.5.KAM ú-kà-an šu-ma lá uk-ta-in-šu-nu 0.33333 ma-na 5 GÍN.TA KÙ.BABBAR ṣa-ru-pá-am a-na e-lá-ma a-šùr-la-ma-sí i-ša-qal ITU.1.KAM ṣí-ip-um li-mu-um PUZUR₄-(d)MU | In the presence of En(n)ānum, of Lā-qēpum. For 11 -textiles Aššur-lamassī will confirm to Elamma (the liability of) Āmur-Ištar and Ali-ahum within 5 months. If he does not confirm their liability Aššur-lamassī will pay to Elamma 25 shekels of silver apiece. Month VII, eponymy of Puzur-Nirah. |\n| IGI a-bu-sa-lim DUMU i-dí-ìr-ra IGI i-dí-a-šùr DUMU pu-ur-sí-sí-im | In the presence of Abu-šalim son of Iddin-Erra, of Iddin-Aššur, son of Bursa / isum |\n\n<hr>\n\nLook into past discussions, especially into [the round-up](https://www.kaggle.com/competitions/deep-past-initiative-machine-translation/discussion/668402) posted by @jackvd \n\nHope this helps and good luck!"
  },
  "source": "meta"
}