Datasets:

drelhaj
/

Multilingual-Summarization

Error code:   DatasetGenerationCastError
Exception:    DatasetGenerationCastError
Message:      An error occurred while generating the dataset

All the data files must have the same columns, but at some point there are 3 new columns ({'document_text', 'doc_id', 'summary'}) and 6 missing columns ({'summary_1', 'doc_ids', 'cluster_id', 'summary_2', 'documents_text', 'summary_3'}).

This happened while the csv dataset builder was generating data using

hf://datasets/drelhaj/Multilingual-Summarization/single_docs_train.csv (at revision 8411f7595caa61894e54007291ecf643446344d2)

Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1831, in _prepare_split_single
                  writer.write_table(table)
                File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 714, in write_table
                  pa_table = table_cast(pa_table, self._schema)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2272, in table_cast
                  return cast_table_to_schema(table, schema)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2218, in cast_table_to_schema
                  raise CastError(
              datasets.table.CastError: Couldn't cast
              language: string
              doc_id: string
              document_text: string
              summary: string
              -- schema metadata --
              pandas: '{"index_columns": [{"kind": "range", "name": null, "start": 0, "' + 729
              to
              {'cluster_id': Value('string'), 'language': Value('string'), 'doc_ids': Value('string'), 'documents_text': Value('string'), 'summary_1': Value('string'), 'summary_2': Value('string'), 'summary_3': Value('string')}
              because column names don't match
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1455, in compute_config_parquet_and_info_response
                  parquet_operations = convert_to_parquet(builder)
                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1054, in convert_to_parquet
                  builder.download_and_prepare(
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 894, in download_and_prepare
                  self._download_and_prepare(
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 970, in _download_and_prepare
                  self._prepare_split(split_generator, **prepare_split_kwargs)
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1702, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1833, in _prepare_split_single
                  raise DatasetGenerationCastError.from_cast_error(
              datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
              
              All the data files must have the same columns, but at some point there are 3 new columns ({'document_text', 'doc_id', 'summary'}) and 6 missing columns ({'summary_1', 'doc_ids', 'cluster_id', 'summary_2', 'documents_text', 'summary_3'}).
              
              This happened while the csv dataset builder was generating data using
              
              hf://datasets/drelhaj/Multilingual-Summarization/single_docs_train.csv (at revision 8411f7595caa61894e54007291ecf643446344d2)
              
              Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

cluster_id string	language string	doc_ids string	documents_text string	summary_1 string	summary_2 string	summary_3 string
M002	arabic	M0020;M0021;M0022;M0023;M0024;M0025;M0026;M0027;M0028;M0029	<DOC id=M0020> 15 من البحرية الملكية تم القبض عليهم تحت تهديد السلاح من قبل الحرس الإيراني. الجمعة 23 مارس 2007 تم القبض على خمسة عشر من افراد البحرية الملكية البريطانية من قبل السلطات الايرانية تحت تهديد السلاح في الخليج الفارسي قبالة السواحل العراقية ، وفقا لوزارة الدفاع البريطانية. البحرية الملكية تصر على انهم كان...	القي القبض على خمسة عشر من افراد البحرية الملكية البريطانية من قبل السلطات الايرانية تحت تهديد السلاح في الخليج الفارسي قبالة السواحل العراقية. أعلن الرئيس الإيراني محمود أحمدي نجاد إنه سيتم الإفراج عن خمسة عشر من افراد البحرية البريطانية المقبوض عليهم .وكان افراد البحرية البريطانية الخمسة عشر المعتقلين من قبل إيران ق...	تم القاء القبض على 15 جندياً من البحرية البريطانية. الحكومة البريطانية تصر على أن الجنود كانوا في مهمة رسمية لإجراء التفتيش للسفن القادمة للعراق في المياه الاقليمة العراقية. في حين أن الإعلام الايراني تحدث في البدأعلى أن الجنود كانوا في المياه الاقليمية العراقية و لكن سريعاً ما تم تغيير البيان إلى أن الجنود كانوا في ال...	تم القبض على خمسة عشر من افراد البحرية الملكية البريطانية من قبل السلطات الايرانية تحت تهديد السلاح في الخليج الفارسي قبالة السواحل العراقية وذكرت الحكومة الإيرانية أنها ليست المرة الأولى التي تدخل فيها السفن البريطانية المياه الاقليمية الايرانية. Rajanews.com ، وهو موقع الأخبار الذي يديره بعض من أنصار أحمدي نجاد قالوا...
M002	chinese	M0020;M0021;M0022;M0023;M0024;M0025;M0026;M0027;M0028;M0029	<DOC id=M0020> 15名皇家海军水兵在枪口威胁下被伊朗士兵捕获 2007年3月23日，星期五据英国国防部消息，在伊拉克海岸的波斯湾，15名英国皇家海军人员在枪口威胁下被伊朗当局捕获。皇家海军坚持认为事发当时他们在伊拉克水域活动。伊朗政府最初提供的事发位置是在伊拉克水域。在英国政府做出解释之后，伊朗对事发位置做出修改，伊朗提供的新位置是在伊拉克水域以外而且在伊朗领海范围内。伊朗政府随后发表声明，这已经不是英国船只第一次进入伊朗水域。据联合国安理会第1723号决议，来自22型“康沃尔”号导弹护卫舰的水兵和海军陆战队员正在检查一艘被认为向伊拉克走私车辆的船只，不过在检查之后马上消除了嫌疑。 2006年，“康沃尔”号导...	伊拉克海岸的波斯湾，15名英国皇家海军人员在枪口威胁下被伊朗当局捕获。他们将要面临间谍罪指控。但英国表示如果数日内伊朗不释放水手们和海军的话，将准备进入“一个不同的阶段”。英国已经冻结了所有与伊朗之间的双边贸易活动，直到被伊朗扣押的15名英国水手和海军被释放。但伊朗“推迟”释放英国士兵。欧盟已经发表声明，要求释放被扣留的人员，并称如果伊朗拒绝，将采取“适当措施”。英国夏令时中午12点，15名被扣押人员被释放，回国后描述了他们被逮捕和扣留的经过。调查发现，伊朗抓获皇家海军人员行为乃是一个失误。	07年3月23日15名皇家海军水兵在枪口威胁下被伊朗士兵捕获。被伊朗扣押的英国水手“将因间谍罪被审判”。皇家海军坚持认为事发当时他们在伊拉克水域活动。英国政府表示如果数日内伊朗不释放水手们和海军的话，英国将准备进入“一个不同的阶段”。英国冻结与伊朗的贸易活动。因伊拉克巴士达附近出现枪声伊朗“推迟”释放英国士兵。欧盟要求释放扣留在伊朗的英国船员。4月4日艾哈迈迪内贾德宣布将释放英国海军人员。英国海军人员回家后详细描述了他们被伊朗逮捕和扣留的经过。英国对伊朗抓获皇家海军人员行为的回应被指出失误。	2007年3月23日，15名英国皇家海军人员在伊拉克水域执行例行巡逻任务时，被伊朗士兵捕获。被伊朗革命卫队逮捕的英国水手将要面临间谍罪指控。欧盟要求释放扣留在伊朗的英国船员。英国已经冻结了所有与伊朗之间的双边贸易活动。伊朗也发表声明，如果英国转向联合国安理会或者进一步冻结与伊朗的关系，他们将会进一步推迟释放士兵。英国表示，如果数日内不释放水手的话，将准备进入“一个不同的阶段”。伊朗宣布将释放英国海军人员。15名被伊朗扣留的英国海军人士得到释放，并在记者招待会上，描述他们被伊朗逮捕和扣留的经过。
M002	czech	M0020;M0021;M0022;M0023;M0024;M0025;M0026;M0027;M0028;M0029	<DOC id=M0020> 15 námořníků královského námořnictva zajato před hlavněmi íránské stráže Pátek, 23. března 2007 Patnáct příslušníků Britského královského námořnictva bylo zajato íránskými ozbrojenými autoritami v Perském zálivu nedaleko iráckého pobřeží, podle britského ministerstva obrany. Královské námořnictvo trvá ...	Patnáct příslušníků Britského královského námořnictva bylo zajato 23. března 2007 íránskými ozbrojenými autoritami v Perském zálivu nedaleko iráckého pobřeží, podle britského ministerstva obrany. Královské námořnictvo trvá na tom, že operovali v iráckých vodách. Íránská vláda původně udala pozici incidentu uvnitř iráck...	V pátek 23. března 2007 bylo patnáct příslušníků Britského královského námořnictva zajato íránskými ozbrojenými autoritami v Perském zálivu nedaleko iráckého pobřeží. Incident se stal blízko vodní cesty Shatt al-Arab. Po komentáři bristké vlády Írán zveřejnil změněnou pozici incidentu, která byla mimo irácké vody a uvn...	23. března 2007 íránské ozbrojené autority zajali v Perském zálivu patnáct britských námořníků z fregaty HMS Cornwall, kteří prověřovali loď podezřelou z pašování. Írán tvrdí, že Britové vpluli do íránských vod a pokud jim bude prokázán záměr, budou obviněni ze špionáže. Špionáž je v Íránu trestána smrtí. Incident se o...
M002	english	M0020;M0021;M0022;M0023;M0024;M0025;M0026;M0027;M0028;M0029	<DOC id=M0020> 15 Royal Navy sailors captured at gunpoint by Iranian guards Friday, March 23, 2007 Fifteen British Royal Navy personnel have been captured by Iranian authorities at gunpoint in the Persian Gulf off the Iraqi coast, according to the British Ministry of Defence. The Royal Navy insists that they were ope...	Two years after the seizure of Royal Navy personnel by Iran, two inquiries, that examined the British Ministry's of Defence handling, identified weaknesses in training, communications and the handling of intelligence as well as "collective failure of judgement". The fifteen sailors and marines, from the frigate HMS ...	In March 2007 a British frigate with 15 Navy personnel, including a woman, have been captured by Iranian authorities, while they were investigating a ship suspected of smuggling cars in Iraq , on charges that they entered illegally into Iran's territorial waters. The detainees were taken in Tehran and unofficial inform...	Fifteen British Royal Navy personnel was been captured by Iranian authorities at gunpoint in the Persian Gulf off the Iraqi coast. According to Britain, a ship that was believed to be smuggling cars into Iraq, was been checked. In accordance with Iran, a British ship was approaching an Iranian site, which formerly bel...
M002	french	M0020;M0021;M0022;M0023;M0024;M0025;M0026;M0027;M0028;M0029	<DOC id=M0020> 15 marins de la Royal Navy capturés à main armée par des gardes iraniens Vendredi 23 Mars 2007 Selon le ministère britannique de la défense, quinze marins de la Royal Navy ont été capturés à main armée par les autorités iraniennes dans le Golfe Persique près des côtes Irakiennes. La Royal Navy insiste ...	Le 23 mars 2007, 15 soldats et marins de la Royal Navy Britannique ont été capturés par les gardes côtes Iraniens près de l'embouchure du Chatt el-Arab. Ils étaient en train d'inspecter un navire suspecté d'importer illégalement des voitures en Irak quand ils ont été arrêtés par les bateaux Iraniens et emmenés vers la ...	Quinze marins de la Royal Navy ont été capturés à main armée par les autorités iraniennes dans le Golfe Persique près des côtes Irakiennes. L'Iran soutient qu'ils étaient dans les eaux territoriales Iraniennes alors que le Royaume Uni insiste qu'ils opéraient en eaux Iraqiennes selon les termes d'un mandat des Nations ...	Le 23 mars 2007, quinze marins de la Royal Navy ont été capturés par les autorités iraniennes dans le Golfe persique. Ces marins étaient alors en train d'inspecter un navire. L'Iran soutient que la capture a été faite dans les eaux iraniennes et que ce n'était pas la première fois que les navires britanniques entraie...
M002	greek	M0020;M0021;M0022;M0023;M0024;M0025;M0026;M0027;M0028;M0029	"<DOC id=M0020>\n15 ναύτες του Βασιλικού πολεμικού ναυτικού αι(...TRUNCATED)	"Δυο χρόνια μετά την αρπαγή του προσωπικού του Βασιλικ(...TRUNCATED)	"Τον Μάρτιο του 2007 βρετανική φρεγάτα με 15 άτομα πλήρωμα(...TRUNCATED)	"Βρετανικό δεκαπενταμελές πλήρωμα αιχμαλωτίστηκε από (...TRUNCATED)
M002	hebrew	M0020;M0021;M0022;M0023;M0024;M0025;M0026;M0027;M0028;M0029	"<DOC id=M0020>\n15 חיילי הצי הבריטי המלכותי נתפסו באיומי אקדח (...TRUNCATED)	"ביום שישי, 23 במרץ, 2007, נשבו 15 אנשי צוות בריטי, 8 מתוכם מלח(...TRUNCATED)	"ב 23 במרץ נחטפו באיומי רובים וכלי נשק 15 מלחים איראנים ל(...TRUNCATED)	"חמישה עשר אנשי צוות מהצי הבריטי המלכותי נתפסו על ידי ה(...TRUNCATED)
M002	hindi	M0020;M0021;M0022;M0023;M0024;M0025;M0026;M0027;M0028;M0029	"<DOC id=M0020>\n१५रायल नौसेना नविक बन्दूक की नो(...TRUNCATED)	"शुक्रवार, मार्च २३, २००७ को संयुक्त रा(...TRUNCATED)	"क्षेत्रीय क्षेत्र विवाद ईरान और ईराक (...TRUNCATED)	"१५ ब्रिटिश नाविक और मरीनों को ईरान द्(...TRUNCATED)
M002	romanian	M0020;M0021;M0022;M0023;M0024;M0025;M0026;M0027;M0028;M0029	"<DOC id=M0020>\n15 marinari ai Marinei Regale capturați de gardienii irakieni sub amenințarea arm(...TRUNCATED)	"Cincisprezece marinari și pușcași marini din Fregata şi Marina Regală a Regatului Unit, între(...TRUNCATED)	"Cincisprezece marinari și pușcași marini din Fregata Regală a Regatului Unit au fost reținuți(...TRUNCATED)	"În martie 2007, Gărzile Revoluționare Iraniene au reținut cincisprezece marinari și pușcași (...TRUNCATED)
M002	spanish	M0020;M0021;M0022;M0023;M0024;M0025;M0026;M0027;M0028;M0029	"<DOC id=M0020>\n15 marineros de la Marina Real capturados a punta de pistola por guardias iraníes\(...TRUNCATED)	"Quince elementos de la Marina Real Británica han sido capturados por autoridades iraníes. El inci(...TRUNCATED)	"15 marineros e infantes de marina fueron capturados por guardias fronterizos iraníes el 23 de marz(...TRUNCATED)	"Quince miembros de la Marina Real Británica han sido capturados por autoridades iraníes en el Gol(...TRUNCATED)

End of preview.

MultiLing Multilingual Summarisation Corpus

The MultiLing Multilingual Summarisation Corpus is a comprehensive multilingual benchmark for single-document and multi-document abstractive summarisation, originally created for the MultiLing 2011 and MultiLing 2013 shared tasks held under ACL.
This release consolidates, cleans, and reformats the original resources into a standard, machine-readable dataset suitable for modern sequence-to-sequence and large language model research.

The corpus covers ten languages:

Arabic, Chinese, Czech, English, French, Greek, Hebrew, Hindi, Romanian, Spanish

and contains both:

Single-document summarisation pairs (source article, 1 gold summary)
Multi-document summarisation clusters (10 related articles, 3 human-written abstractive summaries)

All texts originate from WikiNews, following the Creative Commons BY licence. See http://multiling.iit.demokritos.gr/

✨ Key Features

Single-Document Summarisation

40 languages in the original collection; the present release includes the main ten used for MultiLing 2013.
For each language, every document has:
- One source article
- One human abstractive summary
Fully parallel across languages.

Multi-Document Summarisation

Each topic consists of 10 articles describing an event sequence.
Topics appear consistently across all languages that contributed to that year’s task.
Each cluster includes three human-written abstractive summaries, produced independently.
Human summaries were constrained to 240–250 words (or equivalent byte limits for Chinese).

Parallel & Comparable Structure

The corpus was originally designed to allow:

Cross-lingual and multilingual summarisation
Comparative analyses of summarisation difficulty across languages
Multilingual evaluation of automatic summarisation metrics (ROUGE, AutoSummENG-MeMoG, NPowER)

📘 Source and Citation

This dataset is derived from the corpus described in:

Li L, Forăscu C, El-Haj M, Giannakopoulos G. (2013)
Multi-Document Multilingual Summarization Corpus Preparation, Part 1: Arabic, English, Greek, Chinese, Romanian.
In Proceedings of the MultiLing 2013 Workshop on Multilingual Multi-Document Summarization, pp. 1–12.
ACL 2013, Sofia, Bulgaria.
PDF: https://aclanthology.org/W13-3101.pdf

Please cite the paper above when using this dataset.

🗂 Dataset Structure

Single-Document Format

Each sample includes:

language — ISO folder name (ar, en, fr, etc.)
doc_id
document_text
summary

Multi-Document Format

Each sample includes:

cluster_id — e.g. M000, M014, M103
language
doc_ids — list of the ten document identifiers
documents_text — concatenated with <DOC id=…> wrappers
summary_1, summary_2, summary_3 — three reference human summaries

All files are provided in CSV and JSONL, with train/dev/test splits.

🔧 Recommended Use Cases

Multilingual abstractive summarisation
Cross-lingual evaluation of LLMs
Multi-document summarisation research
Training summarisation models on parallel news texts
Research on multilingual evaluation metrics
Cross-lingual transfer learning
Low-resource summarisation investigations

📊 Splits

The dataset is released with deterministic:

train
validation
test

splits for single-document and multi-document subsets.
For multi-document summarisation, splits are cluster-based to prevent data leakage.

📥 Loading the Dataset

from datasets import load_dataset

ds = load_dataset("YOUR_DATASET_NAME", "multi")
# or
ds = load_dataset("YOUR_DATASET_NAME", "single")

🔒 Licence

All texts originate from WikiNews under Creative Commons BY 2.5/3.0 licences. This consolidated dataset is released under CC-BY-4.0.

🙏 Acknowledgements

MultiLing is the result of a large international community effort involving contributors from more than ten universities and research centres. This cleaned and repackaged release builds on that original work to make the corpus more accessible for modern NLP research.

Downloads last month: 95