Skip to main content
Powered by ShareScore

Find research datasets worth reusing

Search datasets from major research repositories and use ShareScore to quickly assess how well each record supports discovery, access, and reuse.

5

datasets available to search

ShareScore release 0.9.0

Reset

Dataset results

5 results for β€œAncient Languages”

Learn how ShareScore rates datasets β†—
zenodo40/100

ACHILLES: Ancient and Historical Language Evaluation Set

<p>The dataset used in the <a href="https://sigtyp.github.io/st2024.html">SIGTYP 2024 Shared Task on Word Embedding Evaluation for Ancient and Historical Languages</a>. The task included four problems; problems 1-3 were offered in both <a href="https://codalab.lisn.upsaclay.fr/competitions/16822">constrained</a> and <a href="https://codalab.lisn.upsaclay.fr/competitions/16818">unconstrained</a> tracks on CodaLab, while problem 4 was only a part of the unconstrained track.</p> <ol> <li>POS-tagging</li> <li>Lemmatisation</li> <li>Morphological feature prediction</li> <li>Mask filling <ul> <li>Word-level</li> <li>Character level</li> </ul> </li> </ol> <p>For problems 1-3, data from <a href="http://hdl.handle.net/11234/1-5150">Universal Dependencies v.2.12</a> was used for Ancient Greek, Ancient Hebrew, Classical Chinese, Coptic, Gothic, medieval Icelandic, Latin, Old Church Slavonic, Old East Slavic, Old French and Vedic Sanskrit. Old Hungarian texts, annotated to the same standard as UD corpora, were added to the dataset from the <a href="http://oldhungariancorpus.nytud.hu/en-codices.html">MGTSZ website</a>. In Old Hungarian data, tokens which were POS-tagged PUNCT were altered so that the form matched the lemma to simplify complex punctuation marks used to approximate manuscript symbols; otherwise, no characters were changed.</p> <p>As the ISO 639-3 standard does not distinguish between historical stages of Latin, as it does between other languages like Irish, but it was desirable to approximate this distinction for Latin, we further split Latin data. This resulted in two Latin datasets: Classical and Late Latin, and Medieval Latin. This split was dictated by the composition of the&nbsp;<a href="http://perseusdl.github.io/treebank_data/">Perseus</a> and<a href="http://dev.syntacticus.org/proiel.html"> PROIEL</a> treebanks that served as a source for Latin UD treebanks.</p> <p>Historical forms of Irish were only included in mask filling challenges (problem 4), as the quantity of historical Irish text data which has been tokenised and annotated to a single standard to date is insufficient for the purpose of training models to perform morphological analysis tasks. The texts were drawn from <a href="https://celt.ucc.ie/publishd.html">CELT</a>, <a href="http://corpas.ria.ie/index.php?fsg_function=1">Corpas Stairi&uacute;il na Gaeilge</a>, and digital editions of the <a href="http://www.stgallpriscian.ie/">St. Gall glosses</a> and the <a href="https://wuerzburg.ie/">W&uuml;rzburg glosses</a>. Each Irish text taken from CELT is labelled "Old", "Middle" or "Early Modern" in accordance with the language labels provided in CELT metadata. Because CELT metadata relating to language stages and text dating is reliant on information provided by a variety of different editors of earlier print editions, this metadata can be inconsistent across the corpus and on occasion inaccurate. To mitigate complications arising from this, texts drawn from CELT were included in the dataset only if they had a single Irish language label and if the dates provided in CELT metadata for the text match the expected dates for the given period in the history of the Irish language.</p> <p>The upper temporal boundary was set at 1700 CE, and texts created later than this date were not included in the dataset. The choice of this date is driven by the fact that most of the historical language data used in word embedding research dates back to the 18th century CE or later, and our intention was to focus on the more challenging and yet unaddressed data. The resulting datasets for each language were then shuffled at the sentence level and split into training, validation and test subsets at the ratio of 0.8 : 0.1 : 0.1.</p> <p>A <a href="https://github.com/sigtyp/ST2024/blob/main/list_of_text_sources.md">detailed list of text sources</a> for each language in the dataset, as well as <a href="https://github.com/sigtyp/ST2024/tree/main#data">other metadata</a> and the <a href="https://github.com/sigtyp/ST2024/tree/main#data-format-for-tasks-1-3">description of data formats</a> used for each problem, is provided on the <a href="https://github.com/sigtyp/ST2024/">Shared Task's GitHub</a>. The structure of the dataset is as follows:</p> <pre>πŸ“‚ <strong>morphology</strong> <em>(data for problems 1-3)</em> β”œβ”€β”€ πŸ“‚ test<br> &nbsp; &nbsp; &nbsp; β”œβ”€β”€ πŸ“‚ ref <em>(reference data used in CodaLab competitions)</em><br> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; β”œβ”€β”€ πŸ“‚ lemmatisation<br> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; β”œβ”€β”€ πŸ“‚ morph_features<br> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; └── πŸ“‚ pos_tagging<br> &nbsp; &nbsp; &nbsp; └── πŸ“‚ src <em>(source test data with labels)</em> &nbsp; &nbsp;β”œβ”€β”€ πŸ“‚ train &nbsp; &nbsp;└── πŸ“‚ valid<br><br>πŸ“‚ <strong>fill_mask_word</strong> <em>(data for problem 4a)</em><br> β”œβ”€β”€ πŸ“‚ test<br> &nbsp; &nbsp; &nbsp; β”œβ”€β”€ πŸ“‚ ref <em>(reference data used in CodaLab competitions)</em><em><br> &nbsp; &nbsp; &nbsp; └── </em>πŸ“‚ src<em> (source test data with labels in 2 different formats)<br> β”œβ”€β”€ </em>πŸ“‚ json<br> &nbsp; └── πŸ“‚ tsv<br> β”œβ”€β”€ πŸ“‚ train <em>(train data in 2 different formats)</em><br><em> β”œβ”€β”€ </em>πŸ“‚ json<br> &nbsp; └── πŸ“‚ tsv<br> &nbsp; └── πŸ“‚ valid <em>(validation data in 2 different formats)</em><br><em> β”œβ”€β”€ </em>πŸ“‚ json<br> &nbsp; └── πŸ“‚ tsv<br><br>πŸ“‚ <strong>fill_mask_char </strong>(data for problem 4b) </pre> <pre> β”œβ”€β”€ πŸ“‚ test<br> &nbsp; &nbsp; &nbsp; β”œβ”€β”€ πŸ“‚ ref <em>(reference data used in CodaLab competitions)</em><em><br> &nbsp; &nbsp; &nbsp; └── </em>πŸ“‚ src<em> (source test data with labels in 2 different formats)<br> β”œβ”€β”€ </em>πŸ“‚ json<br> &nbsp; └── πŸ“‚ tsv<br> β”œβ”€β”€ πŸ“‚ train <em>(train data in 2 different formats)</em><br><em> β”œβ”€β”€ </em>πŸ“‚ json<br> &nbsp; └── πŸ“‚ tsv<br> &nbsp; └── πŸ“‚ valid <em>(validation data in 2 different formats)</em><br><em> β”œβ”€β”€ </em>πŸ“‚ json<br> &nbsp; └── πŸ“‚ tsv<br><br></pre> <p>We would like to thank Ekaterina Melnikova for suggesting the name for the dataset.</p>

opencc-by-4.0Feb 2024View details β†’
zenodo40/100

Supplementary materials of "Calibrated weighted permutation test detects ancient language connections in the Circumpolar area (Chukotian-Nivkh and Yukaghir-Samoyedic)"

<p>Supplementary materials of the paper:&nbsp;Alexei S. Kassian, George Starostin, Mikhail Zhivlov, Sergey A. Spirin. Calibrated weighted permutation test detects ancient language connections in the Circumpolar area (Chukotian-Nivkh and Yukaghir-Samoyedic).&nbsp;<em>Journal of Historical Linguistics</em>, 2024, <a href="https://doi.org/10.1075/jhl.00014.kas">https://doi.org/10.1075/jhl.00014.kas</a>.&nbsp;</p> <p>The Supplement includes the datasets, descriptions, comments and all technical files which makes our linguistic experiment fully reproducible:</p> <ul> <li>Supplement text (pdf): 1. Consonant classes; 2. Weighted permutation test; 3. Negative control and calibration of permutation test outcomes; 4. Pairwise comparison between Yeniseian, Burushaski and all the Na-Dene wordlists; 5. CC-matches in statistically significant pairs (Chukotian-Itelmen, Nivkh, Samoyed, Yukaghir, Yeniseian, Na-Dene, Burushaski); 6. Basic information on the language families and groups; 7. Linguistic comments on individual Swadesh forms; 8. Transcription system.</li> <li>Supplement tables (xls): 1. Stability of the Swadesh concepts; 2. 110-item wordlists for the languages of the Circumpolar area; 3. Weighted permutation test for the negative control group; 4. Consonant classes.</li> <li>Technical files of the permutation test: 1. Circumpolar languages; 2. Negative control group (incl. dataset); 3. Positive control group; 4. Yeniseian-Dene-Burushaski hypothesis</li> </ul> <p>The software package is available at <a href="https://github.com/dmkrylov/starling-permutation-test">https://github.com/dmkrylov/starling-permutation-test</a></p>

opencc-by-4.0Aug 2023View details β†’
zenodo40/100

Ancient Greek language models

<p>In this repository, we release a series of vector space models of Ancient Greek, trained following different architectures and with different hyperparameter values.&nbsp;</p> <p>Below is a breakdown of all the models released, with an indication of the training method and hyperparameters. The models are split into &lsquo;<strong>Diachronica&rsquo; </strong>and &lsquo;<strong>ALP&rsquo; </strong>models, according to the published paper they are associated with.</p> <blockquote> <p>[<strong>Diachronica</strong>:] Stopponi, Silvia, Nilo Pedrazzini, Saskia Peels-Matthey, Barbara McGillivray &amp; Malvina Nissim. Forthcoming. Natural Language Processing for Ancient Greek: Design, Advantages, and Challenges of Language Models, <em>Diachronica</em>.</p> <p>[<strong>ALP</strong>:] Stopponi, Silvia, Nilo Pedrazzini, Saskia Peels-Matthey, Barbara McGillivray &amp; Malvina Nissim. 2023. Evaluation of Distributional Semantic Models of Ancient Greek: Preliminary Results and a Road Map for Future Work. <em>Proceedings of the Ancient Language Processing Workshop associated with the 14th International Conference on Recent Advances in Natural Language Processing (RANLP 2023)</em>. 49-58. Association for Computational Linguistics (ACL). https://doi.org/10.26615/978-954-452-087-8.2023_006</p> </blockquote> <h1><em>Diachronica</em> models</h1> <h2>Training data</h2> <p>Diorisis corpus (Vatri &amp; McGillivray 2018). Separate models were trained for:</p> <ol> <li>Classical subcorpus</li> <li>Hellenistic subcorpus</li> <li>Whole corpus</li> </ol> <p>Models are named according to the (sub)corpus they are trained on (i.e. <code>hel_</code> or <code>hellenestic</code> is appended to the name of the models trained on the Hellenestic subcorpus, <code>clas_</code> or <code>classical</code> for the Classical subcorpus, <code>full_</code> for the whole corpus).</p> <h2>Models</h2> <h3><strong><em>Count-based</em></strong></h3> <blockquote> <p>Software used: LSCDetection (Kaiser et al. 2021;&nbsp;<a href="https://github.com/Garrafao/LSCDetection">https://github.com/Garrafao/LSCDetection</a>)</p> </blockquote> <p>a.&nbsp;&nbsp;&nbsp;&nbsp; With Positive Pointwise Mutual Information applied (folder PPMI spaces). For each model, a version trained on each subcorpus after removing stopwords is also included (<code>_stopfilt</code> is appended to the model names). Hyperparameter values: <code>window=5</code>, <code>k=1</code>, <code>alpha=0.75</code>.</p> <p>b.&nbsp;&nbsp;&nbsp;&nbsp; With both Positive Pointwise Mutual Information <em>and</em> dimensionality reduction with Singular Value Decomposition applied (folder PPMI+SVD spaces). For each model, a version trained on each subcorpus after removing stopwords is also included (<code>_stopfilt</code> is appended to the model names). Hyperparameter values: <code>window=5</code>, <code>dimensions=300</code>, <code>gamma=0.0</code>.</p> <h3><strong><em>Word2Vec</em></strong></h3> <blockquote> <p>Software used: CADE (Bianchi et al. 2020;&nbsp;<a href="https://github.com/vinid/cade">https://github.com/vinid/cade</a>).</p> </blockquote> <p>a.&nbsp;&nbsp;&nbsp;&nbsp; Continuous-bag-of-words (CBOW). Hyperparameter values: <code>size=30</code>, <code>siter=5</code>, <code>diter=5</code>, <code>workers=4</code>, <code>sg=0</code>, <code>ns=20</code>.</p> <p>b.&nbsp;&nbsp;&nbsp;&nbsp; Skipgram with Negative Sampling (SGNS). Hyperparameter values: <code>size=30</code>, <code>siter=5</code>, <code>diter=5</code>, <code>workers=4</code>, <code>sg=1</code>, <code>ns=20</code>.</p> <h3><strong><em>Syntactic word embeddings</em></strong></h3> <p>Syntactic word embeddings were also trained on the Ancient Greek subcorpus of the PROIEL treebank (Haug &amp; J&oslash;hndal 2008), the Gorman treebank (Gorman 2020), the PapyGreek treebank (Vierros &amp; Henriksson 2021), the Pedalion treebank (Keersmaekers et al. 2019), and the Ancient Greek Dependency Treebank (Bamman &amp; Crane 2011) largely following the SuperGraph method described in Al-Ghezi &amp; Kurimo (2020) and the Node2Vec architecture (Grover &amp; Leskovec 2016) (see <a href="https://github.com/npedrazzini/ancientgreek-syntactic-embeddings#graph-based-syntactic-word-embeddings">https://github.com/npedrazzini/ancientgreek-syntactic-embeddings</a> for more details). Hyperparameter values: window=1, min_count=1.</p> <h1><em>ALP</em> models</h1> <h2>Training data</h2> <p>Archaic, Classical, and Hellenistic portions of the Diorisis corpus (Vatri &amp; McGillivray 2018) merged, stopwords removed according to the list&nbsp;made by Alessandro Vatri, available at https://figshare.com/articles/dataset/Ancient_Greek_stop_words/9724613.</p> <h2>Models</h2> <h3><strong><em>Count-based</em></strong></h3> <blockquote> <p>Software used: LSCDetection (Kaiser et al. 2021;&nbsp;<a href="https://github.com/Garrafao/LSCDetection">https://github.com/Garrafao/LSCDetection</a>)&nbsp;</p> </blockquote> <p>a.&nbsp;&nbsp;&nbsp;&nbsp; With Positive Pointwise Mutual Information applied (folder ppmi_alp).&nbsp; Hyperparameter values: <code>window=5</code>, <code>k=1</code>, <code>alpha=0.75</code>. Stopwords were removed from the training set.</p> <p>b.&nbsp;&nbsp;&nbsp;&nbsp; With both Positive Pointwise Mutual Information <em>and</em> dimensionality reduction with Singular Value Decomposition applied (folder ppmi_svd_alp). Hyperparameter values: <code>window=5</code>, <code>dimensions=300</code>, <code>gamma=0.0</code>. Stopwords were removed from the training set.</p> <h3><strong><em>Word2Vec</em></strong></h3> <blockquote> <p>Software used: Gensim library (ŘehΕ―Ε™ek and Sojka, 2010)</p> </blockquote> <p>a.&nbsp;&nbsp;&nbsp;&nbsp; Continuous-bag-of-words (CBOW). Hyperparameter values: <code>size=30</code>, <code>window=5</code>, <code>min_count=5</code>, <code>negative=20</code>, <code>sg=0</code>. Stopwords were removed from the training set.</p> <p>b.&nbsp;&nbsp;&nbsp;&nbsp; Skipgram with Negative Sampling (SGNS). Hyperparameter values: <code>size=30</code>, <code>window=5</code>, <code>min_count=5</code>, <code>negative=20</code>, <code>sg=1</code>. Stopwords were removed from the training set.</p> <h1><strong>References</strong></h1> <p>Al-Ghezi, Ragheb &amp; Mikko Kurimo. 2020. Graph-based syntactic word embeddings. In Ustalov, Dmitry, Swapna Somasundaran, Alexander Panchenko, Fragkiskos D. Malliaros, Ioana HulpuΘ™, Peter Jansen &amp; Abhik Jana (eds.), <em>Proceedings of the Graph-based Methods for Natural Language Processing (TextGraphs)</em>, 72-78.</p> <p>Bamman, D. &amp; Gregory Crane. 2011. The Ancient Greek and Latin dependency treebanks. In Sporleder, Caroline, Antal van den Bosch &amp; Kalliopi Zervanou (eds.), <em>Language Technology for Cultural Heritage. Selected Papers from the LaTeCH [Language Technology for Cultural Heritage] Workshop Series. Theory and Applications of Natural Language Processing</em>, 79-98. Berlin, Heidelberg: Springer.</p> <p>Gorman, Vanessa B. 2020. Dependency treebanks of Ancient Greek prose. <em>Journal of Open Humanities Data</em> 6(1).</p> <p>Grover, Aditya &amp; Jure Leskovec. 2016. Node2vec: scalable feature learning for networks. In <em>Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD &lsquo;16)</em>, 855-864.</p> <p>Haug, Dag T. T. &amp; Marius L. J&oslash;hndal. 2008. Creating a parallel treebank of the Old Indo-European Bible translations. In <em>Proceedings of the Second Workshop on Language Technology for Cultural Heritage Data (LaTeCH)</em>, 27&ndash;34.</p> <p>Keersmaekers, Alek, Wouter Mercelis, Colin Swaelens &amp; Toon Van Hal. 2019. Creating, enriching and valorizing treebanks of Ancient Greek. In Candito, Marie, Kilian Evang, Stephan Oepen &amp; Djam&eacute; Seddah (eds.), <em>Proceedings of the 18th International Workshop on Treebanks and Linguistic Theories</em> <em>(TLT, SyntaxFest 2019)</em>, 109-117.</p> <p>Kaiser, Jens, Sinan Kurtyigit, Serge Kotchourko &amp; Dominik Schlechtweg. 2021. Effects of Pre- and Post-Processing on type-based Embeddings in Lexical Semantic Change Detection. In <em>Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics</em>.</p> <p>Schlechtweg, Dominik, Anna H&auml;tty, Marco del Tredici &amp; Sabine Schulte im Walde. 2019. A Wind of Change: Detecting and Evaluating Lexical Semantic Change across Times and Domains. In <em>Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics</em>, 732-746, Florence, Italy. ACL.</p> <p>Vatri, Alessandro &amp; Barbara McGillivray. 2018. The Diorisis Ancient Greek Corpus: Linguistics and Literature.&nbsp;<em>Research Data Journal for the Humanities and Social Sciences</em>&nbsp;3, 1, 55-65, Available From: Brill&nbsp;<a href="https://doi.org/10.1163/24523666-01000013" target="_blank" rel="noopener">https://doi.org/10.1163/24523666-01000013</a></p> <p>Vierros, Marja &amp; Erik Henriksson. 2021. PapyGreek treebanks: a dataset of linguistically annotated Greek documentary papyri. <em>Journal of Open Humanities Data</em> 7.</p>

opencc-by-4.0Sep 2023View details β†’
zenodo36/100

Dataset for the paper "Ideology and identity in grammar: A diachronic-quantitative approach to language standardisation processes in Ancient Greek"

<p>This is the dataset for the paper &quot;Ideology and identity in grammar: A diachronic-quantitative approach to language standardisation processes in Ancient Greek&quot;</p>

opencc-by-4.0Jun 2021View details β†’
zenodo24/100

Harnessing Natural Language Processing to Deciphers Ancient Temple Texts

<p>This paper explores the use of Natural Language Processing (NLP) and Optical Character Recognition (OCR) to digitize and translate ancient Indian temple inscriptions. By applying machine learning to ancient languages like Sanskrit and Tamil, the research offers methods to decode and reconstruct eroded texts. Semantic analysis tools are also used to uncover the cultural and historical significance of the inscriptions. This interdisciplinary approach combines data science with cultural preservation, providing new ways to understand ancient texts.</p>

opencc-by-4.0Oct 2024View details β†’

ScienceDex guides

Understand access before you commit

These curated guides explain access requirements, typical timelines, costs, and reuse considerations for widely used research datasets.

Compare curated datasets

Allen Brain Atlas

Allen Brain Atlas is an Allen Institute collection of brain map atlases, datasets, APIs, and analysis tools covering mouse, human, and non-human primate brain resources.

allen-brain-atlas
neuroscienceopenDocumentation, web resources, and API references are available online.
Last verified 2026-04-30Open record

Annotated Behaviour and Observability Dataset (ABODe)

ABODe is a University of Edinburgh DataShare dataset for behavior classification in group-housed mice using home-cage video, identities, bounding boxes, ground-plate positions, and annotator labels.

abode-home-cage
behavioral-neuroscienceopenThe DataShare record exposes download links for annotations, documentation, license text, and the zipped per-snippet data directory.
Last verified 2026-04-30Open record

DANDI Archive for NWB datasets

DANDI is a BRAIN Initiative archive for publishing and sharing neurophysiology data, including electrophysiology, optophysiology, and behavioral data packaged as NWB and related standards.

dandi-nwb
electrophysiologyopenPublished Dandiset metadata and archive endpoints are available through the production DANDI API.
Last verified 2026-04-30Open record

International Brain Laboratory public data

The International Brain Laboratory public data releases expose standardized mouse decision-making experiments, including Neuropixels recordings, widefield calcium imaging, behavior, and session metadata accessed through the ONE API.

ibl
behavioral-neuroscienceopenPublic sessions can be searched and loaded from the IBL public data server through ONE.
Last verified 2026-04-29Open record

OpenNeuro

OpenNeuro is a free, open platform for sharing neuroimaging datasets, with public search, dataset pages, and download paths for web, S3, DataLad, and the OpenNeuro CLI.

openneuro
neuroscienceopenPublished datasets are available on demand over the internet.
Last verified 2026-04-29Open record

ScienceDex for faculty

Your academic CV and NIH biosketch

Make your CV in your school's format and your NIH biosketch from it, then download both as Word files.