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.
4
datasets available to search
ShareScore release 0.9.0
Dataset results
4 results for “Tigrinya”
Tigrinya Dialect Identification (TDI)
<p><strong>The Tigrinya Dialect Identification (TDI)</strong> dataset contains text on three Tigrinya dialects or varieties namely: Z, D, and L. The purpose of this dataset is to study dialect identification for Tigrinya using machine learning.</p> <p>For the Z variety, we used snippets from the book ኽልተ ዛንታት (Kilte Zantatat). For the L variety, we used book chapters from ፋቶ (Fato) and ዕርቂ እንደርታ (Erqi Enderta). For the D variant, we could not find a book. Instead, we collected data from two Facebook users, Akeza Awalom and Guraya Asadi Raya that consistently write in that variety. Sentences collected for each dialect were translated to the other dialect with expert native speakers in the target dialect.</p> <p><br> <strong>Source by Dialect</strong></p> <table align="left"> <tbody> <tr> <td> <p><strong>Dialect</strong></p> </td> <td> <p><strong>Source</strong></p> </td> <td> <p><strong>No. sentences</strong></p> </td> </tr> <tr> <td> <p>Z</p> </td> <td> <p>Kilte Zantatat</p> </td> <td> <p>1041</p> </td> </tr> <tr> <td> <p>L</p> </td> <td> <p>Fato </p> </td> <td> <p>764</p> </td> </tr> <tr> <td> </td> <td>Erqi nderta</td> <td>405</td> </tr> <tr> <td> <p>D</p> </td> <td> <p><a href="https://www.facebook.com/akeza.awealom">Akeza Awalom</a></p> </td> <td> <p>224</p> </td> </tr> <tr> <td> </td> <td><a href="https://www.facebook.com/guraya.asadiraya">GualRaya</a></td> <td>530</td> </tr> </tbody> </table> <p><br> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p><strong>Acknowledgements</strong></p> <p> </p> <p>Special thanks to Meles Solomon, who provided us with his book, ዕርቂ እንደርታ (Erqi Enderta). He also helped with translations to the L dialect. Thanks also goes to Tesfay Gebreegziabher and Gidey Gebrekidan for allowing us to use their books Fato and Kilte Zantatat respectively. Many thanks to Teklay Berhane, Abeba Asemu, Haftu Abadi, Tsegay Kinfe, Moges Bekru, Kahsay Berhe Adhana, Kibrom Mulugeta, Tsegazeab Kidanu, Tsgab Weldemariam, Abu W Debay, Solomon Shibabaw, Hagos Hiete for their valuable contributions as translators.</p>
Tigrinya Analogy Test for evaluating Word Embeddings
<p><strong>Tigrinya Analogy Test for evaluating Word Embeddings</strong></p> <p>This is a Tigrinya version of the Google Analogy Test set, which is used to evaluate English word-embedding models. The analogy test is a well-established strategy to empirically evaluate the quality of word-embedding models. More information about the English task can be found at the <a href="https://aclweb.org/aclwiki/Google_analogy_test_set_(State_of_the_art)">ACL Wiki</a>.</p> <p>This data is was first machine translated then manually verified by a native speaker to reduce errors.</p> <p>Some aspects of the original analogy test is focused on English and may not transfer well to other languages, such as those related to grammar or morphology. Therefore, we have discarded examples that became irrelevant in Tigrinya when adapting the task. Finally, there are a total of <strong>18465</strong> entries in the Tigrinya Analogy Test set, while the source English data has <strong>19544</strong> entries.</p> <p>An entry is dropped if the translations led to one of the following conditions:</p> <ol> <li>If the source word pair map to one Tigrinya word, for example, lucky & luckiest both correspond to ዕድለኛ.</li> <li>If the source word results in a multi-word expression. For example, grandson (ወዲ ጓል / ወዲ ወዲ), granddaughter (ጓል ጓል / ጓል ወዲ). This because the typical word-embedding approaches such as <em>word2vec</em> are not designed to predict multi-word phrases.</li> </ol> <p> </p> <p><strong>Test Sections</strong></p> <p>The test includes a series of semantic and syntactic analogies divided up into subsections including world capitals, currencies, family, tense, and plurality. The test contains the following sections:</p> <ol> <li>capital-world</li> <li>currency</li> <li>city-in-state</li> <li>family</li> <li>gram1-adjective-to-adverb</li> <li>gram2-opposite</li> <li>gram3-comparative</li> <li>gram4-superlative</li> <li>gram5-present-participle</li> <li>gram6-nationality-adjective</li> <li>gram7-past-tense</li> <li>gram8-plural</li> <li>gram9-plural-verbs</li> </ol> <p> </p> <p><strong>Examples:</strong></p> <ul> <li>Semantic section of World Capitals: “ኣስመራ: ኤርትራ as ፓሪስ: ?” and if the model responds correctly it will return: “ፈረንሳ”.</li> <li>Semantic section of Family section: “ሰብኣይ: ሰበይቲ as ወዲ: ጓል”.</li> <li>Syntax section with tense, a sample analogy might be “Walk: Walked as Run: Ran”.<br> </li> </ul> <p><strong>Evaluation</strong></p> <p>The final accuracy of a model is the proportion of the questions that the model answers correctly.<br> Generally, a better-quality model would answer more questions correctly than a model of lower quality.<br> However, note that a model with low performance on this analogy test, might still contain useful information, but may not be robust or good enough for more complex tasks.</p> <p> </p> <p><strong>Limitations</strong></p> <ul> <li>The analogy test could be a good indicator of the quality of word-embeddings, but it should be used with caution when comparing models trained on varying domains of data. It shall not be expected to generalize equally to all domains.</li> <li>The final score can be affected by the size, vocabulary, and domain of the text with which the models are trained on. For example, this may not be a good benchmark to compare models trained on news text <em>vs</em> posts on social media.</li> <li>Even though a manual sanity check was performed, we note that the semi-automatic construction of the Tigrinya test set might contains errors. If you discover any, you are welcome to contribute back by either opening an <em>Issue</em> at the GitHub repo, <a href="https://github.com/fgaim/tigrinya-analogy-test">https://github.com/fgaim/tigrinya-analogy-test</a>.</li> </ul> <p> </p> <p><strong>Citation</strong></p> <p>If you use this resource in your research, please cite it accordingly.</p>
TLMD: Tigrinya Language Modeling Dataset
<p>A monolingual dataset built for Tigrinya language modeling. To the best of our knowledge, this is the largest dataset for Tigrinya of its kind. The data was collected from various sources across the web including news, blogs, and books. The largest portion of the data, ~75%, comes from over 2150 issues of the <em>Haddas Ertra</em> newspaper and other magazines published by <a href="https://www.shabait.com">www.shabait.com</a>.</p> <p>Data Statistics:</p> <ul> <li>Total size: ~0.5GB</li> <li>Around 40 million tokens</li> <li>Over 2 million lines</li> <li>367 unique characters</li> <li>Train split: 98%, 1.97 million lines</li> <li>Validation split: 2%, 43k lines</li> </ul> <p>We have done a light-weight cleanup of the data:<br> - Removal of Tigrinya text with legacy and non-standard encoding systems<br> - Normalization of punctuation and special characters<br> - Removal of redundant white spaces and empty lines<br> - Rejoining or fixing broken sentences when possible<br> - Removal of foreign words</p> <p>We avoid applying any form of tokenization, extensive cleanup, and preprocessing operations in order not to take away potentially useful information, those decisions are left to the use-case researchers or developers.</p> <p>This dataset is shared solely to advance research on natural language processing for Tigrinya. While the dataset authors do not claim any copyright on the content, some of the original sources may do. To use the content for commercial purposes or other forms of redistribution of the data, permission shall be acquired from the original owners, mainly shabait.com.</p>
TIGQA: An Expert-Annotated Question-Answering Dataset in Tigrinya
<div> <h3>What is TIGQA?</h3> <a href="https://github.com/hailaykidu/TigQA-Dataset#what-is-tigqa"></a></div> <p>TigQA is an expert-annotated dataset in Tigrinya, a low-resource language spoken by approximately 10 million speakers in Eritrea and the Tigray region of Ethiopia. Our proposed SQuAD-like dataset contains 2,685 question-answer pairs covering 122 diverse topics such as climate, water, and traffic. These pairs are from 537 context paragraphs in publicly accessible Tigrinya and Biology books, with the answers being provided by teachers from the region.</p> <div> </div> <div> </div>
ScienceDex guides
Understand access before you commit
These curated guides explain access requirements, typical timelines, costs, and reuse considerations for widely used research 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.
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.
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.
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.
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.