--- language: - afr - amh - arb - arq - ary - eng - es - hau - hin - ind - kin - mar - pan - tel dataset_info: - config_name: afr features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: test num_bytes: 65243 num_examples: 375 - name: dev num_bytes: 66249 num_examples: 375 download_size: 95864 dataset_size: 131492 - config_name: amh features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: train num_bytes: 209475 num_examples: 992 - name: test num_bytes: 36637 num_examples: 171 - name: dev num_bytes: 19498 num_examples: 95 download_size: 153682 dataset_size: 265610 - config_name: arb features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: test num_bytes: 110473 num_examples: 595 - name: dev num_bytes: 5846 num_examples: 32 download_size: 72348 dataset_size: 116319 - config_name: arq features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: train num_bytes: 170025 num_examples: 1261 - name: test num_bytes: 79323 num_examples: 583 - name: dev num_bytes: 12181 num_examples: 97 download_size: 149472 dataset_size: 261529 - config_name: ary features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: train num_bytes: 382561 num_examples: 924 - name: test num_bytes: 175568 num_examples: 426 - name: dev num_bytes: 27975 num_examples: 71 download_size: 274828 dataset_size: 586104 - config_name: eng features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: train num_bytes: 844975 num_examples: 5500 - name: test num_bytes: 374647 num_examples: 2600 - name: dev num_bytes: 36697 num_examples: 250 download_size: 868674 dataset_size: 1256319 - config_name: esp features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: train num_bytes: 316713 num_examples: 1562 - name: test num_bytes: 123222 num_examples: 600 - name: dev num_bytes: 28981 num_examples: 140 download_size: 323584 dataset_size: 468916 - config_name: hau features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: train num_bytes: 403474 num_examples: 1736 - name: test num_bytes: 142238 num_examples: 603 - name: dev num_bytes: 49236 num_examples: 212 download_size: 328542 dataset_size: 594948 - config_name: hin features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: test num_bytes: 377385 num_examples: 968 - name: dev num_bytes: 113047 num_examples: 288 download_size: 217493 dataset_size: 490432 - config_name: ind features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: test num_bytes: 68185 num_examples: 360 - name: dev num_bytes: 26579 num_examples: 144 download_size: 68263 dataset_size: 94764 - config_name: kin features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: train num_bytes: 234520 num_examples: 778 - name: test num_bytes: 67211 num_examples: 222 - name: dev num_bytes: 30758 num_examples: 102 download_size: 219256 dataset_size: 332489 - config_name: mar features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: train num_bytes: 555224 num_examples: 1155 - name: test num_bytes: 139343 num_examples: 298 - name: dev num_bytes: 146496 num_examples: 293 download_size: 381039 dataset_size: 841063 - config_name: pan features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: test num_bytes: 307401 num_examples: 634 - name: dev num_bytes: 117984 num_examples: 242 download_size: 166402 dataset_size: 425385 - config_name: tel features: - name: sentence1 dtype: string - name: sentence2 dtype: string - name: label dtype: float64 splits: - name: train num_bytes: 561688 num_examples: 1146 - name: test num_bytes: 145249 num_examples: 297 - name: dev num_bytes: 64775 num_examples: 130 download_size: 347275 dataset_size: 771712 configs: - config_name: afr data_files: - split: test path: afr/test-* - split: dev path: afr/dev-* - config_name: amh data_files: - split: train path: amh/train-* - split: test path: amh/test-* - split: dev path: amh/dev-* - config_name: arb data_files: - split: test path: arb/test-* - split: dev path: arb/dev-* - config_name: arq data_files: - split: train path: arq/train-* - split: test path: arq/test-* - split: dev path: arq/dev-* - config_name: ary data_files: - split: train path: ary/train-* - split: test path: ary/test-* - split: dev path: ary/dev-* - config_name: eng data_files: - split: train path: eng/train-* - split: test path: eng/test-* - split: dev path: eng/dev-* - config_name: esp data_files: - split: train path: esp/train-* - split: test path: esp/test-* - split: dev path: esp/dev-* - config_name: hau data_files: - split: train path: hau/train-* - split: test path: hau/test-* - split: dev path: hau/dev-* - config_name: hin data_files: - split: test path: hin/test-* - split: dev path: hin/dev-* - config_name: ind data_files: - split: test path: ind/test-* - split: dev path: ind/dev-* - config_name: kin data_files: - split: train path: kin/train-* - split: test path: kin/test-* - split: dev path: kin/dev-* - config_name: mar data_files: - split: train path: mar/train-* - split: test path: mar/test-* - split: dev path: mar/dev-* - config_name: pan data_files: - split: test path: pan/test-* - split: dev path: pan/dev-* - config_name: tel data_files: - split: train path: tel/train-* - split: test path: tel/test-* - split: dev path: tel/dev-* task_categories: - text-classification - sentence-similarity --- ## Dataset Description - **Homepage:** https://semantic-textual-relatedness.github.io - **Repository:** [GitHub](https://github.com/semantic-textual-relatedness/Semantic_Relatedness_SemEval2024) - **Paper:** [SemRel2024: A Collection of Semantic Textual Relatedness Datasets for 14 Languages](https://arxiv.org/abs/2402.08638) - **Paper:** [SemEval Task 1: Semantic Textual Relatedness for African and Asian Languages](https://arxiv.org/pdf/2403.18933.pdf) - **Leaderboard:** https://codalab.lisn.upsaclay.fr/competitions/16799#results - **Point of Contact:** [Nedjma Ousidhoum](mailto:nedjma.ousidhoum@gmail.com) ### Dataset Summary SemRel2024 is a collection of Semantic Textual Relatedness (STR) datasets for 14 languages, including African and Asian languages. The datasets are composed of sentence pairs, each assigned a relatedness score between 0 (completely) unrelated and 1 (maximally related) with a large range of expected relatedness values. SemRel2024 dataset was used as part of the SemEval2024 shared task 1. The task aims to evaluate the ability of systems to measure the semantic relatedness between two sentences. ### Languages The SemRel2024 dataset covers the following 14 languages: 1. Afrikaans (_afr_) 2. Algerian Arabic (_arq_) 3. Amharic (_amh_) 4. English (_eng_) 5. Hausa (_hau_) 6. Indonesian (_ind_) 7. Hindi (_hin_) 8. Kinyarwanda (_kin_) 9. Marathi (_mar_) 10. Modern Standard Arabic (_arb_) 11. Moroccan Arabic (_ary_) 12. Punjabi (_pan_) 13. Spanish (_esp_) 14. Telugu (_tel_) **Note**: Spanish test labels are all -1 because the Spanish team retained the gold test labels to avoid contamination problems in future benchmarking. We refer to the [CodaLab contest website](https://codalab.lisn.upsaclay.fr/competitions/15715) to evaluate your predictions, which will remain open. ## Dataset Structure ### Data Instances Each instance in the dataset consists of two text segments and a relatedness score indicating the degree of semantic relatedness between them. ``` { "sentence1": "string", "sentence2": "string", "label": float } ``` - sentence1: a string feature representing the first text segment. - sentence2: a string feature representing the second text segment. - label: a float value representing the semantic relatedness score between sentence1 and sentence2, typically ranging from 0 (not related at all) to 1 (highly related). ## Citation Information If you use the SemRel2024 dataset in your research, please cite the following papers: ``` @misc{ousidhoum2024semrel2024, title={SemRel2024: A Collection of Semantic Textual Relatedness Datasets for 14 Languages}, author={Nedjma Ousidhoum and Shamsuddeen Hassan Muhammad and Mohamed Abdalla and Idris Abdulmumin and Ibrahim Said Ahmad and Sanchit Ahuja and Alham Fikri Aji and Vladimir Araujo and Abinew Ali Ayele and Pavan Baswani and Meriem Beloucif and Chris Biemann and Sofia Bourhim and Christine De Kock and Genet Shanko Dekebo and Oumaima Hourrane and Gopichand Kanumolu and Lokesh Madasu and Samuel Rutunda and Manish Shrivastava and Thamar Solorio and Nirmal Surange and Hailegnaw Getaneh Tilaye and Krishnapriya Vishnubhotla and Genta Winata and Seid Muhie Yimam and Saif M. Mohammad}, year={2024}, eprint={2402.08638}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` ``` @inproceedings{ousidhoum-etal-2024-semeval, title = "{S}em{E}val-2024 Task 1: Semantic Textual Relatedness for African and Asian Languages", author = "Ousidhoum, Nedjma and Muhammad, Shamsuddeen Hassan and Abdalla, Mohamed and Abdulmumin, Idris and Ahmad,Ibrahim Said and Ahuja, Sanchit and Aji, Alham Fikri and Araujo, Vladimir and Beloucif, Meriem and De Kock, Christine and Hourrane, Oumaima and Shrivastava, Manish and Solorio, Thamar and Surange, Nirmal and Vishnubhotla, Krishnapriya and Yimam, Seid Muhie and Mohammad, Saif M.", booktitle = "Proceedings of the 18th International Workshop on Semantic Evaluation (SemEval-2024)", year = "2024", publisher = "Association for Computational Linguistics" } ```