{
 "id": "mteb/SciDocsRR",
 "owner": "mteb",
 "name": "SciDocsRR",
 "downloads": 9702,
 "likes": 0,
 "last_modified": "2025-05-04T16:11:52.000Z",
 "stated_licence": "cc-by-4.0",
 "stated_from": "tag",
 "licence_name": null,
 "licence_link": null,
 "file_licence": null,
 "licence_files": [],
 "sources": [
  {
   "id": "mteb/scidocs",
   "licence": "cc-by-sa-4.0",
   "found": true
  }
 ],
 "gated": false,
 "tasks": [
  "text-ranking"
 ],
 "citation": "@inproceedings{cohan-etal-2020-specter,\n  abstract = {Representation learning is a critical ingredient for natural language processing systems. Recent Transformer language models like BERT learn powerful textual representations, but these models are targeted towards token- and sentence-level training objectives and do not leverage information on inter-document relatedness, which limits their document-level representation power. For applications on scientific documents, such as classification and recommendation, accurate embeddings of documents are a necessity. We propose SPECTER, a new method to generate document-level embedding of scientific papers based on pretraining a Transformer language model on a powerful signal of document-level relatedness: the citation graph. Unlike existing pretrained language models, Specter can be easily applied to downstream applications without task-specific fine-tuning. Additionally, to encourage further research on document-level models, we introduce SciDocs, a new evaluation benchmark consisting of seven document-level tasks ranging from citation prediction, to document classification and recommendation. We show that Specter outperforms a variety of competitive baselines on the benchmark.},\n  address = {Online},\n  author = {Cohan, Arman  and\nFeldman, Sergey  and\nBeltagy, Iz  and\nDowney, Doug  and\nWeld, Daniel},\n  booktitle = {Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics},\n  doi = {10.18653/v1/2020.a",
 "signal": "conflict",
 "reason": "The declared licence (cc-by-4.0) may conflict with a source dataset's reported share-alike licence (cc-by-sa-4.0)."
}