{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/mining-discourse-markers-for-unsupervised","title":"Mining Discourse Markers for Unsupervised Sentence Representation Learning","arxiv_id":"1903.11850","date":"2019-03-28","proceeding":"NAACL 2019 6","authors":["Damien Sileo","Tim Van-De-Cruys","Camille Pradel","Philippe Muller"],"abstract":"Current state of the art systems in NLP heavily rely on manually annotated\ndatasets, which are expensive to construct. Very little work adequately\nexploits unannotated data -- such as discourse markers between sentences --\nmainly because of data sparseness and ineffective extraction methods. In the\npresent work, we propose a method to automatically discover sentence pairs with\nrelevant discourse markers, and apply it to massive amounts of data. Our\nresulting dataset contains 174 discourse markers with at least 10k examples\neach, even for rare markers such as coincidentally or amazingly We use the\nresulting data as supervision for learning transferable sentence embeddings. In\naddition, we show that even though sentence representation learning through\nprediction of discourse markers yields state of the art results across\ndifferent transfer tasks, it is not clear that our models made use of the\nsemantic relation between sentences, thus leaving room for further\nimprovements. Our datasets are publicly available\n(https://github.com/synapse-developpement/Discovery)","url_abs":"http://arxiv.org/abs/1903.11850v1","url_pdf":"http://arxiv.org/pdf/1903.11850v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"mining-discourse-markers-for-unsupervised","repo_url":"https://github.com/synapse-developpement/Discovery","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"relation-classification","task_name":"Relation Classification"},{"task_slug":"representation-learning","task_name":"Representation Learning"},{"task_slug":"sentence","task_name":"Sentence"},{"task_slug":"sentence-embeddings","task_name":"Sentence Embeddings"}],"methods":[],"datasets_introduced":[{"slug":"discovery","name":"Discovery","full_name":""}],"methods_introduced":[],"results":[{"leaderboard":"/sota/relation-classification-on-discovery-dataset","task":"Relation Classification","dataset":"Discovery","model":"BERT","rank_in_archive_order":1,"of":1,"metrics":{"1:1 Accuracy":"20.6"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1903.11850","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}