{"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/evaluating-unsupervised-text-classification","title":"Evaluating Unsupervised Text Classification: Zero-shot and Similarity-based Approaches","arxiv_id":"2211.16285","date":"2022-11-29","proceeding":null,"authors":["Tim Schopf","Daniel Braun","Florian Matthes"],"abstract":"Text classification of unseen classes is a challenging Natural Language Processing task and is mainly attempted using two different types of approaches. Similarity-based approaches attempt to classify instances based on similarities between text document representations and class description representations. Zero-shot text classification approaches aim to generalize knowledge gained from a training task by assigning appropriate labels of unknown classes to text documents. Although existing studies have already investigated individual approaches to these categories, the experiments in literature do not provide a consistent comparison. This paper addresses this gap by conducting a systematic evaluation of different similarity-based and zero-shot approaches for text classification of unseen classes. Different state-of-the-art approaches are benchmarked on four text classification datasets, including a new dataset from the medical domain. Additionally, novel SimCSE and SBERT-based baselines are proposed, as other baselines used in existing work yield weak classification results and are easily outperformed. Finally, the novel similarity-based Lbl2TransformerVec approach is presented, which outperforms previous state-of-the-art approaches in unsupervised text classification. Our experiments show that similarity-based approaches significantly outperform zero-shot approaches in most cases. Additionally, using SimCSE or SBERT embeddings instead of simpler text representations increases similarity-based classification results even further.","url_abs":"https://arxiv.org/abs/2211.16285v2","url_pdf":"https://arxiv.org/pdf/2211.16285v2.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":"evaluating-unsupervised-text-classification","repo_url":"https://github.com/sebischair/lbl2vec","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"BSD-3-Clause"}},{"paper_slug":"evaluating-unsupervised-text-classification","repo_url":"https://github.com/sebischair/medical-abstracts-tc-corpus","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"text-classification","task_name":"Text Classification"},{"task_slug":"unsupervised-text-classification","task_name":"Unsupervised Text Classification"},{"task_slug":"zero-shot-text-classification","task_name":"Zero-Shot Text Classification"},{"task_slug":"text-classification-1","task_name":"text-classification"}],"methods":[{"method_slug":"lbl2transformervec","method_name":"Lbl2TransformerVec"},{"method_slug":"lbl2vec","method_name":"Lbl2Vec"},{"method_slug":"sbert","method_name":"SBERT"},{"method_slug":"simcse","method_name":"SimCSE"},{"method_slug":"skip-gram-word2vec","method_name":"Skip-gram Word2Vec"}],"datasets_introduced":[{"slug":"medical-abstracts","name":"Medical Abstracts","full_name":"Medical Abstracts Text Classification Dataset"}],"methods_introduced":[{"slug":"lbl2transformervec","name":"Lbl2TransformerVec","full_name":"Lbl2TransformerVec"}],"results":[{"leaderboard":"/sota/unsupervised-text-classification-on-1","task":"Unsupervised Text Classification","dataset":"20NewsGroups","model":"Lbl2TransformerVec","rank_in_archive_order":2,"of":2,"metrics":{"F1-score":"64,69"},"uses_additional_data":false},{"leaderboard":"/sota/unsupervised-text-classification-on-ag-news","task":"Unsupervised Text Classification","dataset":"AG News","model":"Lbl2TransformerVec","rank_in_archive_order":2,"of":2,"metrics":{"F1-score":"83,79"},"uses_additional_data":false},{"leaderboard":"/sota/unsupervised-text-classification-on-medical","task":"Unsupervised Text Classification","dataset":"Medical Abstracts","model":"Lbl2TransformerVec","rank_in_archive_order":1,"of":2,"metrics":{"F1-score":"56.46"},"uses_additional_data":false},{"leaderboard":"/sota/unsupervised-text-classification-on-medical","task":"Unsupervised Text Classification","dataset":"Medical Abstracts","model":"Lbl2Vec","rank_in_archive_order":2,"of":2,"metrics":{"F1-score":"43.03"},"uses_additional_data":false},{"leaderboard":"/sota/unsupervised-text-classification-on-yahoo","task":"Unsupervised Text Classification","dataset":"Yahoo! Answers","model":"Lbl2TransformerVec","rank_in_archive_order":1,"of":1,"metrics":{"F1-score":"55.84"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2211.16285","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}