{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,19]],"date-time":"2026-05-19T07:16:18Z","timestamp":1779174978874,"version":"3.51.4"},"reference-count":48,"publisher":"MIT Press","license":[{"start":{"date-parts":[[2024,11,7]],"date-time":"2024-11-07T00:00:00Z","timestamp":1730937600000},"content-version":"vor","delay-in-days":311,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["direct.mit.edu"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2024,11,4]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>Topic taxonomy discovery aims at uncovering topics of different abstraction levels and constructing hierarchical relations between them. Unfortunately, most prior work can hardly model semantic scopes of words and topics by holding the Euclidean embedding space assumption. What\u2019s worse, they infer asymmetric hierarchical relations by symmetric distances between topic embeddings. As a result, existing methods suffer from problems of low-quality topics at high abstraction levels and inaccurate hierarchical relations. To alleviate these problems, this paper develops a Box embedding-based Topic Model (BoxTM) that maps words and topics into the box embedding space, where the asymmetric metric is defined to properly infer hierarchical relations among topics. Additionally, our BoxTM explicitly infers upper-level topics based on correlation between specific topics through recursive clustering on topic boxes. Finally, extensive experiments validate high-quality of the topic taxonomy learned by BoxTM.<\/jats:p>","DOI":"10.1162\/tacl_a_00712","type":"journal-article","created":{"date-parts":[[2024,11,7]],"date-time":"2024-11-07T20:18:44Z","timestamp":1731010724000},"page":"1401-1416","update-policy":"https:\/\/doi.org\/10.1162\/mitpressjournals.corrections.policy","source":"Crossref","is-referenced-by-count":4,"title":["Self-supervised Topic Taxonomy Discovery in the Box Embedding Space"],"prefix":"10.1162","volume":"12","author":[{"given":"Yuyin","family":"Lu","sequence":"first","affiliation":[{"name":"School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China. luyy37@mail2.sysu.edu.cn"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Hegang","family":"Chen","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China. chenhg25@mail2.sysu.edu.cn"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Pengbo","family":"Mao","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China. maopb@mail2.sysu.edu.cn"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yanghui","family":"Rao","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China. raoyangh@mail.sysu.edu.cn"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Haoran","family":"Xie","sequence":"additional","affiliation":[{"name":"School of Data Science, Lingnan University, Hong Kong SAR. hrxie@ieee.org"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Fu Lee","family":"Wang","sequence":"additional","affiliation":[{"name":"School of Science and Technology, Hong Kong Metropolitan University, Hong Kong SAR. pwang@hkmu.edu.hk"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Qing","family":"Li","sequence":"additional","affiliation":[{"name":"Department of Computing, the Hong Kong Polytechnic University, Hong Kong SAR. csqli@comp.polyu.edu.hk"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"281","published-online":{"date-parts":[[2024,11,4]]},"reference":[{"key":"2024110720183674100_bib1","first-page":"9649","article-title":"Boxe: A box embedding model for knowledge base completion","volume-title":"NeurIPS","author":"Abboud","year":"2020"},{"key":"2024110720183674100_bib2","first-page":"1027","article-title":"k-means++: The advantages of careful seeding","volume-title":"SODA","author":"Arthur","year":"2007"},{"issue":"2","key":"2024110720183674100_bib3","doi-asserted-by":"publisher","first-page":"261","DOI":"10.1093\/BIOMET\/67.2.261","article-title":"Logistic-normal distributions: Some properties and uses","volume":"67","author":"Atchison","year":"1980","journal-title":"Biometrika"},{"key":"2024110720183674100_bib4","first-page":"12316","article-title":"Modeling heterogeneous hierarchies with relation-specific hyperbolic cones","volume-title":"NeurIPS","author":"Bai","year":"2021"},{"key":"2024110720183674100_bib5","first-page":"17","article-title":"Hierarchical topic models and the nested Chinese restaurant process","volume-title":"NIPS","author":"Blei","year":"2003"},{"key":"2024110720183674100_bib6","first-page":"993","article-title":"Latent Dirichlet allocation","volume":"3","author":"Blei","year":"2003","journal-title":"Journal of Machine Learning Research"},{"key":"2024110720183674100_bib7","doi-asserted-by":"publisher","first-page":"155","DOI":"10.1007\/978-3-642-80328-4_13","article-title":"Coefficient of variation","volume-title":"Applied Multivariate Statistics in Geohydrology and Related Sciences","author":"Brown","year":"1998"},{"key":"2024110720183674100_bib8","first-page":"288","article-title":"Reading tea leaves: How humans interpret topic models","volume-title":"NIPS","author":"Chang","year":"2009"},{"key":"2024110720183674100_bib9","doi-asserted-by":"publisher","first-page":"2139","DOI":"10.1007\/s11280-021-00963-7","article-title":"Hierarchical neural topic modeling with manifold regularization","volume":"24","author":"Chen","year":"2021","journal-title":"World Wide Web"},{"key":"2024110720183674100_bib10","doi-asserted-by":"publisher","first-page":"2343","DOI":"10.18653\/v1\/2021.acl-long.182","article-title":"Tree-structured topic modeling with nonparametric neural variational inference","volume-title":"ACL\/IJCNLP","author":"Chen","year":"2021"},{"key":"2024110720183674100_bib11","first-page":"182","article-title":"Improving local identifiability in probabilistic box embeddings","volume-title":"NeurIPS","author":"Dasgupta","year":"2020"},{"key":"2024110720183674100_bib12","first-page":"4171","article-title":"BERT: Pre-training of deep bidirectional transformers for language understanding","volume-title":"NAACL-HLT","author":"Devlin","year":"2019"},{"key":"2024110720183674100_bib13","doi-asserted-by":"publisher","first-page":"439","DOI":"10.1162\/tacl_a_00325","article-title":"Topic modeling in embedding spaces","volume":"8","author":"Dieng","year":"2020","journal-title":"Transactions of the Association for Computational Linguistics"},{"key":"2024110720183674100_bib14","first-page":"2903","article-title":"Sawtooth factorial topic embeddings guided gamma belief network","volume-title":"ICML","author":"Duan","year":"2021"},{"key":"2024110720183674100_bib15","first-page":"547","article-title":"Topicnet: Semantic graph-guided topic discovery","volume-title":"NeurIPS","author":"Duan","year":"2021"},{"key":"2024110720183674100_bib16","first-page":"226","article-title":"A density-based algorithm for discovering clusters in large spatial databases with noise","volume-title":"KDD","author":"Ester","year":"1996"},{"issue":"5814","key":"2024110720183674100_bib17","doi-asserted-by":"publisher","first-page":"972","DOI":"10.1126\/science.1136800","article-title":"Clustering by passing messages between data points","volume":"315","author":"Frey","year":"2007","journal-title":"Science"},{"key":"2024110720183674100_bib18","article-title":"Bertopic: Neural topic modeling with a class-based tf-idf procedure","author":"Grootendorst","year":"2022","journal-title":"CoRR"},{"key":"2024110720183674100_bib19","doi-asserted-by":"publisher","first-page":"869","DOI":"10.1109\/CVPR.2019.00096","article-title":"Scops: Self-supervised co-part segmentation","volume-title":"CVPR","author":"Hung","year":"2019"},{"key":"2024110720183674100_bib20","doi-asserted-by":"publisher","first-page":"800","DOI":"10.18653\/v1\/2020.acl-main.73","article-title":"Tree-structured neural topic model","volume-title":"ACL","author":"Isonuma","year":"2020"},{"key":"2024110720183674100_bib21","doi-asserted-by":"publisher","first-page":"925","DOI":"10.1145\/3485447.3511935","article-title":"Taxoenrich: Self-supervised taxonomy completion via structure-semantic representations","volume-title":"WWW","author":"Jiang","year":"2022"},{"key":"2024110720183674100_bib22","doi-asserted-by":"publisher","first-page":"783","DOI":"10.1145\/2396761.2396861","article-title":"Modeling topic hierarchies with the recursive chinese restaurant process","volume-title":"CIKM","author":"Kim","year":"2012"},{"key":"2024110720183674100_bib23","doi-asserted-by":"publisher","first-page":"530","DOI":"10.3115\/v1\/E14-1056","article-title":"Machine reading tea leaves: Automatically evaluating topic coherence and topic model quality","volume-title":"EACL","author":"Lau","year":"2014"},{"key":"2024110720183674100_bib24","doi-asserted-by":"publisher","first-page":"2819","DOI":"10.1145\/3485447.3512002","article-title":"Taxocom: Topic taxonomy completion with hierarchical discovery of novel topic clusters","volume-title":"WWW","author":"Lee","year":"2022"},{"key":"2024110720183674100_bib25","doi-asserted-by":"publisher","first-page":"1279","DOI":"10.18653\/v1\/2020.coling-main.110","article-title":"Embedding semantic taxonomies","volume-title":"COLING","author":"Lees","year":"2020"},{"key":"2024110720183674100_bib26","article-title":"Smoothing the geometry of probabilistic box embeddings","volume-title":"ICLR","author":"Li","year":"2019"},{"issue":"5","key":"2024110720183674100_bib27","doi-asserted-by":"publisher","first-page":"365","DOI":"10.1037\/h0031643","article-title":"Measures of response agreement for qualitative data: Some generalizations and alternatives","volume":"76","author":"Light","year":"2011","journal-title":"Psychological Bulletin"},{"key":"2024110720183674100_bib28","doi-asserted-by":"publisher","first-page":"2185","DOI":"10.18653\/v1\/D18-1242","article-title":"Knowledge base question answering via encoding of complex query graphs","volume-title":"EMNLP","author":"Luo","year":"2018"},{"key":"2024110720183674100_bib29","doi-asserted-by":"publisher","first-page":"6826","DOI":"10.1007\/978-3-031-01914-2_5","article-title":"Weakly-supervised hierarchical text classification","volume-title":"AAAI","author":"Meng","year":"2019"},{"key":"2024110720183674100_bib30","doi-asserted-by":"publisher","first-page":"3143","DOI":"10.1145\/3485447.3512034","article-title":"Topic discovery via latent space clustering of pretrained language model representations","volume-title":"WWW","author":"Meng","year":"2022"},{"key":"2024110720183674100_bib31","doi-asserted-by":"publisher","first-page":"1908","DOI":"10.1145\/3394486.3403242","article-title":"Hierarchical topic mining via joint spherical tree and text embedding","volume-title":"KDD","author":"Meng","year":"2020"},{"issue":"11","key":"2024110720183674100_bib32","doi-asserted-by":"publisher","first-page":"39","DOI":"10.1145\/219717.219748","article-title":"Wordnet: A lexical database for English","volume":"38","author":"Miller","year":"1995","journal-title":"Communications of the ACM"},{"key":"2024110720183674100_bib33","doi-asserted-by":"publisher","first-page":"633","DOI":"10.1145\/1273496.1273576","article-title":"Mixtures of hierarchical topics with pachinko allocation","volume-title":"ICML","author":"Mimno","year":"2007"},{"key":"2024110720183674100_bib34","doi-asserted-by":"publisher","first-page":"6345","DOI":"10.18653\/v1\/P19-1640","article-title":"Topic modeling with wasserstein autoencoders","volume-title":"ACL","author":"Nan","year":"2019"},{"key":"2024110720183674100_bib35","first-page":"6341","article-title":"Poincar\u00e9 embeddings for learning hierarchical representations","volume-title":"NIPS","author":"Nickel","year":"2017"},{"key":"2024110720183674100_bib36","doi-asserted-by":"publisher","first-page":"1532","DOI":"10.3115\/v1\/D14-1162","article-title":"Glove: Global vectors for word representation","volume-title":"EMNLP","author":"Pennington","year":"2014"},{"key":"2024110720183674100_bib37","first-page":"1278","article-title":"Stochastic backpropagation and approximate inference in deep generative models","volume-title":"ICML","author":"Rezende","year":"2014"},{"key":"2024110720183674100_bib38","doi-asserted-by":"publisher","first-page":"1728","DOI":"10.18653\/v1\/2020.emnlp-main.135","article-title":"Tired of topic models? Clusters of pretrained word embeddings make for fast and good topics too!","volume-title":"EMNLP","author":"Sia","year":"2020"},{"key":"2024110720183674100_bib39","doi-asserted-by":"publisher","first-page":"1579","DOI":"10.1609\/aaai.v28i1.8938","article-title":"On dataless hierarchical text classification","volume-title":"AAAI","author":"Song","year":"2014"},{"key":"2024110720183674100_bib40","article-title":"Cluster analysis: Basic concepts and algorithms","volume-title":"Introduction to Data Mining","author":"Steinbach","year":"2005","edition":"1st edn."},{"key":"2024110720183674100_bib41","doi-asserted-by":"publisher","first-page":"8138","DOI":"10.18653\/v1\/2020.acl-main.724","article-title":"Cluhtm - semantic hierarchical topic modeling based on cluwords","volume-title":"ACL","author":"Viegas","year":"2020"},{"key":"2024110720183674100_bib42","doi-asserted-by":"publisher","DOI":"10.7275\/20638273","volume-title":"Geometric representation learning","author":"Vilnis","year":"2021"},{"key":"2024110720183674100_bib43","doi-asserted-by":"publisher","first-page":"263","DOI":"10.18653\/v1\/P18-1025","article-title":"Probabilistic embedding of knowledge graphs with box lattice measures","volume-title":"ACL","author":"Vilnis","year":"2018"},{"key":"2024110720183674100_bib44","doi-asserted-by":"publisher","first-page":"936","DOI":"10.1137\/1.9781611977653.ch105","article-title":"Hierarchical neural topic model with embedding cluster and neural variational inference","volume-title":"SDM","author":"Wang","year":"2023"},{"key":"2024110720183674100_bib45","first-page":"37335","article-title":"Effective neural topic modeling with embedding clustering regularization","volume-title":"ICML","author":"Xiaobao","year":"2023"},{"key":"2024110720183674100_bib46","doi-asserted-by":"publisher","first-page":"4226","DOI":"10.1145\/3534678.3539125","article-title":"Contrastive cross-domain recommendation in matching","volume-title":"KDD","author":"Xie","year":"2022"},{"key":"2024110720183674100_bib47","first-page":"31557","article-title":"Hyperminer: Topic taxonomy mining with hyperbolic embedding","volume-title":"NeurIPS","author":"Yishi","year":"2022"},{"key":"2024110720183674100_bib48","doi-asserted-by":"publisher","first-page":"2701","DOI":"10.1145\/3219819.3220064","article-title":"Taxogen: Unsupervised topic taxonomy construction by adaptive term embedding and clustering","volume-title":"KDD","author":"Zhang","year":"2018"}],"container-title":["Transactions of the Association for Computational Linguistics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/direct.mit.edu\/tacl\/article-pdf\/doi\/10.1162\/tacl_a_00712\/2478615\/tacl_a_00712.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/direct.mit.edu\/tacl\/article-pdf\/doi\/10.1162\/tacl_a_00712\/2478615\/tacl_a_00712.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,11,7]],"date-time":"2024-11-07T20:18:48Z","timestamp":1731010728000},"score":1,"resource":{"primary":{"URL":"https:\/\/direct.mit.edu\/tacl\/article\/doi\/10.1162\/tacl_a_00712\/125175\/Self-supervised-Topic-Taxonomy-Discovery-in-the"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024]]},"references-count":48,"URL":"https:\/\/doi.org\/10.1162\/tacl_a_00712","relation":{},"ISSN":["2307-387X"],"issn-type":[{"value":"2307-387X","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2024]]},"published":{"date-parts":[[2024]]}}}