{"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/measuring-lda-topic-stability-from-clusters","title":"Measuring LDA Topic Stability from Clusters of Replicated Runs","arxiv_id":"1808.08098","date":"2018-08-24","proceeding":null,"authors":["Mika Mäntylä","Maëlick Claes","Umar Farooq"],"abstract":"Background: Unstructured and textual data is increasing rapidly and Latent\nDirichlet Allocation (LDA) topic modeling is a popular data analysis methods\nfor it. Past work suggests that instability of LDA topics may lead to\nsystematic errors. Aim: We propose a method that relies on replicated LDA runs,\nclustering, and providing a stability metric for the topics. Method: We\ngenerate k LDA topics and replicate this process n times resulting in n*k\ntopics. Then we use K-medioids to cluster the n*k topics to k clusters. The k\nclusters now represent the original LDA topics and we present them like normal\nLDA topics showing the ten most probable words. For the clusters, we try\nmultiple stability metrics, out of which we recommend Rank-Biased Overlap,\nshowing the stability of the topics inside the clusters. Results: We provide an\ninitial validation where our method is used for 270,000 Mozilla Firefox commit\nmessages with k=20 and n=20. We show how our topic stability metrics are\nrelated to the contents of the topics. Conclusions: Advances in text mining\nenable us to analyze large masses of text in software engineering but\nnon-deterministic algorithms, such as LDA, may lead to unreplicable\nconclusions. Our approach makes LDA stability transparent and is also\ncomplementary rather than alternative to many prior works that focus on LDA\nparameter tuning.","url_abs":"http://arxiv.org/abs/1808.08098v1","url_pdf":"http://arxiv.org/pdf/1808.08098v1.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":"measuring-lda-topic-stability-from-clusters","repo_url":"https://github.com/M3SOulu/Measuring-LDA-Topic-Stability","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"clustering","task_name":"Clustering"}],"methods":[{"method_slug":"lda","method_name":"LDA"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}