Je remets une couche, mais si vous êtes lassés des LDA et voulez explorer des thèmes dans un corpus de textes, #Bertopic en #Python est une solution ++

@axlmrin.bsky.social a fait une super leçon orientée sciences sociales www.css.cnrs.fr/the-general-...

(+ dispo sur #ActiveTigger en no code :)

🚨 update du mardi soir #ActiveTigger www.css.cnrs.fr/active-tigger/

On a essayé de rendre l'ensemble un peu plus ergonomique... à vous de nous dire si nous avons réussi.

#annotation #bertopic #finetune #computationalsocialsciences

LLM-Assisted Topic Reduction Boosts BERTopic for Social Media Data

Researchers merged BERTopic topics using LLMs, testing on three Twitter/X datasets with four LLMs and noting fewer, more distinct topics; the study was submitted on 18 Sep 2025. getnews.me/llm-assisted-topic-reduc... #bertopic #llm #socialmedia

Had a great time presenting my poster exploring people's thoughts & feelings about death w/ #BERTopic this evening at the Society for Research in Psychopathology Annual Meeting! Thanks to everyone who stopped by to chat– it was wonderful learning more about others' work and nerding out! 💀🪦
#SRP2025

Week 3: Stable Topics with BERTopic / Articles / dracoblue.net In Week 1 (extraction) and Week 2 (embeddings + KMeans in BigQuery ML) we laid the groundwork. This week I built a Python BERTopic stage whose IDs stay stable across runs by mapping BERTopic’s interna...

TopicWatchdog – Week 3: KMeans worked, but cluster IDs kept jumping across retrains. This week I added a Python BERTopic stage with a BigQuery registry → stable topic IDs!

Blog: dracoblue.net/dev/topicwat...

#TopicWatchdog #BERTopic #BigQuery #Clustering #MachineLearning #FediScience

Incoming feature for #ActiveTigger : compute easily exploratory #bertopic on your data to identify topics before moving on to annotation task