https://media.commonshub.social/videos/watch/4a2b1df9-5c72-45d3-9b31-4890c6198542 Davide Eynard argues that timeline algorithms should serve users' goals rather than maximize platform engagement. The talk demonstrates a personal, local approach to search, clustering, and recommendation using Mastodon data and locally computed embeddings, allowing posts to be analyzed without leaving the user's computer. Davide Eynard — FOSDEM 2025, Social Web Devroom Original recording courtesy of FOSDEM. Licensed CC BY 2.0 Belgium. Source: FOSDEM event page Sun, 04 Oct 2026 20:11:46 GMT https://validator.w3.org/feed/docs/rss2.html PeerTube - https://media.commonshub.social https://media.commonshub.social/lazy-static/avatars/accbb06d-5e11-4bda-b0ea-ae4aee8f29d1.png https://media.commonshub.social/videos/watch/4a2b1df9-5c72-45d3-9b31-4890c6198542 All rights reserved, unless otherwise specified in the terms specified at https://media.commonshub.social/about and potential licenses granted by each content's rightholder.