New DUMoE framework models evolving user interests on social media

Researchers have developed DUMoE, a new framework designed to learn user representations from social media data that accounts for evolving preferences. The model addresses 'interest drift' by integrating static profiles, short-term behaviors, and long-term dependencies. It also employs a sparse mixture-of-experts approach to disentangle multiple user interests, with experiments showing its superiority over existing methods in predicting user interests and interactions. AI

IMPACT This research could improve personalization and recommendation systems by better understanding and predicting user behavior shifts.

RANK_REASON The cluster contains a single academic paper detailing a new model and its experimental results. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX
• Connected Papers
• CORE Recommender
• DagsHub
• DUMoE
• Gotit.pub
• Hugging Face
• IArxiv Recommender
• Litmaps
• ScienceCast
• scite Smart Citations

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →