← Back to the feed
AI & softwareResearch findings

UniVA: A Unified Value Alignment Framework for Advertising Recommendation

Original published 2026.05.07 16:41 KSTAdded to Blackdeck 2026.10.21 00:33 KST

What has changed?

UniVA introduces a unified framework that jointly considers user relevance and commercial value in advertising recommendations

Key points

  1. UniVA proposes a unified framework that jointly considers user relevance and commercial value in advertising recommendations.
  2. Commercial SID tokenization and value-aware decoding integrate business attributes and bid information into the generation process.
  3. Value-aware constrained serving restricts decoding to valid request paths using personalized tries, ensuring high-value ad visibility.

Connect it to your work

UniVA aligns commercial value across SID construction, decoding, and online serving processes

What remains unproven

This research note is based on an abstract; check the original for test conditions and application scope.

Originals used for this note

Links open the original publisher in a new tab.

  1. arxivUniVA: Unified Value Alignment for Generative Recommendation in Online Advertising at Tencent