Digest: Personalized Task Dependency Graphs for Mitigating Signal Erosion in Multi-Task Recommendation
Personalized Task Dependency Graphs for Mitigating Signal Erosion in Multi-Task Recommendation
A CIKM 2026 paper on personalized task dependency graphs for mitigating signal erosion in multi-task recommendation.
Research area
User reactivation, generative recommendation, Semantic-ID tokenizer diagnostics, dynamic retrieval, graph embedding, counterfactual watch-time prediction, and efficient CTR modeling form the main technical line.
Publication record
- Authors
- F Liu, T Wu, Y Fang, W Zhou, Z Zhou, W Chen, Q Mei, J Zhou, H Chang, Y Cao, J Ding
- Venue
- Proceedings of the ACM International Conference on Information and Knowledge Management 2026
- Year
- 2026
- Area
- Recommender Systems
Problem
Multi-task recommendation can suffer when learning signals from one task erode or distort the signal needed by another task.
Approach
The paper studies personalized task dependency graphs as a way to model task relationships at a finer granularity than a single global dependency structure.
Scope
The public record should be read as a CIKM 2026 publication; implementation details and non-public evaluation material are not repeated on the homepage.
Takeaway
The work belongs to the recommendation line on reliable multi-objective modeling under changing user, item, and task signals.