Search papers, labs, and topics across Lattice.
This paper introduces UniDot, a unified architecture that integrates feature-interaction models and sequential models for post-click conversion prediction in recommendation systems. By leveraging a shared token space and a novel macro-block design that facilitates parallel processing of feature interactions and user behavior sequences, UniDot achieves efficient inference while maintaining high predictive performance. The model's innovative use of a dual optimizer and multi-path mutual learning enabled it to secure the runner-up position in the Industrial track of the TAAC KDD Cup 2026, demonstrating its effectiveness in real-world applications.
UniDot's architecture unifies feature interactions and sequential modeling, achieving state-of-the-art performance in industrial recommendation tasks with unprecedented efficiency.
Industrial recommenders rely on two model families that have evolved largely independently: feature-interaction models over multi-field user/item features, and sequential models over user-behavior histories. Production systems couple them only loosely. To unify the two, we present UniDot, a novel architecture for post-click conversion prediction built from the factorization-machine (FM) point of view: the embedding inner product---which powers collaborative filtering and lets a recommender generalize to unseen user--item pairs---is the same primitive as attention's query dot key scoring, so a single dot-product of tokens can underlie both feature interaction and sequence modeling. UniDot tokenizes non-sequential fields and multi-domain behavioral sequences into one shared token space and stacks a single macro-block in which a token-mixing bus and a sequence-retrieval bus (item tokens cross-attending the histories) run in parallel and exchange state each layer through an MLP-Mixer fusion, while an FM Highway carries explicit per-layer dot-product interactions around the residual stack directly to the classifier. The sequence side is embedded once per forward pass and shared by all consumers, bounding inference latency. Trained with a dual sparse/dense (Adagrad + Muon) optimizer, an auxiliary conversion-delay head, and multi-path mutual learning, UniDot finished as the runner-up on the Industrial track of the TAAC KDD Cup 2026.
