DIIT: A Domain-Invariant Information Transfer Method for Industrial Cross-Domain Recommendation

TL;DR

DIIT, an end-to-end Domain-Invariant Information Transfer method for industrial cross-domain recommendation, is proposed, which designs two extractors to fully extract domain-invariant information from the latest source domain models at the domain level and the representation level respectively.

Abstract

Cross-Domain Recommendation (CDR) have received widespread attention due to their ability to utilize rich information across domains. However, most existing CDR methods assume an ideal static condition that is not practical in industrial recommendation systems (RS). Therefore, simply applying existing CDR methods in the industrial RS environment may lead to low effectiveness and efficiency. To fill this gap, we propose DIIT, an end-to-end Domain-Invariant Information Transfer method for industrial cross-domain recommendation. Specifically, We first simulate the industrial RS environment that maintains respective models in multiple domains, each of them is trained in the incremental mode. Then, for improving the effectiveness, we design two extractors to fully extract domain-invariant information from the latest source domain models at the domain level and the representation level respectively. Finally, for improving the efficiency, we design a migrator to transfer the extracted information to the latest target domain model, which only need the target domain model for inference. Experiments conducted on one production dataset and two public datasets verify the effectiveness and efficiency of DIIT.

Figures (6)

• Figure 1: Visualization of the data distribution shift over domains (a) and time (b) using PCA. Area under curve represents CTR, and the data is collected from our production system.
• Figure 2: Model Architecture of the proposed DIIT. DIIT consists of three parts: A warm start module, a domain-invariant information extractor module and a domain-invariant information migrator module. Note that only the target domain model in the bottom right region is used for inference. The solid arrows in violet and black represent the sample flow of the mix dataset and the target domain dataset respectively. The larger arrows in orange and white represent incremental update of the source domain models and initialization of the target domain model respectively.
• Figure 3: The t-SNE visualization of representations before and after the domain-invariant information extractors on the production dataset.
• Figure 4: Hyper-parameter experiment results of DIIT with DNN as the backbone on the production dataset.
• Figure 5: An illustration of how DIIT works in the industrial RS environment.