Abstract
The issue of data sparsity poses a significant challenge to recommendersystems. In response to this, algorithms that leverage side information such asreview texts have been proposed. Furthermore, Cross-Domain Recommendation(CDR), which captures domain-shareable knowledge and transfers it from a richerdomain (source) to a sparser one (target), has received notable attention.Nevertheless, the majority of existing methodologies assume a Euclideanembedding space, encountering difficulties in accurately representing richertext information and managing complex interactions between users and items.This paper advocates a hyperbolic CDR approach based on review texts formodeling user-item relationships. We first emphasize that conventionaldistance-based domain alignment techniques may cause problems because smallmodifications in hyperbolic geometry result in magnified perturbations,ultimately leading to the collapse of hierarchical structures. To address thischallenge, we propose hierarchy-aware embedding and domain alignment schemesthat adjust the scale to extract domain-shareable information withoutdisrupting structural forms. The process involves the initial embedding ofreview texts in hyperbolic space, followed by feature extraction incorporatingdegree-based normalization and structure alignment. We conducted extensiveexperiments to substantiate the efficiency, robustness, and scalability of ourproposed model in comparison to state-of-the-art baselines.