Social Knowledge for Cross-Domain User Preference Modeling

2026-03-10

AI Summary

This work proposes a social embedding–based approach for cross-domain personalized recommendation in zero-shot scenarios where user feedback is absent in the target domain. By constructing a joint embedding space of users and popular entities from large-scale social networks, the method leverages cosine similarity to predict user preferences over candidate entities in the target domain. It is the first to incorporate social embeddings into cross-domain preference modeling, revealing latent demographic characteristics encoded in the embeddings and their correlation with cross-domain preferences. Furthermore, the study explores the potential of large language models in learning user social representations. Experimental results demonstrate that the proposed method significantly outperforms popularity-based baselines and effectively enables zero-shot cross-domain recommendation.

Technology Category

• Application Domains: Humanities & Computational Social Science
• Machine Learning: Learning Preferences or Rankings
• Cognitive Modeling & Cognitive Systems: Social Cognition And Interaction

Application Category

• User Modeling, Personalization and Recommendation: Social recommender systems and personalization for the social Web
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

We demonstrate that user preferences can be represented and predicted across topical domains using large-scale social modeling. Given information about popular entities favored by a user, we project the user into a social embedding space learned from a large-scale sample of the Twitter (now X) network. By representing both users and popular entities in a joint social space, we can assess the relevance of candidate entities (e.g., music artists) using cosine similarity within this embedding space. A comprehensive evaluation using link prediction experiments shows that this method achieves effective personalization in zero-shot setting, when no user feedback is available for entities in the target domain, yielding substantial improvements over a strong popularity-based baseline. In-depth analysis further illustrates that socio-demographic factors encoded in the social embeddings are correlated with user preferences across domains. Finally, we argue and demonstrate that the proposed approach can facilitate social modeling of end users using large language models (LLMs).

Problem

• cross-domain
• user preference modeling
• zero-shot
• social knowledge
• personalization

Innovation

• cross-domain preference modeling
• social embedding
• zero-shot personalization
• large language models
• user representation