AI-Driven Personalization across Domains for Local Categorical Query Understanding and Context -Aware Retrieval

This paper explores the cross-domain application of AI-driven personalization in structured search scenarios that combine intent understanding with spatial and categorical constraints across dining, lodging, and leisure experiences. By integrating LLM-based coordination with reinforcement learning and user memory modules, the system continuously learns from users’ long-term preferences and interaction history to support complex, context-rich needs. Experimental evaluations show that memory-enhanced personalization improved result helpfulness by 17.25% and increased transactional referrals by 4.16% in lodging-related searches, while also achieving measurable satisfaction gains in dining and leisure domains. The study demonstrates that crossdomain LLM personalization frameworks with user memory can effectively capture evolving user intents within local categorical contexts, enhance contextual reasoning, and advance the design of adaptive information service systems in the digital economy