When LLM-Inferred User Context Adds Value in Production Streaming Recommendation

2026-09-30

AI Summary

This study addresses the unclear applicability of large language model (LLM)-generated profiles versus aggregated embeddings for streaming media recommendation by systematically evaluating both user representation strategies in a production environment. Methodologically, LLMs are leveraged to generate natural language summaries, while an attention mechanism integrates long- and short-term behavioral features. The findings reveal that representation effectiveness depends significantly on user consumption patterns: habitual users benefit more from aggregated profiles, whereas exploratory users gain greater advantages from the semantic context provided by LLMs. Accordingly, this work proposes a dynamic profile selection strategy conditioned on user consumption states. This approach effectively circumvents the limitations inherent in either method alone, yielding substantial improvements in overall recommendation performance.

Abstract

Contextual information in recommender systems is shifting from static, predefined variables toward latent representations inferred from behavior. Large language models support this shift by rendering an unstructured interaction history as a natural-language summary, which yields a thematic user context that can be encoded and used in place of an aggregate profile. The conditions under which such generated profiles outperform aggregate embeddings have received limited characterization mainly at the domain level. We evaluate semantic user-profiling strategies on a production streaming platform, ranking against the full catalog. The evaluation covers a 2*2 design space crossing representation type (aggregate or LLM-generated) with contextual scope (holistic history or attention-fused short-term and long-term contexts). The relative ordering of the two representation types is conditional on the user's consumption regime. Aggregate profiles are consistently stronger under habitual consumption, which characterizes approximately four-fifths of the population, while LLM-generated profiles are stronger for exploratory users whose subsequent interactions diverge semantically from their history. We also observe a popularity-attractor effect in LLM-generated profiles, which modestly raises within-list diversity while substantially lowering catalog coverage and reducing novelty. These results indicate that a context-aware system can select a profiling strategy from the inferred consumption regime rather than applying one representation to all users.

Problem

• Recommender Systems
• Large Language Models
• User Profiling
• Streaming Recommendation
• Contextual Information

Innovation

• LLM-inferred user context
• streaming recommendation
• consumption regime
• semantic user profiling
• context-aware profiling