Investigation of Multi-Platform Media Publishing with AI Personalization Engines

Abstract

Conventional media publishing mechanisms do not respond to personalised user preferences and are limited to operating on a single platform at a time. In this paper, the investigation examines an AI-driven recommendation mechanism that incorporates cross-platform personalisation. Our model consists of a holistic framework for consolidating behavioural data from multiple sources, implementing a contextual bandit strategy with platform-based reward prediction, and adapting content dynamically to platform-specific constraints. Our prototype involves twelve distinct users, thirty-seven different multimodal content pieces, and 157 actual user interactions. Descriptive analyses of observed user behaviours are performed. The results indicate that our proposed model attains a descriptive CTR of 18.9% and a seven-day retention rate of 58.3% in prototype experiments. Consistency across platforms (ρ = 0.81) and novelty (27.8%) are also quantified. Although, due to small samples, no generalisable conclusions can be drawn, our findings indicate that cross-platform personalisation is feasible and can exceed the performance of baseline single-platform models.