Personalization in educational AI systems often fails when learner profiles are sparse, fragmented, or incoherent conditions that disproportionately affect learners. Existing interoperability standards such as xAPI enable data portability but offer limited support for selective and contextual data enrichment. This paper introduces a Learner-Centric Adaptive Enrichment Framework that empowers learners to store and manage their educational data through personal online data stores. When an educational system encounters uncertainty in a learner model, it can upon consent retrieve only the most relevant and timely data for model enhancement. The proposed pipeline integrates dynamic feature selection and incremental learning techniques to update models efficiently and ethically. We detail the system workflow, discuss implementation challenges, and highlight how adaptive interoperability can enable more responsive, fair, and privacy-conscious learning support.