Revisiting Data Mining for Facebook's Transition to Meta: Challenges and Future Directions

Abstract

Facebook's rebranding to Meta represents a strategic transition from a dominant social-media platform toward an integrated, immersive ecosystem built around virtual reality (VR), augmented reality (AR), and interconnected digital spaces. This shift substantially widens the scope, type, and complexity of user-generated data, extending it beyond clicks and text into spatial movement, gesture-based input, device telemetry, and cross-platform behavioural signals. This paper examines which data-mining (DM) techniques are best suited to interpreting this new multi-modal data environment, focusing on clustering and association-rule mining (ARM), and the challenges that accompany their application. Drawing on a structured review of academic and industry literature, the paper discusses how clustering enables behavioural segmentation of VR/AR users while ARM uncovers cross-platform usage patterns that support retention, cross-selling, and seamless user-journey design. The review also identifies four recurring challenges in Meta's data-mining pipeline: privacy and regulatory exposure, data quality and heterogeneity, real-time processing demands, and model interpretability/algorithmic bias. The paper concludes that clustering and ARM offer complementary analytical foundations for Meta but require reinforcement through privacy-preserving data mining (PPDM), anonymization, secure multiparty computation (SMC), incremental and streaming algorithms, and transparent, bias-audited model governance if Meta is to responsibly realise its metaverse ambitions.