DIGITWISE: Digital Twin-based Modeling of Adaptive Video Streaming Engagement

TL;DR

DIGITWISE tackles the variability in user engagement during adaptive video streaming by introducing a digital twin framework that learns per-user sensitivities from session data and uses them to predict engagement with a unified model. The system comprises a sensitivity plane (per-user DTs) and an engagement plane (concatenated features and sensitivities fed into an XGBoost predictor), enabling scenario-based optimization of ABR and encoding parameters. Key findings show that incorporating user sensitivities reduces engagement-prediction MAE by up to $5.8\%$ and can increase predicted engagement by up to $8.6\%$ through parameter tuning, with a public dataset to support reproducibility. This approach provides a practical, flexible tool for personalizing streaming experiences and informing design decisions in content provisioning and delivery.

Abstract

As the popularity of video streaming entertainment continues to grow, understanding how users engage with the content and react to its changes becomes a critical success factor for every stakeholder. User engagement, i.e., the percentage of video the user watches before quitting, is central to customer loyalty, content personalization, ad relevance, and A/B testing. This paper presents DIGITWISE, a digital twin-based approach for modeling adaptive video streaming engagement. Traditional adaptive bitrate (ABR) algorithms assume that all users react similarly to video streaming artifacts and network issues, neglecting individual user sensitivities. DIGITWISE leverages the concept of a digital twin, a digital replica of a physical entity, to model user engagement based on past viewing sessions. The digital twin receives input about streaming events and utilizes supervised machine learning to predict user engagement for a given session. The system model consists of a data processing pipeline, machine learning models acting as digital twins, and a unified model to predict engagement. DIGITWISE employs the XGBoost model in both digital twins and unified models. The proposed architecture demonstrates the importance of personal user sensitivities, reducing user engagement prediction error by up to 5.8% compared to non-user-aware models. Furthermore, DIGITWISE can optimize content provisioning and delivery by identifying the features that maximize engagement, providing an average engagement increase of up to 8.6%.

Figures (14)

• Figure 1: Conceptual architecture of video streaming with DIGITWISE enhancements.
• Figure 2: DIGITWISE architecture.
• Figure 3: Inputs and outputs of the Data Processing component.
• Figure 4: DIGITWISE pipeline.
• Figure 5: Service tuning architecture.