Paper Type
Short
Paper Number
PACIS2025-1510
Description
Predicting potential blockbusters, e.g., digital content with explosive consumption growth, is important for E-commerce platforms. However, blockbusters face predictability challenges and the historical explosion patterns may not necessarily apply to future fashion trends. This paper introduces a novel framework for blockbuster prediction called Invariants-Enhanced Variants Learning (IEVL) that leverages a coupled optimization framework. The estimation of the probability of content becoming a blockbuster can benefit from insights derived from past blockbusters and the vast corpus of non-blockbuster content. We validate our framework using data from China’s leading e-commerce platform, digital content provided by Alibaba, demonstrating the effectiveness of our framework compared with benchmark approaches. Our research provides insights into both forecasting methodologies and e-commerce platform practices.
Recommended Citation
Wu, Ziyue and Chen, Xi, "Coupling Variants with Invariants: Excavating Blockbusters from Content Performance Footprints" (2025). PACIS 2025 Proceedings. 1.
https://aisel.aisnet.org/pacis2025/aiandml/aiandml/1
Coupling Variants with Invariants: Excavating Blockbusters from Content Performance Footprints
Predicting potential blockbusters, e.g., digital content with explosive consumption growth, is important for E-commerce platforms. However, blockbusters face predictability challenges and the historical explosion patterns may not necessarily apply to future fashion trends. This paper introduces a novel framework for blockbuster prediction called Invariants-Enhanced Variants Learning (IEVL) that leverages a coupled optimization framework. The estimation of the probability of content becoming a blockbuster can benefit from insights derived from past blockbusters and the vast corpus of non-blockbuster content. We validate our framework using data from China’s leading e-commerce platform, digital content provided by Alibaba, demonstrating the effectiveness of our framework compared with benchmark approaches. Our research provides insights into both forecasting methodologies and e-commerce platform practices.
Comments
AI ML