Research on time series modeling and high performance computing for dynamic valuation of data assets
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
With the acceleration of data capitalization, traditional static valuation methods can no longer meet the needs of dynamic, objective and accurate. Aiming at the dual separation between the existing data asset valuation model and the computing level, this paper proposes a time series modeling method integrating domain knowledge, and realizes the dynamic valuation of data asset value by combining the cloud-side collaborative high-performance computing architecture. Firstly, the multi-dimensional information of data assets is integrated by constructing a multi-modal feature engineering layer; Secondly, the Gated Spatio-temporal Unit (GSTU) is introduced to capture the temporal evolution and spatial correlation of data value. Finally, the real-time transition detection is carried out by using CUSUM control chart to realize the hot update of the model. In addition, this paper designs a cloud-side collaborative high-performance computing architecture based on Ray+Kubernetes, and through tensor slicing, gradient accumulation and sparse communication compression, the training efficiency and reasoning speed of the model are significantly improved. The empirical study shows that this method is superior to the traditional model in terms of valuation accuracy and calculation efficiency, and can effectively deal with unexpected events and sudden policy changes, which provides a new idea and method for dynamic valuation of data assets.
Fields of Study & Subjects
Big Data and Digital Economy
Technology's Impact on Media
Data Stream Mining Techniques
Publication details
- Journal / source
- IET conference proceedings.
- Volume
- 2025
- Issue
- 47
- Pages
- 315–320
- Publication date
- 1 March 2026
- Language
- EN