.png){kind=icon}
.png){kind=icon}
Return
Uncertainty oriented pattern extracting and analyzing via sliding window control
.png){kind=icon}
PRE
AI
DOI：10.1016/j.eswa.2025.129309.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Uncertain pattern analysis is one of the diverse data analysis techniques that discover hidden information in stream data. The uncertain pattern analysis method integrated into intelligent systems aims to discover insightful patterns from uncertainty-driven data by considering the existential probability of the information comprising the data. In the consistently accumulated stream data, it is necessary to consider the latest information more important than the old information. Previous pattern analysis methods reflect the information included in all scanned data in global data structures to analyze uncertain patterns from uncertainty-driven data. Specifically, the latest list-based uncertain pattern analysis methods do not adequately discover hidden insights reflected in the recent trend from stream data. Motivated by the limitation, we present a novel list-based uncertain pattern analysis method with the sliding window control. The proposed method removes information related to the oldest data, which does not properly reflect recent trends, and analyzes result patterns from the latest transactions. Pruning strategies satisfying the anti-monotone property are presented, which are suitable for the sliding window technique. The experimental results demonstrate that the proposed method has superior runtime, memory usage, and scalability compared to other comparison methods. Moreover, additional evaluations demonstrate that the proposed method has completeness and effectiveness for analyzing large-scale data with practical applicability in real-world environments where numerous streams occur.
Keywords:
uncertain pattern analysis
sliding window
stream data
pruning strategies
anti-monotone property
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W
Organization
No organization information available