Incremental mining of periodic patterns of inadequately informed communication data based on fuzzy segmentation of time series
Abstract Considering the communication data, insufficient information leads to fuzzy time series cycle length, making it impossible to accurately capture real cycle change patterns or recognize new ones. In this regard, we investigate an incremental mining method for cycle patterns in insufficient-i...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-07-01
|
| Series: | Discover Computing |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s10791-025-09665-4 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract Considering the communication data, insufficient information leads to fuzzy time series cycle length, making it impossible to accurately capture real cycle change patterns or recognize new ones. In this regard, we investigate an incremental mining method for cycle patterns in insufficient-information communication data based on fuzzy time series segmentation. This method downgrades the insufficient information communication data to address uneven length issues. Utilizing the improved Gath-Geva algorithm, the downgraded data is partitioned into different periodic time series data via fuzzy clustering. Subsequently, the partial weekly incremental pattern mining algorithm with a moving window scans the segmented data, extracting new periodic patterns using the maximum sub-pattern pandering tree algorithm. After scanning the different cycle time series data obtained from the segmented communication data, we employ the incremental mining algorithm to mine frequent cycle patterns and update incremental pattern information. The experimental results show that in terms of data dimensionality reduction, the time series of communication data with insufficient original high-dimensional information is downgraded to within 0.4 times of the original dimensionality, indicating that the proposed method successfully compresses the original high-dimensional data into a low-dimensional space, effectively reducing the computational complexity and improving the data processing speed. Moreover, the number of periodical pattern segments of communication data mined from the data is consistent, verifying the the method’s stability. The proposed method can accurately identify periodic patterns and effectively discover new patterns for communication data with different source and destination IP addresses. |
|---|---|
| ISSN: | 2948-2992 |