Discovery of Frequent Episodes in Event Sequences
Data Mining and Knowledge Discovery
Finding event-oriented patterns in long temporal sequences
PAKDD'03 Proceedings of the 7th Pacific-Asia conference on Advances in knowledge discovery and data mining
A tree structure for event-based sequence mining
Knowledge-Based Systems
| Hi-index | 0.00 |
A major task of traditional temporal event sequence mining is to predict the occurrences of a special type of event (called target event) in a long temporal sequence. Our previous work has defined a new type of pattern, called event-oriented pattern, which can potentially predict the target event within a certain period of time. However, in the event-oriented pattern discovery, because the size of interval for prediction is pre-defined, the mining results could be inaccurate and carry misleading information. In this paper, we introduce a new concept, called temporal feature, to rectify this shortcoming. Generally, for any event-oriented pattern discovered under the pre-given size of interval, the temporal feature is the minimal size of interval that makes the pattern interesting. Thus, by further investigating the temporal features of discovered event-oriented patterns, we can refine the knowledge for the target event prediction.