2008 · 153 citations · 13 references
ComputingComputer ScienceData MiningKnowledge RepresentationMachine LearningEvent DetectionEvent RepresentationFrequent Pattern MiningTemporal Data MiningTemporal Knowledge RepresentationAllen's Interval AlgebraEvent ClassificationInterval-based ClassificationInterval-based RepresentationSearch Space PruningTemporal ClassificationTemporal Pattern DiscoveryTemporal Pattern Mining
Existing temporal pattern mining assumes events are instantaneous, but real-world events have durations and complex relationships, and the current hierarchical representation based on Allen's interval algebra is lossy, failing to fully recover exact relationships. The authors augment the hierarchical representation to achieve lossless interval relationships, design the IEMiner algorithm to discover frequent temporal patterns, and use these patterns to build the IEClassifier for distinguishing closely related classes. The approach models interval relationships with a hierarchical extension of Allen's algebra, augments it for lossless representation, and employs the IEMiner algorithm—using two optimization techniques to prune the search space—to discover frequent patterns that feed into the IEClassifier. Experiments on synthetic and real-world datasets demonstrate the approach’s efficiency, scalability, and the IEClassifier’s improved accuracy.
Existing temporal pattern mining assumes that events do not have any duration. However, events in many real world applications have durations, and the relationships among these events are often complex. These relationships are modeled using a hierarchical representation that extends Allen's interval algebra. However, this representation is lossy as the exact relationships among the events cannot be fully recovered. In this paper, we augment the hierarchical representation with additional information to achieve a lossless representation. An efficient algorithm called IEMiner is designed to discover frequent temporal patterns from interval-based events. The algorithm employs two optimization techniques to reduce the search space and remove non-promising candidates. From the discovered temporal patterns, we build an interval-based classifier called IEClassifier to differentiate closely related classes. Experiments on both synthetic and real world datasets indicate the efficiency and scalability of the proposed approach, as well as the improved accuracy of IEClassifier.
13
Corinna Cortes, Vladimir Vapnik · Machine Learning · 1995 · 33K citations · Full text
Maintaining knowledge about temporal intervals
James F. Allen · Communications of the ACM · 1983 · 7.5K citations · Full text
Steven L. Salzberg, Alberto M. Segre · 1994 · 5.8K citations
R. K. Agrawal, Ramakrishnan Srikant · 2002 · 5.1K citations
Integrating classification and association rule mining