Incremental sequential rule mining with streaming input traces
| Autor: | Andriy Drozdyuk, Scott Buffett, Michael Fleming |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2020 |
| Předmět: |
Association rule learning
Computer science Computation 02 engineering and technology Extension (predicate logic) data mining Reuse computer.software_genre streaming data Set (abstract data type) incremental sequential rule mining 020204 information systems Streaming data 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Data mining Sequential Pattern Mining computer Sequential rule mining |
| Zdroj: | Advances in Artificial Intelligence ISBN: 9783030473570 Canadian Conference on AI |
| DOI: | 10.1007/978-3-030-47358-7_8 |
| Popis: | Traditional static pattern mining techniques, such as association rule mining and sequential pattern mining, perform inefficiently when applied to streaming data when regular updates are required, since there is significant repetition in the computation. Incremental mining techniques instead reuse information that has been previously extracted, and apply newly received data to compute the updated set of patterns. This paper proposes a new algorithm for incrementally mining sequential rules with streaming data. An existing rule mining algorithm, ERMiner is presented, and an incremental extension, called IERMiner is proposed and demonstrated. Experiments show that IERMiner significantly decreases the run time required to update the set of patterns when compared to running ERMiner on the full dataset each time. 33rd Canadian Conference on Artificial Intelligence, Canadian AI 2020, May 13–15, 2020, Ottawa, ON, Canada Series: Lecture Notes in Computer Science |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|