Formal concept analysis (FCA) is increasingly applied to data mining problems, essentially as a formal framework for mining reduced representations (bases) of target pattern families. Yet most of the FCA-based miners, closed pattern miners, would only extract the patterns themselves out of a dataset, whereas the generality order among patterns would be required for many bases. As a contribution to the topic of the (precedence) order computation on top of the set of closed patterns, we present a novel method that borrows its overall incremental approach from two algorithms in the literature. The claimed innovation consists of splitting the update of the precedence links into a large number of lower-cover list computations (as opposed to a single upper-cover list computation) that unfold simultaneously. The resulting method shows a good improvement with respect to its counterpart both on its theoretical complexity and on its practical performance. It is therefore a good starting point for the design of efficient and scalable precedence miners.
Weitere Kapitel dieses Buchs durch Wischen aufrufen
Bitte loggen Sie sich ein, um Zugang zu diesem Inhalt zu erhalten
Sie möchten Zugang zu diesem Inhalt erhalten? Dann informieren Sie sich jetzt über unsere Produkte:
- Yet a Faster Algorithm for Building the Hasse Diagram of a Concept Lattice
- Springer Berlin Heidelberg
Neuer Inhalt/© ITandMEDIA