2013 | OriginalPaper | Buchkapitel
Mining Interesting Itemsets in Graph Datasets
verfasst von : Boris Cule, Bart Goethals, Tayena Hendrickx
Erschienen in: Advances in Knowledge Discovery and Data Mining
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Traditionally, pattern discovery in graphs has been mostly limited to searching for frequent subgraphs, reoccurring patterns within which nodes with certain labels are frequently interconnected in exactly the same way. We relax this requirement by claiming that a set of labels is interesting if they often occur in each other’s vicinity, but not necessarily always interconnected by exactly the same structures. Searching for such itemsets can be done both in datasets consisting of one large graph, and in datasets consisting of many graphs. We present novel methods dealing with both possible settings. Our algorithms benefit from avoiding computationally costly isomorphism checks typical for subgraph mining, as well as from a greatly reduced search space, as we consider only itemsets, and not all possible edges and paths that can connect them.