Beyond intratransaction association analysis

Abstract

In this paper, we extend the scope of mining association rules from traditional single-dimensional intratransaction associations, to multidimensional intertransaction associations. Intratransaction associations are the associations among items with the same transaction , where the notion of the transaction could be the items bought by the same customer , the events happened on the same day , and so on. However, an intertransaction association describes the association relationships among different transactions , such as “if(company) A's stock goes up on day 1, B's stock will go down on day 2, but go up on day 4.” In this case, whether we treat company or day as the unit of transaction, the associated items belong to different transactions. Moreover, such an intertransaction association can be extended to associate multiple contextual properties in the same rule, so that multidimensional intertransaction associations can be defined and discovered. A two-dimensional intertransaction association rule example is “After McDonald and Burger King open branches, KFC will open a branch two months later and one mile away,” which involves two dimensions: time and space . Mining intertransaction associations poses more challenges on efficient processing than mining intratransaction associations. Interestingly, intratransaction association can be treated as a special case of intertransaction association from both a conceptual and algorithmic point of view. In this study, we introduce the notion of multidimensional intertransaction association rules, study their measurements— support and confidence—and develop algorithms for mining intertransaction associations by extension of Apriori. We overview our experience using the algorithms on both real-life and synthetic data sets. Further extensions of multidimensional intertransaction association rules and potential applications are also discussed.

Logical Linked Data Compression

“…Multi-Dimensional Association Rules Although association mining was originally studied for mining transactions for only one attribute (ex:Product), much research has been performed to extend it across multiple attributes [16,17,28,29]. In this study, RDF datasets are viewed as multi-dimensional transaction databases by treating each property as an attribute and a subject as an identifier.…”

“…In this study, RDF datasets are viewed as multi-dimensional transaction databases by treating each property as an attribute and a subject as an identifier. Similar to intra-transaction and inter-transaction associations [17], we define intra-property and inter-property associations for RDF datasets. Intra-property association refers to an association among different object values for a given property while inter-property association refers to association between multiple properties.…”

First-order temporal pattern mining with regular expression constraints

“…The remaining msps are tested for validity with respect to the original RE-condition R. 7 A sequence s is valid with respect to a state q if there exists a path in the automaton corresponding to s, starting at q and ending at a final state. 8 A sequence s is legal with respect to a state q it there is a path in the automaton corresponding to s and starting at q.…”

“…Different kinds of sequential patterns and more general temporal patterns ( [4,5,6,7,8]) have been proposed, as well as general formalisms and algorithms for expressing and mining these patterns have been developed ( [9,10]). Most of these patterns are specified by formalisms which are, in some extent, reducible to Propositional Temporal Logic.…”

Online Mining in Sensor Networks

“…Alternatively, we can consider sensory attributes correlations as inter-transaction association rules [13]. For example, a rule or correlation says that "if node A's attribute x goes up at time point 1, B's attribute y will (with an 80% possibility) go up at time point 2 and C's attribute x will go down at time point 3".…”