Mining Association Rules in Hypertext Databases

In this workweproposea generalisation of the notion of associationrule in the contextof flat transactions to that of a compositeassociation rule in the context of a structured directed graph, such as the world-wide-web.Thetechniques proposed aimat finding patterns in Cheuser behaviourwhen traversing such a hypertext system. Weredefine the concepts of confidence and support for composite association rules, and two algorithms to mine such rules are proposed. Extensive experiments with randomdata were conducted and the results showthat, in spite of the worst-casecomplexity analysis whichindicates exponential behaviour, in practice the algorithms' complexity, measuredin the numberof iterations performed, is linear in the numberof nodestraversed.

• Type: Conference
• Year: 1998
• Where: KDD
• Authors: José Borges, Mark Levene
• Keywords: Contextof Flat Transactions, Data Mining, KDD 1998, Structured Directed Graph, World-wide-web.Thetechniques Proposed Aimat