Association rules is a descriptive data mining technique, that has acquired major interest in the last years, with applications in several areas, such as: electronic and traditional commerce, securities, health care and geo-processing. This technique allows identifying intratransactions patterns in a database. An association rule describes how much the presence of a set of attributes in a database's record implicates in the presence of other distinct set of attributes in the same record. The possibility of discovering all the existent associations in a database transaction is the most relevant aspect of the technique. However, this characteristic determines the generation of a large number of rules, hindering the capacity of a human user in analyzing and interpreting the extracted knowledge. The use of rule measures is important to analyze the knowledge. In this context we investigate, firstly, the intensity of several objective measures to act as filters for rule sets. Next, we analyze how the combination of these measures can be used to identify the more interesting rules. Finally, we apply the proposed technique to a rule set, to illustrate its use in the post-processing phase. © Springer-Verlag Berlin Heidelberg 2004.
CITATION STYLE
Melanda, E. A., & Rezende, S. O. (2004). Combining quality measures to identify interesting association rules. In Lecture Notes in Artificial Intelligence (Subseries of Lecture Notes in Computer Science) (Vol. 3315, pp. 441–453). Springer Verlag. https://doi.org/10.1007/978-3-540-30498-2_44
Mendeley helps you to discover research relevant for your work.