Including the user in the knowledge discovery loop: interactive itemset-driven rule extraction

We introduce a user-driven approach to mining association rules, integrated into a visualization system, called I2E, that allows miners to depart from a reduced and representative subset of rules to interactively explore the whole rule space. A visualization of the space of k-itemsets displayed after each iteration of Apriori allows the miner to guide rule extraction by exploring the space of itemsets. Miners can discard itemsets considered not relevant and define clusters of related itemsets to perform rule filtering, so that uninteresting rules are removed while preserving itemset coverage in the resulting rule set. This reduced set provides a starting point to explore the rule space with an interface that supports pairwise comparisons between rules, according to some defined criteria. We describe the results obtained from applying the proposed approach and its supporting system on a case study with a real dataset containing information on cattle commercialized in Brazil.

[1]  Alípio Mário Jorge.  Hierarchical Clustering for Thematic Browsing and Summarization of Large Sets of Association Rules , 2004, SDM.

[2]  Li Yang,et al.  Visualizing Frequent Itemsets, Association Rules, and Sequential Patterns in Parallel Coordinates , 2003, ICCSA.

[3]  Fabrice Guillet,et al.  A 2D-3D visualization support for human-centered rule mining , 2007, Comput. Graph..

[4]  Maria Cristina Ferreira de Oliveira,et al.  An Itemset-Driven Cluster-Oriented Approach to Extract Compact and Meaningful Sets of Association Rules , 2007, Sixth International Conference on Machine Learning and Applications (ICMLA 2007).

[5]  Tomasz Imielinski,et al.  Mining association rules between sets of items in large databases , 1993, SIGMOD Conference.

[6]  Bart Goethals,et al.  A priori versus a posteriori filtering of association rules , 1999, 1999 ACM SIGMOD Workshop on Research Issues in Data Mining and Knowledge Discovery.

[7]  Gavriel Salvendy,et al.  Design and evaluation of visualization support to facilitate association rules modeling , 2006, Int. J. Hum. Comput. Interact..

[8]  Ayhan Demiriz,et al.  A Framework for Visualizing Association Mining Results , 2006, ISCIS.

[9]  Solange Oliveira Rezende,et al.  Obtaining and Evaluating Generalized Association Rules , 2007, ICEIS.

[10]  Gavriel Salvendy,et al.  Design and evaluation of visualization support to facilitate decision trees classification , 2007, Int. J. Hum. Comput. Stud..

[11]  B. Shekar,et al.  Interestingness of association rules in data mining: Issues relevant to e-commerce , 2005 .

[12]  Heikki Mannila,et al.  Finding interesting rules from large sets of discovered association rules , 1994, CIKM '94.

[13]  Xiangji Huang,et al.  Hierarchical grouping of association rules and its application to a real-world domain , 2006, Int. J. Syst. Sci..