Multi-Level Association Rules Mining Algorithm Based on Semantic Relativity
Traditional multi-level association rules mining approaches are based only on database contents. The relations of items in itemset are considered rarely. It leads to generate a lot of meaningless itemsets. Aiming at the problem,multi-level association rules mining algorithm based on semantic relativity is proposed. Domain knowledge is described by Ontology. Every item is seen as a concept in Ontology. Semantic relativity is used to measure the semantic meaning of itemsets. Minimum support of itemset is set according to its length and semantic relativity. Semantic related minimum support with length-decrease is defined to filter meaningless itemsets. Experiments results showed that the method in the paper can improve the efficiency of multi-level association rules mining and generated meaningful rules.
L. Zhang and Z. C. Wang, "Multi-Level Association Rules Mining Algorithm Based on Semantic Relativity", Key Engineering Materials, Vols. 460-461, pp. 363-368, 2011