This document summarizes a research paper that proposes a multidimensional data mining algorithm to determine association rules across different granularities. The algorithm addresses weaknesses in existing techniques, such as having to rescan the entire database when new attributes are added. It uses a concept taxonomy structure to represent the search space and finds association patterns by selecting concepts from individual taxonomies. An experiment on a wholesale business dataset demonstrates that the algorithm is linear and highly scalable to the number of records and can flexibly handle different data types.