« zurück

Kolloquium | Zero-suppressed BDDs and recent applications for data mining and knowledge discovery

18.09.10 | MZH 3420 | 14:00 Uhr

Prof. Shin-ichi Minato from Graduate School of Information Science and Technology, Hokkaido University, Sapporo, Japan

Binary Decision Diagrams (BDDs) are an efficient data structure for representing Boolean functions. Much research has been done on BDD manipulation in VLSI logic design since 1990's. After 2000's, we found that BDD-based techniques can also be applied effectively to data mining and knowledge discovery. Especially, Zero-suppressed BDDs (ZDDs) are suitable for handling sets of sparse combinations that often appear in many practical database analyses. In this talk, we will show our recent activities in BDD-based knowledge discovery techniques. First, we describe the ``LCM over ZDDs'' algorithm, which achieves very fast frequent itemset mining. Then, we present some useful post-processing methods for analyzing the results of the frequent itemset mining.