Discovery of association rules from large databases of itemsets is an important data mining problem. Association rules are usually stored in database systems for future use in decision support systems. The problem of retrieval of association rules and item sets may be recognized as the subset search problem in databases. The problem is not well supported by SQL language and traditional database indexing techniques.
In the talk a new index structure, called Group Bitmap Index, will be presented and its performance will be compared with traditional indexing techniques: B+-tree and bitmap indexes. It will be shown that proposed index enables faster subset search and significantly outperforms the traditional indexing methods.
Referent: Prof. Dr. Tadeuzs Morzy,
Vorstand Datenbanksystem-Labor,
Institut für Computer Science
Technische Universität Poznan,
Polen
Zeitpunkt: Freitag, 9. Januar 1998, 14 Uhr c. t.
Ort: HS 3 der Universität Klagenfurt