| Mining for frequent patterns in transactional databases has been studied for more than a decade. Many algorithms have been developed to mine static databases. There are a few incremental algorithms, FUP2 and SWF, that allow both addition and deletion of transactions. However, they are not efficient because they have to rescan the whole dataset at least once. They are not suitable in real time situations where transactions are added or deleted constantly and frequent patterns mining could be required at any time.; In this thesis, we propose a novel data structure called CATS Tree. CATS Tree extends the idea of FP-Tree to improve storage compression and allow frequent pattern mining without generation of candidate itemsets. The proposed algorithms allow mining with a single pass over the database as well as addition or deletion of transactions in the finest granularity at any given time. |
