A Novel Incremental Algorithm for Frequent Itemsets Mining in Dynamic Datasets
CIARP '08 Proceedings of the 13th Iberoamerican congress on Pattern Recognition: Progress in Pattern Recognition, Image Analysis and Applications
Algorithms for mining frequent itemsets in static and dynamic datasets
Intelligent Data Analysis
| Hi-index | 0.01 |
Frequent patterns mining has been studied popularly in KDD research. However, little work has been done on incremental updating frequent patterns mining. In a real transaction database, as time changing many new data may be inserted into previous database. But it is hard to handle incremental updating problems with FP-growth algorithm. In this paper, a novel incremental updating pattern tree (INUP_Tree) structure is presented, which is constructed by scanning database only once. And a new frequent pattern mining method (IUF_Mine) based on conditional matrix is developed. When database is updated, only the new added records will be scanned. Besides, original conditional matrix can be adequately used to speed up the new mining process, so the mining efficiency is improved. The experiment result shows that the IUF_Mine method is more efficient and faster than the FP-growth.