关键词: 高维数据，稀疏数据，频繁项集，存储结构

Abstract: The traditional mining algorithms arc not applicable to mine high-dimensional sparse dataset,a new frequent itemsets mining algorithm based on high-dimensional sparse dataset named FIRS (Frequent mining algorithm based on High-dimensional Sparse dataset) was proposed in this paper. FIHS adopts a new data structure to store frequcnt itemsets, using this structure can reduce the storage space and the cost of counting. FIHS can avoid generating infrectuent candidate itemsets through optimizing the operation of connection and pruning,which rectuires scan the dataset once. What's more,just by applying ANIX)R operation,frequcnt K+1-itemsets can be created according to frequent K-itemsets, and the maintenance of the data structure is simple. According to theoretical analysis and experiments, the improved algorithm enjoys many advantages aiming at high-dimensional sparse dataset, such as quick mining, less memory spacc,etc.

Key words: High-dimensional data,Sparse data,Frequent itemsets,Data structure