Abstract:

Association rules mining is one of the most active branch of data mining. Many association rules mining algorithms need to scan the database many times and produce a large number of candidate items. Aiming at the problem of scanning database several times, a probabilityweighted association rules mining algorithm based on vector is proposed. It sets the weight value of item by computing the probability and saves the transaction records via the matrixvector structure by scanning the database only once. In addition, it employs different cutting strategies and computing ways of weighted support and confidence. Experimental results show that the new algorithm can improve the mining efficiency distinctly.

Key words: data mining;probability;vector;weighted association rules;cutting strategies