Efficient Strategies for Mining Negative Association Rules

点击翻译按钮获取中文摘要

Negative association rules (NAR) catch mutually-exclusive correlations among items. They play important roles just as traditional association rules (TAR) do. For example, in stock market surveillance based on alert logs, NARs detect which alerts are false. There are essential differences between mining TARs and NARs because NARs are hidden in infrequent itemsets. This research will develop efficient strategies for mining NARs in databases. These strategies are expected to be about ten times faster than existing ones. This project will deliver database-independent and high-performance mining algorithms for decision-making. The results can benefit Australian marketing and financial companies as well as health and security departments for smart information use.