Efficient Computation of Measurements of Correlated Patterns in Uncertain Data

摘要

One of the most important tasks in data mining is to discover associations and correlations among items in a huge database. In recent years, some studies have been conducted to find a more accurate measure to describe correlations between items. It has been proved that the newly developed measures of all-confidence and bond perform much better in reflecting the true correlation relationship than just using support and confidence in categorical database. Hence, several efficient algorithms have been proposed to mine correlated patterns based on all-confidence and bond. However, as the data uncertainty become increasingly prevalent in various kinds of real-world applications, we need a brand new method to mine the true correlations in uncertain datasets with high efficiency and accuracy. In this paper, we propose effective methods based on dynamic programming to compute the expected all-confidence and expected bond, which could serve as a slant in finding correlated patterns in uncertain datasets.