A Support-Less Confidence-Based Association Rule Mining Algorithm Using Relevance
Traditionally, Association Rule Mining (ARM) algorithms work with the concept of frequent item sets. These algorithms are useful in determining the high frequency patterns in data. This data can then be profitably used by businesses to benefit from commonly seen user trends. Hence, the results displayed are those having high occurrence in the transaction database. In certain cases, these results may not be relevant to the user. Sometimes, high confidence rules may need to be generated from item sets which have low frequency i.e. low support and hence may be ignored by traditional ARM algorithms.