Interesting association rules are those which are novel and actionable.
By definition novel rules are defined as those which represent some information which is not commonly known or accepted. Such rules are often rarely found.
Valid, novel and useful association rules are referred to as actionable rules.
Support and confidence alone are not enough to measure interestingness for association rules. Usually because these factors are almost always pre-determined by domain experts.
Large support and confidence thresholds mean only a small number of association rules, which are usually known facts or "folklores".
Small support and confidence thresholds mean too many association rules.