Association Rule Mining for Temporal Data without Lifespans

There are significant challenges in finding association rules for temporal data due to the associations' temporal factors. In particular, fuzzy temporal association rule mining, a widely applied human explainable method to reveal such rules by discovering fuzzy associations among temporal data, encounters a problematic issue caused by unknown lifespans of the patterns and associations, including start time and end time during the period of effective duration. More specifically, temporal data could have their unique lifespans. These lifespans are essential to the discovery of such temporal association rules. Current work only deals with objects that have directly observable lifespans. However, objects whose lifespans are usually unknown can only be indirectly estimated according to their effects. This thesis, therefore, aims to cope with unknown lifespans so that the proposed methods can serve the purpose of revealing temporal rules.