Mining relationships among interval-based events for classification[1]英文文献资料.pdf

Mining Relationships Among Interval-based Events for Classification Dhaval Patel Wynne Hsu Mong Li Lee School of Computing National University of Singapore Singapore {dhaval,whsu,leeml}@.sg ABSTRACT temporal patterns from instantaneous events, that is, events Existing temporal pattern mining assumes that events do with no duration. This assumption allows the discovered pattern to be simplified to an ordered sequence of events, not have any duration. However, events in many real world such as “fever → stomach ache → vomit”. However, such applications have durations, and the relationships among these events are often complex. These relationships are mod- sequential patterns are inadequate to express the complex eled using a hierarchical representation that extends Allen’s temporal relationships in domains such as medical, multi- interval algebra. However, this representation is lossy as the media, meteorology and finance where the events’ durations exact relationships among the events cannot be fully recov- could play an important role. ered. In this paper, we augment the hierarchical representa- For example, it has been observed that in many diabetic tion with additional information to achieve a lossl