A framework for knowledge-based temporal abstraction

Reviewer: D. Charles Hair

Shahar describes his research into using a knowledge-based approach to abstract higher-level concepts from time-stamped data. The approach is particularly valuable in areas where more time-stamped data is available for analysis than a user can handle. The author describes both the theoretical underpinning of the approach and an implementation that was used to evaluate the approach. The primary focus of this work has been in clinical medical settings. Some evaluation work has also been done in an engineering domain. Evaluation has indicated that this approach facilitates knowledge acquisition, reduces work for developers, provides a useful explicit representation of domain-specific knowledge, and makes system maintenance easier. This paper is clearly written paper, and the work it presents should be of particular interest to those pursuing knowledge-based projects in medical informatics and in time-oriented domains.