The computation of the similarities between event sequences is important for many fields because many activities follow a sequential order. For instance, an industrial plan that triggers different types of alarms due to detected event sequences or the treatment sequence that a patient receives while he/she is hospitalized. With the appropriate tools and techniques to compute the similarity between two event sequences we may be able to detect patterns or regularities in event data and so be able to perform predictions or recommendations based on detected similar sequences. The present work is intended to describe the implementation of two event sequence similarity measures in myCBR, with the purpose of creating a similarity measurement approach for complex domains that employ the use of event sequences. Besides, an initial experimentation is performed in order to study if the proposed measures and measurement approach are able to predict future situations based on similar event sequences.
展开▼
