Monitoring Behavioral Transitions in Cognitive Rehabilitation with Multi-Model, Multi-Window Stream Mining

摘要

This paper describes how quality metrics over stream-mined models can identify potential changes in user goal attainment, as a user learns a personalized emailing system. A sequence of mined models is generated from sequential segments of logged user email commands. When the quality of some models varies significantly from nearby models - as defined by quality metrics - then the user's behavior is flagged as a potentially significant change. This paper describes how this technique works in its application on a case study of cognitive rehabilitation via emailing.