Model Based Dialogue Control for Smartwatches

The presented approach solves the problem of organizing a well-structured dialogue between a smartwatch wearer and a health assistance app running on the smartwatch. Especially for emergency situations, when the app has concluded the presence of several threatening health or security hazards. Such situations require an effective, joint handling of all acute health and security hazards in one combined dialogue. Otherwise, the domain specific knowledge for concluding present hazards has been isolated deliberately into separate units for efficient maintenance and comprehensibility. It is represented in a declarative way by finite state machines. A multitude of such finite state machines will be executed simultaneously in the app for monitoring all relevant potential health and/or security hazards. Furthermore, the tactical knowledge for dialogue handling of concluded hazards is represented independently from the domain specific knowledge by “critical dialogue sections”, CDSs. A CDS has an internal activity structure for controlling the dialogue with the smartwatch wearer in order the handle the concluded hazard. During the execution of a CDS, the smartwatch I/O and communication devices are exclusively attached to the CDS. Effective handling of such concluded hazards does typically include to establish a speech connection with distant family members or a home emergency call center. This can be achieved via the integrated cellular phone of the smartwatch. The selection of the situationally most appropriate hazard for handling is done via a central scheduler utilizing a blackboard. Our application framework relieves the software developer from dealing with the details and obstacles of effective dialogue implementation and allows to focus on the domain specific knowledge and handling logic of the assistance app.