Intelligibility and Personalisation of Pervasive Computing Applications
Speaker: Johnson Fong
When: 2009-11-25 13:00:00
Venue: 78-420
Host: Prof. Jadwiga Indulska
Abstract:Context-aware applications adapt their behaviour according to changes in user context and user requirements. The design and development of context-aware applications is a very challenging task and therefore, one of the research areas in context-aware computing focuses on middleware and programming toolkits that facilitate the software engineering process of context-aware applications.
The current middleware for context-aware applications are model-based. Models of context information, situation abstractions and user preferences allow to gather and evaluate context information in the middleware, and this simplifies software engineering of context-aware applications. However, the applications will not always behave as users expect, due to (1) imperfect sensing and reasoning of/on context information and (2) inaccurately captured user preferences. Without appropriate mechanisms for explanations and control of application behaviours the usability of context-aware applications is limited.
The proposed thesis contends that the existing middleware solutions for context-aware applications lack properties that make the application adaptive behaviours transparent and extensible, and therefore, cannot provide a suitable balance between user control and autonomy of the adaptable applications. This justifies and motivates the development of (1) a conceptual framework that improves user experiences by supporting intelligibility of context-aware applications and user control of their adaptive behaviours, and (2) a corresponding software infrastructure for application developers (and potentially also users) to facilitate the development of adaptable context-aware applications, by providing engineering design support to automate the application design process.
At the core of this work are (1) methods for explicit exposure and explanation of the adaptation decision making process in context-aware applications, and (2) models that allow users to customize, at run time, the context-aware behaviour according to their individual requirements
Biography:
Type: PhD confirmation seminar
Contact:jaga@itee.uq.edu.au