Intelligent techniques for context-aware systems
Nayat Sánchez-Pi · LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas) · 2011
Nowadays, with advances in communication technologies, researches are focused in the fields of designing new devices with increasing capabilities, implanting software frameworks or middleware to make these devices interoperable. Building better human interfaces is a challenging task and the adoption of Artificial Intelligence (AI) techniques to the process help associating semantic meaning to devices which makes possible the gesture recognition and voice recognition. This thesis is mainly concerned with the open problem in context-aware systems: the evaluation of these systems in Ambient Intelligence (AmI) environments. With regard to this issue, we argue that due to highly dynamic properties of the AmI environments, it should exist a methodology for evaluating these systems taking into account the type of scenarios. However in order to support with a solid ground for that discussion, some elements are to be discussed as well. In particular, we: • use a commercial platform that allows us to design and manage the contextual information of context- aware systems by means of a context manager included in the architecture; • analyze the formal representation of this contextual information by means of a knowledge based system (KBS); • discuss the possible methodologies to be used for modelling knowledge in KBS and our approach; • give reasons why intelligent agents is a valid technique to be applied to systems in AmI environments; • propose a generic multi-agent system (MAS) architecture that can be applied to a large class of envisaged AmI applications; • propose a multimodal user interface and its integration with our MAS; • propose an evaluation methodology for context-aware systems in AmI scenarios. The formulation of the above mentioned elements became necessary as this thesis was developed. The lack of an evaluation methodology for context-aware systems in AmI environments, where so many issues to be covered, took us to the main objective of this thesis. In this regard: • we provide an updated and exhaustive state-of-the-art of this matter; • examine the properties and characteristics of AmI scenarios; • put forward an evaluation methodology and experimentally test our methodology in AmI scenarios. ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------