Description
Date depot: 1 janvier 1900
Titre: Strategies for Context Reasoning in Assistive Livings for the Elderly
Directeur de thèse:
Mounir MOKHTARI (SAMOVAR)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini
Resumé:
One’s interest in ambient intelligence may lie in the ability of an environment to respond in an ap- propriate manner to what is happening within it. It is the reaction of a computerised system to a non-formalised situation that is intriguing. Such systems are by nature able to instantiate a reaction, even complex, to a formalised and recognised situation, even complex. The true challenge is to provide a formalisation for machines to project situational data and make sense of it, i.e. build connections or bindings between it and the rest of the contextual knowledge. We call this challenge “context comprehension”, and divide it into two main aspects: (i) formalising contextual knowledge to project situational data in it, and (ii) reasoning to connect such inter-correlated formalised knowledge or infer new one. The problem being studied in this doctoral work is: What strategies can be put in place to provide context comprehension in assistive livings?
Comprehension is defined in the Random House Kernerman Webster’s College Dictionary as the “capacity of the mind to perceive and understand; [the] power to grasp ideas”. We can see it as the process of simultaneously extracting and constructing meaning through the manipulation of sensed situational data. We use the words extracting and constructing to emphasize both the importance and the insufficiency of the sensed data as a determinant of comprehension. In the field of ambient intelli- gence, it consists in putting in place a translation mechanism between the sensed representation of a situation and its formalised, machine-readable version. This mechanism would probably be constituted of heterogeneous and complementary strategies, allowing the definition of a formalisation (or model), the naive projection of sensed data into the model, and the inference of knowledge into this model.
Leveraging our experience with the traditional approach to Ambient Assisted Living (AAL) which re- lies on a large spread of heterogeneous technologies in deployments, this thesis studies the possibility of a more “stripped down” and complementary approach, where only a reduced hardware subset is deployed, probing a transfer of complexity towards the software side, and enhancing the large scale deployability of the solution. Focused on the reasoning aspects in AAL systems, this work has allowed the finding of a suitable semantic inference engine for the peculiar use in these systems, responding to a need in this scientific community. Considering the coarse granularity of situational data available, ded- icated rule-sets with adapted inference strategies are proposed, implemented, and validated using this engine. A novel semantic reasoning mechanism is proposed based on a cognitively inspired reasoning architecture. Finally, the whole reasoning system is integrated in a fully featured context-aware service framework, powering its context awareness by performing live event processing through complex onto- logical manipulation. The overall system is validated through in-situ deployments in a nursing home as well as private homes over a few months period, which itself is noticeable in a mainly laboratory-bound research domain.
This thesis concentrates its research efforts on the reasoning aspects in smart environments. The disser- tation will focus mainly on the design of the reasoning engine which allowed us to develop an integrated system that has been deployed in real conditions. It is organised around five parts, each subdivided into one to three chapters. The first part provides the background and motivation behind this doctoral work, concluding with the positioning of the work. The second part consists in the conception part of the work. It is composed of an analysis of the related work in rule-based reasoning, a comparison of semantic reasoning engines from an ambient intelligence point of view, the design of a novel rule-set for context comprehension, and its integration into a cognitively inspired reasoning architecture. The third part is more focused towards the implementation aspects of the work. After introducing the enabling technologies, it describes the evolution of the context-aware service framework designed and implemented, focusing on two notable milestones and their respective architectures. Details of the mechanisms implemented are provided, highlighting their contribution to the improvement of the over- all process and performance. The fourth part provides results from the various validations that have been conducted; on one hand in a nursing home in Singapore and on the other hand in three individual homes in France, with the involvement in both countries of several partners from medical, research and engineering background. It also discusses the strategies which could enable the technological transfer into society of AAL solutions. The last part concludes on the work done and provides an overview of the further studies in perspective.
Doctorant.e: Tiberghien Thibaut