My research is built around the results obtained for the ITEA2 FUSE-IT project (https://itea3.org/project/fuse-it.html). My main involvement in this project was the design of a semantic-based data model for smart building description, along with a reasoning mechanism for anomaly detection.
As a first step, A common information base, as a core data model, is defined, which describes and defines formally the main physical and conceptual building elements (namely: assets, spaces, data points, incidents, and key performance indicators), their characteristics and interrelationships, as well as the constraints that apply to them. For instantiation purposes, we relied on a logical framework based on the existential rules, which allows describing any domain as a set of facts, a set of rules, and a set of constraints.
The problem of anomaly detection has been mapped to inconsistency checking in a knowledge base, in combination with a decision support system that analyses the relevance of the raised alerts. The whole process amounts to reasoning under uncertainty. Besides, I’m also working on the quantitative interpretation of preference operators involving fuzzy quantification. The aim of this research is to study how ranking operators can be exploited in the design of decision support systems under uncertainty.
Within the framework of the Home Security Network project (HOSEN), I have been involved in the design and implementation of an algorithm for user data protection and user privacy preservation in the context of Internet of Things applications. This project involves two industrial partners (IOTIC et NetASQ) and La Rochelle university (L3i laboratory of the university of La Rochelle).
My main contribution is an algorithm for user habit identification, which learns continuously the habits of a given user or any other object, and detects any deviation in the current behavior. If the deviation exceeds a tolerable threshold, then an alert is raised and the communication is stopped between the connected object and the external world.
To do so, we designed a knowledge representation formalism and a reasoning mechanism, which are capable to model the semantics of the data exchanged by connected devices at home with the external world, and inferring and applying, on the fly, security policies to protect the communications between the connected object and the external environment, and hence to protect the user's privacy.
Within the framework of the European project EcoBioCap, I was involved in modeling and aggregation of user preferences expressed by the various stakeholders, as textual arguments, in the process of designing and developing biodegradable and eco-efficient packaging. This aggregation process is based on an argumentation approach.
An argumentation-based software has also been implemented, in collaboration with Patricio Mosse, as a final deliverable of the project. A demonstration video is available here. An online version of the system is also provided and available at http://pfl.grignon.inra.fr/EcoBioCapProduction/.
The work related to argument modeling can be seen as a continuity of the research carried out during my thesis. On the one hand, arguments are considered as sources of user preferences supported by arguments, and on the other hand, the aggregation process takes into account bipolarity.
I have also worked on the argument-based explanation of a query result addressed to an inconsistent knowledge base. An explanation module is important from the user's point of view. One of the feedback got from the argumentation work is that users are often confused by the returned results and therefore express a need for justification. The system should be able to explain why a particular recommendation is returned or why an expected result from the user's standpoint is not returned by the system. This work is carried out in collaboration with Abdallah Arioua, a Ph.D. student at INRA UMR IATE, and INRIA/LIRMM Graphik team.