HDR of Noura Faci

Service-Oriented Architectures: High Availability, Trust, Sociability


My research agenda over the last 10 years has mainly revolved around
the topic of robustness of service-oriented applications. This meant discovering, selecting,
and composing services that could be faulty at run-time, and hence could make
these applications fail. To mitigate the risks of failure, we examined diversity appropriateness
for successful service-oriented applications.We raised a couple of questions
that are (i) how to define and configure diversity; and (ii) how to select services so
that fault tolerance requirements are met. A preliminary investigation examined how
replication could work hand-in-hand with diversity so, that, for instance many replicate
services (not necessarily the same from a non-functional perspective) could
be grouped together. The next phase was to focus on how to improve the quality of
service discovery in the presence of faults. The idea was to embrace the principles
of social networks to select the most appropriate services, thus endowing them with
some social characteristics like who collaborates with who. The work carried out
allowed defining and setting up social networks of services and extracting relevant
details from these networks for the needs of discovery. To ensure proper use of social
networks, another avenue of research consisted of looking for ways that would
regulate services’ actions (e.g., establish and maintain networks of contacts) using
the concept of commitments. In fact, robustness at the level of “service” is necessary
but not sufficient. Thus, it is important to include the robustness in the selection
process itself as well, and thus, extends our reflection process. We raised the question
of how to evaluate a service’s trust. Trust systems are mostly based on user/service
experiences when requesting these services. The work resulted in the definition of a
credibility model based on a fuzzy clustering, a mechanism for filtering users with
multiple identities, and a trust model based on probabilistic databases.
It is well known that service-oriented architectures and business process management
go hand in hand for developing enterprise applications. We deemed appropriate
exploring the weaving of social principles like proximity into the design and
execution of business processes so, that, business continuity is guaranteed. The main
concern is how to ensure perfect alignment of Web 2.0 technologies with business
development strategies and best practices. The work carried out fostered the social
relations among an enterprise’s employees to perform better. Indeed, it was noted
that informal relations between people exist in enterprises at strategic, management,
and operational levels. We address the questions of (i) how to design business processes
based on social principles and (ii) how to ensure effective execution of these
processes during resource conflicts. A preliminary research phase was to develop
dedicated networks based on social relations (e.g., supervision and partnership) between
a process’s three components (task, person, and machine) and to analyze the
added value of these networks to enterprises.. These networks capture the different
situations of collaboration between tasks, between persons, and between machines.
The next phase was to address resource management in enterprises. The idea was
to capitalize on dedicated networks to coordinate resource production, consumption,
and use. The work allowed: (i) to categorize resources according to their nature and
type, (ii) to identify conflicts per resource category, and (iii) to propose solutions to
these conflicts by using the appropriate networks. Guaranteeing the stability of these
networks is a necessity. To this end, the proposed approach relies on business and
social commitments to regulate the functioning within these networks.
For the next 4-years, my research agenda will tackle new challenges related to
Internet of Things, (r)evolution of the Web, such as on-the-fly combination of things
and cognitive things. Some early findings have already been shared with the community.

Keywords Service composition, fault tolerance, diversity, trust management systems,
robustness, business process management, social networks.

Defense date: friday, december 14, 2018

Jamal BentaharProfesseur(e)Concordia University, CanadaRapporteur(e)
Khalil DriraDirecteur(trice) de rechercheLAAS-CNRS, ToulouseRapporteur(e)
Daniela GrigoriProfesseur(e)Université Paris-Dauphine, ParisRapporteur(e)
Djamal BenslimaneProfesseur(e) Université Claude Bernard, LyonExaminateur​(trice)
François Charoy Professeur(e)Université de Lorraine, NancyPrésident(e)
Walid Gaaloul Professeur(e) Institut Mines Télécom, ParisExaminateur​(trice)
Franck Morvan Professeur(e) Université Paul Sabatier, ToulouseExaminateur​(trice)
Marta Rukoz Professeur(e)Université Paris Ouest, NanterreExaminateur​(trice)