Semantic Web

Key-words: social network, graph-based knowledge modeling, knowledge-management, learning systems, collaborative work, ontology

The design, sharing and access of the information through Internet becomes more and more an important issue. The research activities conducted in this axis provide a theoretical framework for graph-based knowledge management as well as models and tools to put in practice knowledge management supporting means. During these four years and in the continuation of previous results, one of our main research topic tackle with graph-based Knowledge Representation and Reasoning models for the Semantic Web. The idea of a graph model for the Semantic Web is gradually being adopted but four years ago this approach was quite original in a context where the community of Description Logics was imposing its standards (OWL) and the community of Conceptual Graphs was still ignoring the Semantic Web challenges. Our most important contribution is the generalization process we conducted over various graph models and query languages. In 2008, we participated to the GRIWES color project which collectively provided a preliminary study of a generic model. In the continuation of this project and in collaboration with the Edelweiss team at INRIA we recently proposed an abstract knowledge graph machine interpreting an abstract query language. This language is a generalization of the SPARQL query language for any graph model. We worked also on the assistance and partial automation of the control of conformity of a construction project with regards to building standards. This first implied the modeling of the domain knowledge through ontologies which was four years ago quite innovative for the building domain. Based on these domain ontologies, we modeled the constraints expressed in regulatory standards as a base of semantic queries representing rules. This means the reification of constraints and the capitalization of a query base. A building project is thus validated by matching its model against the queries of the base. By scheduling these queries during the control checking process we account for the usage and therefore represent the control process itself as it is performed by human experts. One of our research topic has been the modeling of the knowledge involved in a learning system. We address the problem of modeling both the domain to be learned, the learner using the learning system, the pedagogical approach adopted in the learning system and the learning system itself. Our ontology-based approach of modeling enables the coexistence of these different kinds of knowledge. We argue that the annotation of learning resources enables their reuse from one learning repository or one learning system to another learning system. We reify and model the learning system and the navigation processes it enables through a base of semantic queries over the knowledge modeled. We adopt semantic web models and technologies and we propose a semantic web-based learning system. Our modeling enables to adapt the system to a chosen pedagogical approach and/or a specific learner profile. We further adapt the navigation possibilities of a user by measuring the relevance of a learning resource to the learner in context with a semantic distance.

Security

Key-words: security properties, secure communication protocols, distributed access control, context-aware access control, access control ontology, intrusion detection, IP traceback, web database, RBAC.

Concerning security management activity, we propose to manage security from a software engineering point of view, i.e. considering the point of view of a developer who is not expert in security. For this kind of developer, it is not easy to model a system including security and/or to add security to a non-secure system/application because of the complexity of security, the lack of knowledge in security field and the lack of time when designing and developing softwares. Our goal is not to develop new security solutions but to propose solutions that will allow non-security expert application designers to easily and efficiently add and integrate security during the development process. We are also interested in the problem of dynamic access control to web databases. We propose (RBAC+), a dynamic access control model to enforce fine-grained access control to web databases. It extends the Role-Based Access Control model standard with the notions of application, application profile and sub-application session. The proposed dynamic access control model enhances the ability of detecting malicious transactions, the dominant cause that demolishes database system, by tracking application users throughout a whole session.

Emotica

Key-words: affective computing, emotion detection, emotion ontology.

Emotion Modelling by Ontology Techniques for Interactive Capture mecanisms with Automatic learning

Biometric techniques are already widely used in security-related applications where physiological characteristics can be used to identify individuals. Such techniques are being applied to other domains as well, for example, domotics, personal health watch, assistance for the elderly or handicapped, interactive video games, public safety, communications, direct marketing, opinion polling. A feature common to all these applications is the detection of a human user's emotions in order to adapt the behaviour of an artificial agent. This raised three questions:

  • Can human emotion characteristics be determined from physiological measurements?
  • Is there a set of readily observable biometric parameters from which human emotional states can be cost-effectively determined?
  • Can these observables and characteristics be modeled for use in interactive adaptive systems?

While significant progress has been made recently in capturing emotional states, this often involves hardware components, such as MRI scanners and encephalography, as well as algorithmic techniques, such as fuzzy logic and nonlinear systems theory, neither of which readily scales for widespread use. The EMOTICA project is a cross-disciplinary research project with the aim of proposing a unified structure for emotion information based on knowledge engineering. The central idea is to establish, for a given task, a correspondence between an emotion state to be detected and a set of physiological responses. This involves two specific research problems: * Collecting emotional data. Emotional phenomena are complex and capturing relevant data can be non-trivial.

  • Analyzing the data for its emotional content, and constructing a representation usable by machine as well as understandable by humans.

Much work has been done in developing complex psychological models of emotions, but it is only quite recently that such models have begun to be machine-usable with the aim of abstraction and re-usability.

In particular, the EMOTICA project aims to:

  • use ontological annotation of biometric data to construct a reference knowledge base of emotional states which would be exploitable in different application domains; and to
  • develop generic software tools for handling emotional events captured by different types of sensors over a variety of emotional tasks.

Requirement Engineering

Key-words: knowledge-management, collaborative work,

Human activities undertaken in organizations rely on information systems. These were first used to collect, disseminate and process information. They are now real knowledge bases. The role of information systems becomes increasingly critical in the functioning of organizations. Information systems engineering aims at providing effective solutions to design, build, use and manage (in terms of evolution and increasing complexity) information systems. Research in this area aims at anticipating methods and tools engineering information systems will need.

To understand, anticipate and align user requirements and, more broadly, stakeholders with available technical solutions (components, services, ERP systems, …) has always been an important and difficult problem in the information system engineering field. The growing number and diversity of stakeholders on the one hand and the rapid evolution of ever-greater range of technical solutions on the other hand explain that this problem remains a major challenge. In this context, our aim is to propose methods and tools to help stakeholders and designers (i.e. technical solutions providers) to articulate their representations and processes.

To facilitate the exploitation of resources (documents, actors or services), the semantic web research community aims at making explicit the knowledge contained into resources. This knowledge is represented by ontologies which structure terms, concepts and relationships of a given domain. Ontologies are often used to extract and represent the meaning of resources. This meaning is expressed through annotations supporting semantic resources indexing in order to formalise and make their content explicit. Resource retrieval relies on the formal manipulation of these annotations and is guided by ontologies. We believe that the semantic web models and languages provide a suitable unified framework to firstly characterize user requirements and technical solutions and also allow reasoning on these characteristics.

Specifically, modeling, capturing and analyzing relations among user requirements pieces as well as between user requirements and technical solutions is of major interest in the requirement engineering field. Thus, our aim is to provide frameworks to formalize, store, index, query and infer from those relations to improve significantly the main activities of requirements engineering (validation, negotiation and evolution).

research_topics/start.txt · Last modified: 2011/01/20 17:00 by mirbel
Trace: research_topics
CC Attribution-Noncommercial-Share Alike 3.0 Unported
Valid CSS Driven by DokuWiki Recent changes RSS feed Valid XHTML 1.0