Overall Objectives
Scientific Foundations
Application Domains
New Results
Contracts and Grants with Industry
Other Grants and Activities

Section: New Results

Keywords : Corporate Memory, Corporate Semantic Web, Cognitive Sciences, Knowledge Acquisition, Knowledge Management, Knowledge Engineering, Ontology, Assistance to the User, Cognitive Psychology, Communication, Co-operation, Human-machine interaction.

Support to Modeling and Building of a Corporate Semantic Web

The objective of this action is to propose methodological and software support for the construction of a corporate memory, thanks to a user-centered approach. We study in particular the construction of a corporate semantic Web and the construction of ontologies and annotations from human and textual sources of expertise or from databases. Moreover, we study how to handle multiple viewpoints or multiple ontologies and how to take into account the life cycle and the evolution of a corporate semantic web. We also study e-Learning as a specific scenario of knowledge management. Recently, we started to study how to tackle the context and privacy of the (possibly mobile) user and how to develop and use ``Corporate Semantic Web Services''.

Designing User-Adapted Semantic Web Applications

Participants : Sophie de Bonis, Olivier Corby, Karine Delêtre, Alain Giboin (resp.), Fabien Gandon, Thierry Grandsart.

Initially concerned by formal and technical aspects, the Semantic Web community recently acknowledged the necessity to take Semantic Web applications' uses and users into account for the applications to be accepted by users and their organizations (cf. International Workshop on Interaction Design and the Semantic Web (2004), and the Workshop on End-user Semantic Web Interaction (2005)). The goal of our actions is clearly to propose methods and models to help the user-oriented design of semantic web applications; in particular, importing and adapting methods from the Human-Computer Interaction (or software ergonomics) and CSCW communities to the Semantic Web community and to the ontology engineering community. This goal was evident in various projects such as KmP, Knowledge Web, QBLS, and UsableIntranet. Below are some of the actions we undertook this year:

Introduction of user-oriented evaluation methods in a toolkit of Semantic Web application designers

We took part in The Knowledge Web deliverable D1.2.3 to do so [57] . This deliverable analyzes and evaluates existing methods for ontology content evaluation according to requirements from an industrial point of view. A classification of methods and tools for the evaluation of ontologies for industrial practice is provided. It concerns methods and tools to: (1) select existing ontologies (possibly from libraries); (2) measure the correspondence between textual sources and the corresponding ontology; (3) evaluate the impact of an ontology on an information retrieval application; (4) check and improve the quality and consistency of ontologies; (5) monitor an ontology in use. The methods and tools intervene at different stages in the industrial life cycle of a software product. For each method and tool, its scientific basis and design purposes as well as its relevance and usefulness for industry are presented.

Adaptation and extension of the ergonomic principles proposed for designing Web applications

The objective is to rely on ergonomic principles (e.g. Nielsen's heuristics, Bastien and Scapin's ergonomic principles) to help design usable Semantic Web applications. We focus in particular on the principles dealing with "semantics" [54] , [53] , [58] . We are currently discussing with Chrisian Bastien (Université Paris V) to analyze the possibility to collaborate on Semantic Web ergonomic principles.

Adaptation of scenario-based methods to the design of Semantic Web applications

We were particularly concerned this year with the story-boarding and mock-up techniques, in order to cope with the question of translating semantics in a Semantic Web interface [44] , [56] , [55] , [58] .

A method allowing to keep in touch with users during a Semantic Web application development project

A major problem met during the design of Semantic Web applications, even by defenders of a user-oriented approach, is to constantly keep in touch with users. Relying on our experience of KmP, we analyzed cases where the contact with users was taken, lost, and taken again, and identified some techniques to constantly keep in touch with users [44] . Our analysis and identification is based on the notion of "use(r) representations", which is related to the notion of "corresponding representations" we previously introduced (see Acacia 2004 activity report).

The "Correspondences" framework

We have published our work on the use of the Correspondences Framework [72] , [73] to analyze and support the coordination between producing and understanding or using electronic documents constituting a part of the memory of some organization [21] .

Corporate memory and semantic web for the Transcriptome Analysis

Keywords : Corporate memory, Semantic Web, Natural Language Processing, Biochip experiments, Ontologies, Semantic annotations.

Participants : Khaled Khelif, Rose Dieng-Kuntz.

This work is carried out in the context of Khaled Khelif's thesis [45] , [46] that illustrates the scenario of experiment memory for a scientific community.

The study of gene expression has been greatly facilitated by biochip technology. Biochips can assess tens of thousands of genes simultaneously and lead to a huge amount of information: for example, information about the roles played by particular genes in drug sensitivity and the effects of drugs on gene expression. In the framework of a collaborative project with biologists working on biochip experiments at IPMC (Institut de Pharmacologie Mol culaire et Cellulaire), we aim at assisting them in their experiments and facilitating their validation and interpretation of obtained results. Our objective is to propose methodological and software support for capitalization and valorisation of knowledge resulting from experiments and techniques to preserve and reuse data (structured documents, semantic information retrieval). We rely on the techniques of semantic web (semantic annotations, ontology...) and knowledge engineering. After studying biologists' needs, we proposed to build an experiment memory and to materialize it through a  corporate semantic web. The main modules of this memory are:

The annotations generated by MeatAnnot were validated by biologists and obtained good scores (82% of precision, 62% of recall and 96% of usefulness).

Our method can be generalized to any life science domain (e.g. chemistry, physics) having similar needs of support to validation and interpretation of experiments results.

The originality of this work consists of (a) the integration of metadata on annotations which gives new ways of reasoning and more information on the annotation base, (b) the use of several technologies (such as NLP, Ontologies, Semantic annotations, Corese) to build a real-world Corporate Semantic Web Application.

Semantic Relation Acquisition from Biomedical Corpora for Ontology Discovery

Keywords : Syntax-semantics interface, ontology and annotation learning, lexico-syntactic information extraction, semantic schema acquisition, NLP, Unification-grammars, causation, Semantic Web.

Participants : Laurent Alamarguy, Rose Dieng-Kuntz, Catherine Faron-Zucker.

This work is performed in the framework of Laurent Alamarguy's PhD thesis [25] , [26] . This work deals with the acquisition of semantic relations from biomedical corpora for the construction of ontology and annotations. This aims at elaborating a methodological support and a tool to enhance the automation of ontology construction and enrichment from linguistic comprehension of texts in order to develop a community memory in biomedical area.

We synthesized a conceptual acquisition method, INSYSE, aimed at proposing some salient semantic relation schemas to domain experts who supervise the ontology and annotation construction.

This method follows the linguistics processing philosophy and is based on several main stages dedicated to the syntax-semantics interface. This interface is elaborated through the formalism of the grammatical parser, PATR-II, based on unification grammars that are performed through lexicons constituted by the results of term extraction and shallow processing that reveals different kinds of abstract lexico-syntactic information, and using grammars that we manually determine upon causative construction of salient semantic markers. This interface generates some candidate conceptual schemas that are transcribed in RDF(S) in order to be used with Corese semantic search engine.

This year, we focused on developing a prototype of INSYSE system, a knowledge extraction tool based on our causal semantic relation acquisition method. Concerning the Natural Language Processing aspect, we worked on enhancing and developing our grammatical rules governing the syntax-semantics interface. Particularly, we aimed at analyzing and modeling linguistic phenomena such as valence alternations designing the transition from active form to passive form, or from nominal form to support-verb form. To do so, we explored again our training corpus more accurately, so as to propose a more relevant and conspicuous linguistic analysis favoring a better modeling. Concerning the realization of our prototype, a workbench homogenizing the different modules of the system has been elaborated, and we particularly paid attention to problems related to the merging between term extraction and shallow syntactic parsing.

Moreover, the realization of the workbench gave us the opportunity to submit a training subject to second year engineering ESSI students. In the framework of this training project, we co-supervised a four-students team which proposed coding solutions mainly based on Java.

A paper detailing the elaboration of the system, submitted to the community of Natural Language Processing in Biomedical area, was published in [26] .

Construction of a multi-point of view Semantic Web

Keywords : Semantic Web, Ontology, Ontology Matching, Multiple Viewpoints.

Participants : Thanh-Le Bach, Rose Dieng-Kuntz.

This work is carried out within the context of Thanh-Le Bach' s PhD.

The objective of this thesis is to build and use a multi-viewpoints semantic web in a heterogeneous organization, based on multi-viewpoints ontologies. These ontologies can be created from existing ontologies by integrating them or merging them. That raises the problem of ontology alignment or matching: finding entities which are conceptually similar in the ontologies to be compared.

During this year, we continued our work on ontology alignment algorithms. We proposed and implemented two new algorithms based on the ontology structural information as well as linguistic information that we can extract from ontologies. The algorithms focus on the matching/alignment of two ontologies represented in OWL, so that they can take advantage of the well-defined underlying meanings of OWL constructs, such as owl:equivalentClass, owl:cardinality... The first algorithm uses local structure information, which can be extracted from an entity description to deduce the similarity between two entities. The second one uses the structure information extracted from the whole ontology structure: it represents OWL ontologies as graphs then finds their largest common subgraph. The similar entities can be deduced from that largest common subgraph [28] . The algorithms are implemented in Java and experimented with testing ontology pairs at I3CON (the Information Interpretation and Integration Conference) . The first results are interesting, but the algorithm still needs to be optimized for better performance when running with larger ontologies. For representing multi-viewpoints ontologies, we also proposed an extension of OWL ontology language, so as to represent multi-viewpoints ontologies based on the C-Vista model proposed by Myriam Ribière [79] .

Management of Corporate Semantic Web Evolution

Keywords : corporate memory, corporate semantic web, evolution, ontology evolution, semantic annotation.

Participants : Phuc-Hiep Luong, Rose Dieng-Kuntz.

This work is being carried out within the framework of Luong Phuc Hiep's PhD, that, based on the analysis of the life cycle of a corporate semantic web (CSW), aims at solving some problems related to its evolution: evolution of each component (resources, ontologies and semantic annotations) as well as evolution of relations among these components.

During this year, we studied the problems of change management and evolution. Organisations evolve in dynamic and changing environments because of the changes in their business, technologies and processes. These changes in the real world often result in needs of modifications of the CSW. Moreover, when one of three main components of a CSW is changed, it might impact the consistency of other components or of the overall system. In this case, other related parts may need to evolve after such changes in order to re-establish the consistency of the CSW. We mainly focus on the influence that ontology modifications can have on semantic annotations based on these ontologies and, in particular, on the evolution of these semantic annotations.

We studied related research on schema evolution in databases and in knowledge-based systems, and on main techniques dedicated for both levels of schema and instance evolution, we studied the change management for distributed ontologies and some existing methods and tools for ontology evolution and ontology versioning. After reviewing this related work, we have tried to give some propositions with the purpose of resolution of inconsistency on the relation between a semantic annotation and the ontology in case of ontology modification. We have proposed the architecture of a system enabling to manage the evolution of such a CSW when its ontologies or the semantic annotations of its resources change. This proposed evolution management system will focus on the evolution of each component, on the relations among these components and on the propagation of the ontology changes towards the semantic annotations depending on this ontology. In [47] , we describe such an evolution management system for a CSW by giving its architecture system working cooperatively with a CSW. We also proposed a model of consistency including some defined invariants and a mechanism of verification allowing to check whether a semantic annotation is still consistent with respect to its ontology. In order to solve inconsistencies of annotations when the ontology changes, we have constructed all the possible solutions for each ontology change operation that might impact the consistency of annotation. Finally, we introduced a process of propagation for ontology changes towards their related semantic annotations. This process supports both versioning and evolution problems for the update of semantic annotations.

We will focus on a real scenario with evolving ontology (e.g. UMLS) and based on RDF(S) language. We will try to formulate invariants of consistency and propagation rules through Corese rules and Corese query language.

Support to Cooperative Work: Application to a Healthcare Network

Participants : Karima Aissiou, Rose Dieng-Kuntz.

Our previous work performed in the framework of the ACI Ligne de Vie project aimed at developing a knowledge management system for a healthcare network, in order to ensure care continuity and support to collaborative work of the actors of the network, was published in [20] , [50] . We extended the virtual staff for enabling to store the results of a session in RDF(S) and to use Corese query language for asking queries about past sessions of virtual staff.

Semantic Web for E-Learning

Keywords : E-learning, Ontologies, Annotations, Pedagogical Resource Retrieval, Pedagogical Resource Composition.

Participants : Sylvain Dehors, Catherine Faron-Zucker, Alain Giboin, Stéphanie Mevel.

This work takes place in the framework of Sylvain Dehors's PhD.

During this year the following aspects have been explored through both bibliographic and experimental activities. The main experiment conducted this year is called QBLS, an acronym for "Question Based Learning System". It results from a fruitful collaboration with a teacher at ESSI school, Jean Paul Stromboni. During this work, we studied both methodological and technical aspects of putting and accessing "online" semantically enriched course material. This experiment fits in a broader reflection on the application of the semantic web, dealing with the following issues: ``What kind of information does a teacher really need? How do I get them? What to do with them?'' To answer those key questions the following points were investigated and concrete solutions proposed.

Ontologies for e-learning

The Semantic Web proposes to represent and share knowledge by means of ontologies. Even if the use of ontologies for learning is described quite extensively in the literature, we found that it is still very unclear how ontologies can be integrated in learning systems and effectively used otherwise than in very basic mechanisms. In the QBLS experiment we focused on that issue and showed the establishment and real usage of a "pedagogical ontology" used to describe learning resources.

Semantic annotations for E-learning

Here we have let aside all the technical and legal information which has been already extensively studied by standardization bodies like IEEE and IMS to focus on annotations with a high pedagogical value. Through the QBLS experiment which started in November 2004 we have proposed a method based on the analysis of the layout of existing documents to extract and generate semantic information about those documents. The rationale behind this method is to automate as much as possible the annotation task. This also turns the usual textual editor (Word or OpenOffice) into a semantic annotation editor.

Semantic Web Tools and architecture

The semantic Web benefits from the effort of numerous research teams around the world as well as the standardization effort supported by W3C. By reviewing the TRIAL Solution project we showed that tools like Corese and standards for web services would greatly enhance performance and readability at the architecture level in not-so-new learning applications. The QBLS experiment also demonstrated the power of the emerging "semantic web server" (Sewese) developed by Acacia allowing a fast and efficient deployment of this application.

Interface for Semantic Web E-learning application

The last but not least of the aspects dealt with during this year is the requirements for user interfaces dedicated to "semantic enhanced learning". By collaborating with a student in ergonomics, we designed a specific interface for QBLS, putting in practice the need for the interface to : (1) closely support the usage scenario of the system and (2) hide the complexity of the semantic mechanisms used by semantic web tools. By staying in close touch with the reality of pedagogical practice we were able to propose a simple yet coherent and efficient system that is described in several publications accepted this year [37] , [35] , [38] , [36] .

Corporate Semantic Web Services

Keywords : Semantic Web, Web Services.

Participants : Moussa Lo, Fabien Gandon.

We were granted an AUF funding for a post-doctorate exchange with University Gaston Berger of St Louis (Senegal). As a result the researcher Moussa Lo visited our team for six months and collaborated with us to start a new research topic: Corporate Semantic Services i.e. web services in corporate semantic web to integrate and unify the access to corporate knowledge, applications and services [43] . This work was an experiment in integrating semantic web services in the existing semantic web server architecture of Corese to provide web applications based on the semantic web services it can identify. Thus, Corese is used as a semantic UDDI (Universal Description, Discovery and Integration) registry and allows us to automatically discover and invoke corporate applications wrapped into semantically annotated web services. Using rules and an extension to the existing semantic web service frameworks, we also demonstrate how (i) to compose the web services with queries on the knowledge stored in the corporate memory to automatically populate the service inputs and (ii) to turn Corese into a composable service of the memory [52] .

Mobility, Context-Awareness and Privacy

Keywords : Context, Awareness, Privacy, Mobility.

Participant : Fabien Gandon.

We continued our cooperation with the Mobile Commerce Laboratory of the University of Carnegie Mellon. Fabien Gandon spent a week in CMU (March 2005) to help finalize and release the e-Wallet engine implementing a secure and unified interface to access knowledge and services published by a person. This work was also integrated in a project with the III institute of Taiwan and the Computing Media and Communication Laboratory of Carnegie Mellon in an application to mobile tour guide in museums. This work was presented at a GTMob meeting ( ) and at a symposium organized by the CNRT Telius (15/06/2005) and will be published in a book chapter to appear.


Logo Inria