Team Edelweiss

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

Section: New Results

Interaction Design


Participant : Freddy Limpens.

During this 3rd year of the PhdD on Multi-points of view semantic enrichment of folksonomies, the focus was mostly on the NiceTag ontology and the achievement of our approach to semantically structure tags that was developed in the PhD dissertation. The NiceTag ontology aims at overcoming the lack of expressivity of current tagging models. Indeed, the model of tag we developed in collaboration with Alexandre Monnin and David Laniado, aims at covering the diversity in form and usages of the tags. This model considers tags primarily as a link, typed according to the use of the tag, between a tagged resource and a sign used to tag. Then we propose using named graph to embody this record and type it in order to account for other dimensions of tag actions. This model has been presented in two french-speaking conference [29] , [24] and an international conference [21] .

The PhD dissertation [14] presents in details our multi-points of view approach to the semantic enrichment of folksonomies [20] . We propose a socio-technical system, grounded on a usage analysis, and combining automatic processing of tags and users' contributions through user-friendly interfaces. Automatic processing of tags allows bootstrapping the process by using a combination of a custom method analyzing tags' labels and adapted methods analyzing the structure of folksonomies. The contributions of users are described thanks to our model SRTag (Semantically Related Tag) that allows supporting diverging points of view, and captured thanks to our user friendly interface allowing the users to structure tags while searching the folksonomy. Conflicts arising between individual points of view are then detected and temporarily solved by an automatic agent, whose outcome is then exploited to help a referent user maintain a global and coherent structuring of the folksonomy. Each individual point of view can then be enriched with the others' contributions, with the global point of view serving as a reference to guaranty a local coherence for all users. The result of our method allows enhancing the navigation within tag-based knowledge systems, but can also serve as a base for building thesauri or ontologies fed by a truly bottom up process, providing therefore a solution to the bottleneck effect of knowledge acquisition.

Mobile Access to the Web of Data

Participant : Luca Costabello.

The PhD Thesis, directed by F. Gandon & I. Herman (CWI and Semantic Web Activity Lead at W3C), is focused on end-user interaction with the Web of Data from mobile devices. As initiatives such as Linked Data grow, we face the challenge of designing novel end-user services that benefits from this heterogeneous, unbound amount of interconnected resources. Moreover, the evolution of mobile devices raises new questions and enables brand new ways of accessing the web of data: mobile user context plays a central role, since multiple data dimensions, such as location or time, are crucial in enabling the design of new tools and interaction paradigms in the mobile environment. In these early steps of the work, efforts have been focused on the state of the art and on choosing the most suitable approach to the problem.

Social Network Analysis

Participants : Guillaume Erétéo, Fabien Gandon, Michel Buffa.

In the first two years of this PhD thesis, we developed a method to conduct a Social Network Analysis (SNA) that takes benefits of the ontological primitives that type the rich graphs formed by the RDF descriptions of online social networks [8] . Building on top of our results on semantic social network analysis, we designed a community detection algorithm that takes benefits of the semantic information of the RDF description of social network. This algorithm not only detects communities, but also labels them with meaningful terms.

Community detection algorithms only consider the graph structure of social networks [45] . Recently, the RAK algorithm [47] proposes to trap random labels in communities by propagating them through links of the graph. In addition, social web applications made folksonomies popular: users annotate online resources with freely chosen keywords called tags, and each tag may label a community of interest composed of all the actors using this tag. Moreover methods were developed to help online communities to enrich folksonomies with semantic relations between tags [14] , which may help refining the labeling of communities.

We defined an algorithm that merges both approaches in order to perform community detection that takes benefit, not only of the link structure of the social network, but also of the community effect and labeling value of folksonomies and their emerging semantics. Our algorithm, called SemTP (Semantic Tag Propagation), is a variant of the RAK algorithm that is described above. SemTP turns the label propagation mechanism of RAK into a semantic propagation of tags. The basis consist in (1) assigning the tags that the actors used to annotate resources, and (2) propagating them by handling semantic relations between tags (e.g. skos:narrower). The Figure 4 shows a toy execution of this algorithm.

Figure 4. Toy example of a semantic tag propagation

We tested and validated this algorithm on the social network of all the Ph.D. thesis funded by the ADEME agency from which we extracted 1853 actors, 13982 relationships, 6583 tags and 3570 skos:narrower relations between 2785 tags.

Knowledge Sharing

Participant : Isabelle Mirbel.

In collaboration with the MODALIS research team at I3S (CNRS & University of Nice-Sophia Antipolis) we proposed an intentional approach to enhance sharing and reuse of in-silico experiments.

In-silico experiments are traditionally automated through scientic workfows. While those systems effectively tackle the complexity of underlying technologies, they themselves do little to ease and organize sharing and reuse. Indeed, scientic workflow models blur the line between user goals and techniques and often mix abstraction levels.

As our aim in this context is to assist knowledge sharing between users of various technical and process modeling levels, we relied on our previous work about intentions and information retrieval, in which we proposed an approach to support knowledge transfer about search processes from experts to novices. In this previous work, starting from an intentional process modeling formalism, we provided an ontology to annotate search processes.

To enhance sharing and reuse of in-silico experiments, we first extended the chosen intentional process modeling formalism in order to support the modeling of in-silico experiments. We extended it in two ways: (1) we made fork specification both explicit and mandatory so as to allow automatic handling of modeled processes and (2) we introduced the notion of parameterized process parts to further ease and enhance reuse and repurposing. Then we extended the ontology previously proposed to annotate search processes in order to reflect the changes we made to the intentional process modeling formalism. We applied our approach to several in-silico experiments represented by scientific worklows on As a result, intentional specifications of in-silico experiments can be shared and exploited by reasoning on their representations.

Models and Methods for Representing Groups of Individuals and Their Activities

Participant : Alain Giboin.

Models and Methods for Representing Collective Personas

Context of the work: ISICIL project.

Individual Personas are user models that are represented as specific, realistic humans. Collective personas are models representing specific, realistic groups of people as such (e.g., teams, communities). We updated our review of the existing methods for elaborating collective personas, and reported remaining issues about how to elaborate such methods, e.g., How to give a realistic aspect to a collective persona? Do we need to elaborate collective personas or is it enough to develop the collective dimensions of individual personas (e.g., personas' relationships, roles, etc.), and/or to develop the collective scenarios related to individual personas acting collectively?

Models and methods for Representing Relationships between Individuals

Context of the work: AVISICIL project, in collaboration with researchers from the Kewi team (I3S, UNS) and from the Laboratoire de Psychologie Cognitive et Sociale (UNS) who are involved in affective computing design projects (designing systems intended to help elderly people maintain their relationships, or autistic children to build relationships with others).

We initiated a literature survey on relationships models and vocabularies: models of, and vocabularies describing interpersonal relationships, and models and vocabularies describing underlying aspect of interpersonal relationships such as emotions. We paid particular attention to vocabularies developed by Semantic Web and Social Web communities (e.g., EmotionML markup language; FOAF ontology).

Models of Shared Representations and Representation Sharing Processes

Criteria for Assessing the Effectiveness of Shared Representations Context of the work: ISICIL project; in collaboration with Florie Bugeaud and Eddie Soulier (UTT).

Any collaborative work depends on the quality of representations shared by collaborators (mental or internalized representations, material or externalized representations), especially representations of the object of the work. In the context of a broader study on the role of shared representations in innovation projects teamwork, we studied how to better characterize the effectiveness of a shared representation, relying on the existing literature surveying the characteristics of effective shared representations, boundary objects, intermediary objects, or related notions (see, [41] , [42] , [48] ). Our aim is to enrich the set of criteria for evaluating the representation effectiveness that have been used hitherto and to structure these criteria in a coherent framework [26] .

Comparing and Bridging Models of Shared Representations and Representation Sharing Processes

Context of the work: GDR CNRS Psycho Ergo( ), Groupe thématique Coopération homme-homme et Coopération home-machine Action de recherche RefCom (Réferentiel commun, Intelligibilité mutuelle, Contexte partagé, Team/Situation Awareness... : quel(s) rapport(s) ?) ; in collaboration with Pascal Salembier (UTT).

The notions of shared representations and representation sharing processes are examined in the research literature from several points of view; this variety of viewpoints gave rise to different conceptualizations, which are referred to using such terms as Common Frame of Reference, Mutual Intelligibility, Shared Context, Team/Situation Awareness, etc. In order to achieve mutual intelligibility between researchers working on such conceptualizations, we elaborated a grid for collaboratively comparing and bridging them. This grid consists of a set of dimensions allowing to characterize shared representations and representation sharing processes (e.g.: individual and collective attitudes towards sharing (social, relational, emotional aspects) ; dynamics of sharing (proactiveness in managing the sharing)) and the approaches used to perform this modeling (e.g., theoretical assumptions ; aim of the modeling) . The grid is currently applied by participants of the RefCom joint action research.

Webmarks: Revisiting the Notion of Bookmarks Toward Contextualized Reference on the Web

Participant : Nicolas Delaforge.

In the early days of the web, we were many to think that we were witnessing the advent of the giant global library. This librarian vision led to a modeling of web content (pages) which was propitious to the development of web tools, concepts and terms directly inherited from the paradigm of documentation. With web applications, web 2.0 and the rise of social web applications, the nature of the web evolved considerably and eventually the documentary vision of it became obsolete or at best very much confined. However, some remnants of this ancient metaphor remain and one of their best representatives certainly is the bookmark.

On the web, every resource is accessed by an URL. By analyzing both the technical and editorial web specificities we show that the web is not, from our definition, a documentary space. Due to their documentation legacy, bookmarks are built on top of the old archivistic principle "access to support give access to content" and they should be reconsidered as well. For instance, as web pages are dynamically generated, the transposition of this principle to the web gives a drastically different result : "access to support potentially generates a new content at every access". In fact, on the web, a bookmark does not give access to a content. It is more like a road sign that points to a web place where something is happening.

We believe we can improve the traditionnal bookmark system by a deeper understanding of the true nature of the web. Starting again from the documents of the W3C Technical Architecture Group (TAG), we propose a model of the web as a superposition of three spaces: the web pages, web of data and web of services. Then we focus on the inherent intentionality behind the act of marking a URL and we show that in most cases, the original intention has nothing to do with the initial documentary practice of book-marking. We finally propose a definition of what we called "webmarks" which allows us to envision innovative and contemporary applications.

We suggest to define several Webmark types : a reference mark, a location mark, an application mark [25] . Each of these marks should be managed in a specific way, with dedicated tools offering adequate functionalities. A "Reference Mark" can be scrapped from the web and stored on a server. Then a stable access and a referencable URL could be garanteed. A "Location Mark" is used when a user wants to mark a social place, like a pin on a map. We could propose a personal web map of the user's social web and provide notifications when social activites are detected. An "Application Mark" works more like a shortcut on a web desktop. It could be placed on a personal dashboard for instance and other specific services can be proposed according to their public API (Google Maps, Facebook,...).


Logo Inria