Section: New Results

ISICIL Platform

Participants : Nicolas Delaforge, Michel Buffa, Fabien Gandon, Alain Giboin.

ISICIL is an ANR project studying social networks and Semantic Web communities to support corporate intelligence.

First, ISICIL proposes a multidisciplinary design of a new form of corporate intelligence. The challenge of this project is to reconcile the new viral Web applications with formal representations of business processes and to integrate them into practical intelligence communities of actors working in the company. We designed, studied and experimented with new tools to support collaborative tasks in business intelligence by leveraging Web 2.0 interfaces (blog, wiki, social bookmarking) for interactions and Semantic Web technologies for interoperability and information processing. ISICIL also allowed to explore new scientific developments of the notion of epistemic cooperation (human interaction oriented toward the development and transmission of knowledge) and to identify usable technological solutions. An ergonomic approach, combining impregnation of ground truth data and freer technological inspirations from bibliographic and webographic sources, was proposed.

Secondly, ISICIL uses typed graphs as models underlying epistemic communities. The entire model relies on a unifying model based on RDF graphs to represent resources and community stakeholders. These models are integrated with bookmarking tools or "Web scraping" the results of which are tagged. The tags used are collected to form folksonomies and a semi-automatic thesaurus structure in these folksonomies. User feedback on this structure is captured when they use the search engine which offers tags related to their keywords and the user can accept, reject or adjust these suggestions. User profiles and links between them, considered as a network, are processed by a series of operators to propose a semantic analysis of social network, for example the computation of indicators of centrality parameterized by an ontology. Merged graphs of structured folksonomies and of social networks finally allow the detection and labeling of epistemic communities. Meanwhile we study how the analysis of user interactions can determine the trust and how to represent and control access to data and their semantics in a social network.

As part of the paradigm of Social Epistemology, ISICIL combined in an ergonomic approach, impregnation of ground truth data, a bibliometric study and technological inspirations to offer patterns and inferences exploiting the Semantic Web social networks to assist corporate intelligence. An open-source platform is available under CeCILL-C licence and was tested at ADEME and Orange. ISICIL is a proof of concept of the compatibility of the Semantic Web formalisms, practices and models of Web 2.0 and the philosophical framework of social epistemology.

The project has resulted in three PhD theses (Florie Bugeaud 2011, Guillaume Erétéo 2011 [98] , Freddy Limpens 2010 [99] ) and publications at the following conferences: ISWC 2009 IEEE / WIC / ACM Web Intelligence 2011, Web Science 2010, WWW 2011, ASWC 2009, COOP 2010, PRO-VE 2009, VSST 2010, EGC 2010, IC 2009 & 2011, Psycho Ergo Days 2010.

The ISICIL project participated to Colloque ANR STIC, January 4-6 2012.