• Accueil
  • Accueil
  • Accueil
  • Accueil



Accueil > Productions scientifiques > HDR / période 2013-2015

« Supporting Service Consumption : Advanced Discovery and Recommendation techniques »

Habilitation à Diriger des Recherches soutenue le vendredi 19 Septembre 2014 à A l’Université Paris VI - Pierre et Marie Curie, Campus Jussieu, 4 place Jussieu Paris 5ème, Tour 55, 2è étage, couloir 55/65, salle 211 par Walid Gaaloul, Maître de conférences dans le département INF.

Titre:Supporting Service Consumption : Advanced Discovery and Recommendation techniques

Composition du jury  :

Prof. Fabio Casati, University of Torento, Italy (rapporteur)
Prof. Marlon Dumas, University of Tartu, Estonia (rapporteur)
Prof. Lionel Seinturier, Université de Lille 1, France (rapporteur)
Prof. Bernd Amann, Université Pierre et Marie Curie, France (examinateur)
Prof. Boualem Benatallah, University of New South Wales, Australia (examinateur)
Prof. Claude Godart, Université de Lorraine, France (examinateur)
Prof. Mohand-Said Hacid, Université Claude Bernard, France (examinateur)


"The tremendous growth in the amount of available (Web) services impulses many researchers on proposing discovery, recommendation and management tools and techniques to help users retrieve services efficiently. Services can be consumed in different contexts : published in distributed registries ; invoked as individual services which provide interfaces to receive inputs and return outputs ; or composed and integrated into service-based processes as new value added composite services. In our work, we aim at facilitating service discovery and management in these three consumption contexts. First, we propose a functionality-driven approach by clustering and organizing registries according to the functionalities of the service they advertise. Second, to recommend services for individual use, we propose a usage-driven approach that takes into account the user usage data which reflect the user interest. Third, to recommend services for process use, we propose a composition-driven recommendation approach that takes into account the relations between services in service-based processes. We develop applications, as a proof of concept, to validate our techniques. We also perform experiments on the data collected by our applications and on large public datasets. Experimental results show that our techniques are feasible, accurate and have good performance in real use-cases."