Past projects

FIORA

Description

Fiora project aims at designing and developing of a customized recommendation engine. This engine will provide the most relevant content for a given user with maximum reliability. For this, the results returned will be tailored to each user depending on several factors such as contextual data related to those users and their environment and the available semantic information.

The main result of this project will be to reconcile within the same engine approaches like case-based reasoning, technical recommendations using collaborative filtering, and the management/validation techniques of recommendations using data mining. A key step for this reconciliation between disjointed disciplines is the construction of a formal application where data and knowledge are handled and processed. In addition, the Fiora project will develop “scalable” technical recommendations coupled with Big Data technologies and formatting data (trellis or ontologies) in order to implement a non-intrusive, scalable and distributed recommendation system.

Finally, the proof of concept will be provided within two distinct domains: “Nutrition & Health” and “e-tourism” applications.

Duration

36 months (December 2012 – November 2015)

Partners

  • ISEP: Raja Chiky, Matthieu Manceny, Zakia Kazi-Aoul
  • Kiolis (leader)
  • Éditions Solar
  • Centrale Recherche SA
  • Paris XIII University

MCUBE

Description

MCUBE is supported by the European Union. Europe is committed in Ile-de-France with the European Regional Development Fund.

Generally, Machine-to-Machine (M2M) applications consist of communication systems from on-site sensors and a main server through wireless networking such as GPRS or EDGE. Application domains of M2M are numerous (healthcare, vehicular and transport applications, tele-surveillance). Likewise, MCUBE is a Multimedia Platform using Machine-to-Machine technology as an implementation of distributed processing service. The treatments use multimedia data such as photos and sounds that are sent to the platform from M2M Gateways. These gateways can be connected to various sensors such as cameras, camcorders, and microphones to collect data and send them to the platform via web services.

MCUBE-Gateway

Treatments require a large amount of data or simply a capacity greater than that provided by the gateways. Treatment can be carried out on the platform and the platform provides an interface to store the results of these treatments. MCUBE platform has been developed with components for M2M such as centralized web platform to store multimedia libraries and data, load balancing platform using cloudizer (see section 3.3), Web services generator, event and rule based programming language (see section 3.2). The various algorithmic libraries, configuration, storage, compilation and deployment are managed from the MCUBE platform.

Duration

42 months (October 2010 – April 2014)

Partners

  • ISEP (leader): Raja Chiky, Bernard Hugueney, Matthieu Manceny
  • WebDyn
  • Cap2020
  • SolarNet
  • Partnerships for supervising 2 theses: CNAM, Telecom ParisTech

BSS

Description

BSS (Bike Sharing System) are very famous today as they are used in more than 400 towns all over the world. However BSS, in particular Vélib, still has some major problems, consisting mainly of finding no bike or no free place in the station when a user wants to take or park a bike.

The objective of the project is to propose new methods to improve the availability of the Vélib system resources. The final goal is to increase user satisfaction and reliability of Vélib. To achieve this goal, it is essential today to ensure a better distribution of bikes between stations by compensating the heterogeneity caused by the strong attraction (proximity to a railway station or to a monument…) of certain areas or the high altitude in other areas.
Two fields of research will be explored: First, we will propose technical incentives to encourage users to take bikes in the most populated stations and place them in the most empty. Then we will focus on the redistribution of bikes between the stations using trucks.

We are now analyzing data issued from Vélib system in order to quantify and well understand the usage of Vélib system today. We noticed many periodicities related to human daily activity. The stations can also be classified into several categories (clusters) according to their attractively (residential area, working area, parks, railway station…)

Duration

24 months (2013-2015)

Partners

  • ISEP: Yousra Chabchoub
  • INRIA Rocquencourt (leader)
  • IFSTTAR
  • LIX
  • Nanterre University