Section: Contracts and Grants with Industry

Industrial Contracts

EPIA: a RNTL Project (2003-2007)

Participants : Semi Gaieb, Yves Lechevallier, Bernard Senach, Doru Tanasa, Brigitte Trousse [ resp ] .

Inria Contract Reference: S04 AO485 00 SOPML00 1

The EPIA project ``Evolution of an Adaptive Information Portal'' got labeled by RNTL 2002, and started on September 2003 until june 2007. Partners are Dalkia, Ever(Mediapps) and Inria.

The objectives of this project are the following:

• Supporting users of Mediapps.Net (tool for selecting canal information of an extranet) via clustering clients. This task started in 2004 and some generic algorithms and pre-processing tools were developed until the beginning of 2005. Some log analysis haven't been done because of the unavailability of real data.

• After understanding the user needs for Net.Portal (construction tool for intranet portals), we finished the specification of the trace of the NetPortal engine (cf. the first version of the deliverable D3: ``Experimental context and trace engine in Net.Portal''.). The result of this work is the description of the Net.Portal relational database schema and the data organization. The specification of the Net.CanalRecommender was stopped and studied in the new context of the eversuite software.

The project was re-oriented mid 2006 in order to take into account Ever wishes and the future integration of Mediapps.net and Net.portal in the eversuite software. This year we studied the trace engine proposed by Ever (written in Java), its use in the context of Net.Portal and the specification of a recommender system for information sources in this context.

MobiVIP: a PREDIT Project (2004-2007)

Participants : Sergiu Chelcea, Christophe Mangeat, Ghuilaine Clouet, Bernard Senach [ co-res.p ] , Brigitte Trousse [ co-resp. ] .

Inria Contract Reference: 2 03 A2005 00 00MP5 01 1

MobiVIP, Individual Public Vehicles for Mobility in town centers, is a research project of Predit 3 (Integration of the Communication and Information systems Group). It involves five research laboratories and seven small business companies (SME), in order to experiment, show and evaluate the impact of the NTIC on a new service for mobility in town centers. This service is made up of small urban vehicles completing existing public transport. The MobiVIP project will develop key technological bricks for the integrated deployment of mobility services in urban environment. The strengths of the project are:

1. the integration between assisted and automatic control, telecommunications, transport modeling, evaluation of service;

2. the demonstrations on 5 complementary experimental sites;

3. the evaluation of possible technology transfer. http://www-sop.inria.fr/mobivip/ .

In December 2004, we finalized in collaboration with B. Senach (Ergomatic Consultants) the deliverable 5.1 [108] which we coordinate with Georges Gallais (Visa Action, Inria Sophia Antipolis). This deliverable aimed at defining a common generic evaluation scenario and proposed a framework to facilitate the identification of the main evaluation dimensions for each planned test or experimentation. The MobiVIP Project has ended in June 2006 for most of the partners but has been extended up to June 2007 for AxIS and some others partners. This continuation will allow us to conduct a large scale experimentation which will start in January 2007 in Antibes.

This year we had two main tasks: 1) the preparation of the deliverables 5.2 [69] , 5.3 [71] and 2) the work related to the task 5.4 and the VU Log experimentation (cf. the task 5.5).

• The deliverable 5.2 [69] addresses the definition of evaluation criteria. From a questionnaire distributed to MobiVIP' partners, 7 evaluation dimensions where identified and about 150 criteria were specified.

• The deliverable 5.3 [71] addresses the use of an information web site for trip planning. AxIS used a specific methodology to conduct this analysis (cf section  6.4.7 ): a human-factors evaluation was realized in parallel to a log analysis. Some results [71] , [70] of the ergonomic analysis suggested to give a deeper look at the logs: suspected difficulties were confirmed by the log' analysis. This combined study of a site' use provided a better insight of the man-machine interface weaknesses and allowed powerful recommendations to improve site quality and user interactions.

  Figure 10. Cartography in CASA

• The task 5.4 addresses the use of data mining technologies to improve travelers' information. Two studies aiming at path classification have been conducted. A first study aiming at defining a semantic distance between paths was conducted in 2005 with simulated data. About thirty routes across a town have been defined. These routes, viewed as potential travels for different professionnal profils (doctors, insurance agent) have been classified with our 2-3AHC algorithm using several similarity distance (Jaccard, Dice) and taking in account attributes such as the number of crossroads, of common parts, proximity of origin. A state of the art on this topic and a review about spatial cognition and mobility, still in progress, will complement the analysis of travelers'behaviours.

  The second one was realized with true data coming from a large survey conducted in railway stations. In the Figure 10 , depicted sections represent the traffic of railway users between stations and downtowns. A section thickness represents the amount of users following it and a section color codes the cluster in which the section has been clustered. There are 13 different transportation modes. Section structure is defined by the distribution of transportation mode used by people following the section. Clustering algorithm processes the percentage of each transportation mode used in a section as a section attribute. With EM clustering, each cluster can be described by averaging the different attributes of all its members. With section structures, this averaging process leads to a "mean structure" for each cluster that helps characterizing them.

• We are still working on the design of the Antibes experimentation (cf. the task 5.5) which will have two stages. The first stage will consist of a study of clients' expectations about new transportation services. This study will precise a previous survey (a questionnaire designed by Loria) done during a demonstration which took place in Nancy. In the second phase of the Antibes experimentation, 5 electric cars will be available in the town center, and their use will be precisely analysed from recorded users' actions.

Eiffel ``E-tourism and Semantic Web'': a RNTL Project (2006-2009)

Participants : Marie-Aude Aufaure, Zeina Jrad, Cyrille Maurice, Yves Lechevallier [ resp. ] , Bernard Senach, Brigitte Trousse.

Inria Contract Reference: 105D1499 00 21173 01 0

The EIFFEL project related to semantic web and e-Tourism was labelled in 2006 by the RNTL program and started this year. Industrial partners are Mondeca and Antidot (leadership) and academic partners are LIRMM and University of Paris X (Nanterre).

The main goal of the Eiffel project is to provide users with an intelligent and multilingual semantic search engine dedicated to the tourism domain. This solution should allow tourism operators and local territories to highlight their resources; the end users will then use a specialised research tool allowing them to organize their trip on the basis of contextualised, specialised, organised and filtered information. Queries and results will be guided by user profiles extracted from usage analysis. These profiles will facilitate the access to distributed and highly heterogeneous data. In this project, AxIS is in charge of the sub-package SP8 and will define new paradigms dedicated to knowledge searching and visualizing, and will extract and exploit users' models and profiles from web logs. AxIs strongly collaborated with Bénédicte Legrand and Michel Sotto from LIP6.

Industrial Contacts

Some contacts during this year:

• VU Log, startup offering software and services dedicated to urban mobility (located in Antibes). Support for an experimentation with a small electric urban vehicule (speed limited to 45 km/h).

• Morocco Telecom: In collaboration with Casablanca University and ENSAM (Meknes), AxIS proposed to Morocco Telecom a web usage study. The research goal of the WRUM project (Web Redesign by Usage Mining) is to extend an existing redesign methodology based on site usage mining and to build a plateform with reusable tracking components. In the study, we will consider the use of several sites to trace the users' navigations and to gain inshight of requirements to improve the sites.

• SAP, Sophia Antipolis related to data mining and data streams (security and environnement problems). Contact: B. Trousse and F. Masseglia