SHERPA project


The research results from the Sherpa project led to their implementation in several software packages which have been applied to various domains.

Hazard management

Elsa: avalanche path analysis

Elsa has been developed jointly with CEMAGREF-Grenoble on top of Shirka. CEMAGREF is responsible for protection against snow avalanche hazard (snow avalanches can destroy buildings, roads, ski resort devices and kill people). The Elsa system is dedicated to support specialists in their avalanche path analysis representing the mountain terrain and simulating meteorological scenarii in order to evaluate the avalanche hazard. The raw terrain is acquired through contour lines, ridges, thalwegs and vegetation areas. It is represented by Shirka objects and presented to the user through a graphic map. Then the user can build a spatial analysis of the path in "panels". A small panel is a connex site area which is homogeneous according to slope and vegetation and is not crossed by a thalweg nor a ridge. It is also of reasonable size. This is the unit which is taken into account in the specialist analysis process. These panels are also represented in Shirka.

The user can then build or select meteorological scenarii which are plausible for the area. These scenarii are simulated on the path. The simulation includes the four phenomena involved in avalanche formation in the starting zone: snow fall, snow drift, fracture, rupture propagation and flowing. Snow fall determines the average snow precipitation on the path; snow drift leads to the accumulation of snow on particular panels depending of the wind orientation and strength; a fracture appears on a particular panel (called trigger) whose snow cover is too unsteady; rupture propagation determines the panels released by the trigger panel. This both quantitative and qualitative simulation provides then the mass and velocity of the avalanche on the flowing zone.

Shirka does not only support the basic storage device for the path and scenarii representation, it also offers inference mechanisms for automatically computing various parameters such as snow height on a panel, rupture limit for a panel, panel average elevation, the amount of snow... It thus constitutes the basic layer of Elsa qualitative simulation. It also provide a reasoning maintenance (TMS) facility which stores the computation results and ensures their validity.

Signal processing

Said: task representation and composition for offshore structure diagnosis

Said is an application of Scarp, developed jointly with the IFREMER institute. It is a knowledge based system for vibration analysis applying signal processing techniques. It is used in order to support the diagnosis of vibrating structures, such as offshore structures. This kind of diagnosis is complex and necessitates frequent user intervention into the problem solving process: the input parameters of the methods used are delicate to estimate, a trial and error process is necessary, relying on the user interpretation of graphic representations.

Medical diagnosis



ColiGene: an object-oriented knowledge base on the regulation mechanssms of gene expression in Escherichia coli

ColiGene is a knowledge base developed with Shirka, designed jointly with a research laboratory of the Lyon university (laboratoire de Biométrie, Génétique et Biologie des Populations). It describes the regulation mechanisms within the genome of Escherichia coli. The related knowledge is on one hand formalized via the definition of object classes, such as gene, protein etc. Textual definitions of these objects or other information concerning the regulation mechanisms are on the other hand integrated via an associated hypertext network. This allows to integrate free text or pictures, taken for example out of the literature, into the knowledge base.



The firstfly base is the outcome of a common project between LGPD and INRIA Rhône-Alpes funded by an "Action coordonnée concertée: sciences du vivant (13: informatique et génome)" of the french ministry of education and research (MENESR).

The Firstfly knowledge base has been developped by Christophe Chemla under the supervision of Bernard Jacq at LGPD and the advices of Jérôme Euzenat.

Genetic expression does not only consist in translation and transcription of genes into proteins because the presence or lack of particular proteins can enable or disable the traduction. These proteins, in turn, are produced by other genes. This phenomenon (named interaction) has been studied for long in the context of D. melanogaster.

The firstfly knowledge base is devoted to the reprensentation of genetic interaction in the D. Melanogaster during its early embryo development (it is restricted to the syncytial blastoderm and cellular blastoderm development stages on the wild animal).

It contains the description of the genes, proteins, interactions that have been observed. . The base also refers to the notions of expression context and network of interaction. It is already possible to navigate through the network or to get the set of possible paths from on product to another. The project is also planned to:

The knowledge base is browsable and usable through the WWW, but it has not been released yet for public consultation.

More information: Christophe Chemla, Une approche bio-informatique des réseaux d'interactions géniques impliqués dans le développement embryonnaire de Drosophila melanogaster, DEA de biologie, Universit é de la méditéranée, Marseille (FR), 1996

Mis-à-jour le 27/01/95