Contribution à la visualisation des connaissances par des graphes dans une mémoire d'entreprise : application sur le serveur Atanor

Abstract : The smooth development of a knowledge management process is based on the use of efficient visualization methods that enable the user to easily understand the knowledge models which are used. Experience feedback of the knowledge management system Atanor, which is knowledge-deployment-oriented in an operational context for complex systems, showed that the tree model currently used for the visualization of the knowledge models is not intuitive. Too many vertices redundancies can lead to reading difficulties and hide some important characteristics. To solve these problems, we propose the Graph'Atanor model which is based on hierarchical graphs.
Moving to a graph-based model raises the issue of its visual representation. The drawings have to remain readable and understandable by all users. They have to satisfy various criteria such as aesthetic ones which model a combinatorial optimization problem consisting, for each layer, in finding an optimal order of the vertices. To solve this problem, we develop a new hybridized genetic algorithm which follows the basic scheme of a genetic algorithm with two major differences: the use of two problem-based crossovers and a hybridization resulting from a local-search strategy.
Computational experiments for standard size graphs show that the genetic algorithm gives better results than the other methods we know. The comparison of the two knowledge models on an industrial example shows that Graph'Atanor based models are not only easier to read but also allow to follow the trace of the users and show critical vertices.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00335934
Contributor : Bruno Pinaud <>
Submitted on : Friday, October 31, 2008 - 11:21:42 AM
Last modification on : Thursday, April 5, 2018 - 10:36:25 AM
Long-term archiving on : Monday, June 7, 2010 - 8:52:57 PM

Identifiers

  • HAL Id : tel-00335934, version 1

Collections

Citation

Bruno Pinaud. Contribution à la visualisation des connaissances par des graphes dans une mémoire d'entreprise : application sur le serveur Atanor. Autre [cs.OH]. Université de Nantes, 2006. Français. ⟨tel-00335934⟩

Share

Metrics

Record views

373

Files downloads

353