Capitalisation et partage de connaissances d'analyse de traces numériques d'activités : Assister le suivi de l'activité dans les environnements de formation à base de simulateur pleine échelle

Abstract : Our research takes place in the field of knowledge engineering. In particularly we focus our study in capitalizing and sharing knowledge of observation and analysis of digital traces. In this context, we base our approach on the concept of modeled trace (M-Trace) developed by the SILEX team. Our approach give the possibility to exploit low levels digital traces in order to extract higher knowledge level through rule-based transformations. These rules modelize the knowldege of observation and analysis of different users. Rules can be capitalized and shared between users. We complete our proposal by providing a synthetic visualization of the knowledge levels with observed elements from the activity. By means of a generic trace model, that we have specified, users can explore the different abstraction level in purposes of investigation in order to better understand and analyze the activity. Our proposals have been implemented in a prototype, called D3KODE (« Define, Discover, and Disseminate Knowledge from Observation to Develop Expertise »), allowing the processing, representation and visualization of traces. D3KODE was applied in the context of professional training on the nuclear power plant full-scope simulator of the EDF group designed to maintain and enhance the knowledge and skills of Nuclear Power Plant control room staff. In such context, the observation, analysis and debriefing of individual and collective interactions of trainees’ operators is a dense activity that require attention and constant alertness of the trainers throughout the simulation, especially for the young trainers who do not have the expertise of confirmed trainers. The amount of data collected during a simulation is big and very low levels. They are difficult to analyse manually in order to extract high level information reflecting the behaviour of trainees. In such a context, understand and follow the activity requires a strong expertise that all trainers don't have. So as to validate our approach, D3KODE was evaluated in a real context according to a comparative protocol conducted with a team of trainers from EDF Group. The evaluation gave significant results to validate our approach and encourage many research opportunities. These research, conducted in partnership with the Training Unit (UFPI) of EDF Group, have been applied in the context of training and skills maintaining of the nuclear power plant control room staff.
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-01099275
Contributor : Olivier Champalle <>
Submitted on : Friday, January 2, 2015 - 12:45:04 PM
Last modification on : Tuesday, January 29, 2019 - 5:16:09 PM
Long-term archiving on : Friday, April 3, 2015 - 10:11:03 AM

Identifiers

  • HAL Id : tel-01099275, version 1

Collections

Citation

Olivier Champalle. Capitalisation et partage de connaissances d'analyse de traces numériques d'activités : Assister le suivi de l'activité dans les environnements de formation à base de simulateur pleine échelle. Informatique [cs]. Universite Claude Bernard Lyon 1, 2014. Français. ⟨NNT : LYON110326029⟩. ⟨tel-01099275v1⟩

Share

Metrics

Record views

237

Files downloads

519