moteur de recherche

Séminaire Algo - Sarah Cohen Boulakia
Séminaire Algo - Sarah Cohen Boulakia
17-mars-2015 14:30
Il y a: 3 yrs

Sarah Cohen Boulakia

Managing Provenance information in Scientific Workflows

Salle de séminaire (4B05R) - Bâtiment Copernic

Scientific workflows management systems are increasingly used to specify and manage bioinformatics experiments. To enhance reproducibility, recording the provenance information (keeping track of all the input data used and parameters settings of tools) in each workflow execution is of paramount importance. Scientific workflows represent experiments by means of graph structures where nodes are the bioinformatics tasks to be executed and edges represent the dataflow. Managing provenance information in scientific workflows is then directly related to querying, comparing and transforming graph structures. In this talk, I will first present an approach which deals with provenance overwhelming, and proposes abstraction mechanisms to help users focus on the most relevant information. Then, I will introduce a technique we developed to help users understand the differences between two executions of the same workflow. Finally, I will present solutions we have designed to address the problem of refactoring scientific workflows.

<- retour: