Composition guidée de services : application aux workflows d’analyse de données en bio-informatique

Abstract : In scientific domains, particularly in bioinformatics, elementary services are composed as workflows to perform complex data analysis experiments. Due to the heterogeneity of resources, the composition of services is a difficult task. Users, when composing workflows, lack assistance to find and interconnect compatible services. Existing solutions use special services manually defined to manage data format conversions between the inputs and outputs of services in workflows, it is difficult for an end user. Managing service incompatibilities with manual converters is time-consuming and heavy. There are automated solutions to facilitate composing workflows but they are generally limited in the guidance and the data adaptation between services they offer. The first contribution of this thesis proposes to systematically detect convertibility from outputs to inputs of services. Convertibility detection relies on a rule system based on an abstraction of input and output types of services. Type abstraction enables to consider the nature and the composition of input and output data. Rules enable decomposition and composition as well as specialization and generalization of types. They also enable to generate data converters to use between services in workflows. The second contribution proposes an interactive approach that enables to guide users to compose workflows by providing suggestions of compatible services and links based on convertibility of input and output types of services. The approach is based on the framework of Logical Information Systems (LIS) that enables safe and guided requests and navigation on data represented with a uniform logic. With our approach, composition of workflows is safe and complete w.r.t. desired properties. The results and experiences, conducted on bioinformatics services and datatypes, show the relevance of our approaches. Our approaches offer adapted mechanisms to manage service incompatibilities in workflows, by taking into account the composite structure of inputs and outputs data. They enable to guide, step by step, users to define well-formed workflows through relevant suggestions.
Document type :
Theses
Complete list of metadatas

Cited literature [15 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-01301681
Contributor : Abes Star <>
Submitted on : Tuesday, April 12, 2016 - 4:27:07 PM
Last modification on : Friday, January 11, 2019 - 4:23:48 PM
Long-term archiving on : Wednesday, July 13, 2016 - 12:01:25 PM

File

These_BA_Mouhamadou.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-01301681, version 1

Citation

Mouhamadou Ba. Composition guidée de services : application aux workflows d’analyse de données en bio-informatique. Bio-informatique [q-bio.QM]. INSA de Rennes, 2015. Français. ⟨NNT : 2015ISAR0024⟩. ⟨tel-01301681⟩

Share

Metrics

Record views

442

Files downloads

120