PhD positions University Paris-12: XML high-performance
 

Doctoral project and scholarship at the University of Paris-12
ALGORITHMS FOR XML:
LARGE DATA SETS, EXTERNAL MEMORY AND SCALABLE PERFORMANCE

Innovimax
Innovimax SARL. Consulting, Training & XML Development
http://www.innovimax.fr
LACL
Laboratoire d'Algorithmique Complexité et Logique, Université de
Paris-12
http://www.univ-paris12.fr/lacl

Innovimax is a startup company based in Paris and offering XML
software solutions, consulting and training. Innovimax is member of
W3C, ISO and AFNOR and is very active in these organizations on all
questions related to XML standards and document verification. Its
customers are: Éditions Francis Lefebvre, La direction des Journaux
Officiels (government publications, under control of the French Prime
Minister), OECD, PTC, etc. Innovimax is member of the Cap Digital
cluster which federates Paris-region leaders in multimedia
technologies. The company actively participates in international
congresses and meetings.

LACL is the University of Paris-12's Computer Science laboratory. LACL
is ranked category « A » research unit by the French Ministry of
Research. One LACL team conducts research on parallel computing and
hosts PC clusters for its experiments and for the benefit of its
industrial partners. This work has been rewarded by an EADS doctoral
thesis prize, national and regional grants. Their techniques for
scalable performance are applied to numerical simulation, machine
learning, databases and automatic software verification. LACL is
member of the System@tic cluster which federates Paris-region leaders
in software R&D.

To support Innovimax's development we wish to engage in a long-term
research project concerned with algorithms for XML transformations.
The project will follow 3 axes:

1) XML sorting and grouping: to find efficient classes of algorithms
for very large number of files and/or very large files, taking into
account various constraints like infinite files (logs) or bi-
directional file access (FileSystem). Consideration may be given to
approximate sorting/grouping to minimize costs.
2) Query algorithms: to combine various improvement factors on XML
queries depending on complexity and reuse. Combinations of the
following are possible: a priori optimization oblivious to document
structure, document model-based rewriting or optimization, document
transformations to improve specific query types. Efficiency thresholds
and hybrid solutions can be designed.
3) Bi-directional processing: XML files can in principle be read in
both directions and this opens the door to optimal processing on large
data sets with BabyStep/GiantStep techniques. Such work can also be
coupled with document projections i.e. design select-by-need temporary
structures.

We seek a qualified Masters-level C.Sc / C.Eng graduate to conduct
this project under joint supervision, leading to a doctorate at the
University of Paris-12 - soon to merge into the new Paris-Est
University.

CONTRACT: The new student will be hired according to the clauses of a
CIFRE contract and register with the Paris-12 SIMME doctoral school.
Contract and doctoral studies duration of 3 years from January 2008.
Necessary resources will be provided by Innovimax-LACL and results
published in international conferences.

Applications: Applicants should send by November 15, 2007 a detailed
CV, MSc-level degree and any other relevant documents. Knowledge of
French not required but language courses will be made available.
Training and experience related to algorithm design and XML structures
will be an asset.

Mohamed Zergaoui
President of Innovimax
innovimax@gmail.com
Gaétan Hains
Director of LACL
gaetan@hains.org