30 July: fyi -- computational linguistics internship, Grenoble (Xerox)

Index of July 2015 | Index of year: 2015 | Full index



Xerox Research Centre Europe, located in Grenoble, is offering the
following internship for Fall 2015:



---------------------------------------------------------------------------------------------

Weighted tree-based transducers for natural language generation in the
context of a dialogue system

---------------------------------------------------------------------------------------------

See:
http://www.xrce.xerox.com/About-XRCE/Internships/Weighted-tree-based-transducers-for-natural-language-generation-in-the-context-of-a-dialogue-system


Contact: Marc Dymetman
marc.dymetman@xrce.xerox.com

Duration: 4 to 5 months
Start Date: As soon as possible after September 2015

Description:

In the context of a human-machine dialogue system being developed at
XRCE, the internship will have aim of exploring the applicability of
weighted/probabilistic finite-state tree transducers (and related
formalisms such as tree-to-string transducers) to the problem of
generating textual utterances from semantic representations produced
by a DM (dialogue manager).

The advantages of such transducers are that they support operations
such as composition and intersection (in particular with language
models), and are well-suited for reversible processing (parsing as the
reverse of generation).

The internship will build on work done at XRCE with string-based
transducers, which it will attempt to adapt and extend towards
tree-based machines, either based on external toolkits (such as
Tiburon [May and Knight 2006]) or on specially developed software.

The ideal candidate should be a Computer Science student at the Master
or (preferably) PhD level, with some experience of weighted tree-based
transducers (or at least string-based transducers), as well as of
associated software tools. Strong programming skills are a requirement
(Python, C++, Java...). Conference publication of the work will be
strongly encouraged.

Please note that applicants must be registered students at a
university or other academic institution and that this establishment
will need to sign an 'Internship Convention' with Xerox before the
student is accepted.

If you would like to submit your application, please send it to
marc.dymetman@xrce.xerox.com or
to
xrce-candidates@xrce.xerox.com. Please
mention "Tree based transducers" in your subject line.

Index of July 2015 | Index of year: 2015 | Full index