The Lattice research unit (http://www.lattice.cnrs.fr/?lang=en <http://www.lattice.cnrs.fr/?lang=en>) in Paris offers a 3 year PhD studentship in the domain of natural language processing. The PhD is related to the ERA-NET Atlantis project (https://ai.vub.ac.be/atlantis <https://ai.vub.ac.be/atlantis>) aiming at developing algorithms to study the emergence of language in a community of robots. The PhD candidate will be based at Lattice (Paris area, France) but will work closely with the Atlantis partners, especially the Sony research laboratory in Tokyo (esp. with Michael Spranger, https://sites.google.com/site/michaelspranger/ <https://sites.google.com/site/michaelspranger/>).
The goal of the PhD is to study the inference of a grammar from semantic utterances or, conversely to learn how to generate sentences from a semantic representation in a specific robotic environment. The task will be defined more precisely at the beginning of the PhD but is related for example to the Semeval 2016 task 8 "Meaning Representation Parsing" based on AMR (abstract meaning representation) defined as follows: "Abstract Meaning Representation (AMR) is a compact, readable, whole-sentence semantic annotation. Annotation components include entity identification and typing, PropBank semantic roles, individual entities playing multiple roles, entity grounding via wikification, as well as treatments of modality, negation, etc." (http://alt.qcri.org/semeval2016/task8/ <http://alt.qcri.org/semeval2016/task8/>)
In the Atlantis framework, the goal is not to be able to infer any grammar from any English utterances but is limited to the language used by robots in their environment, which makes the task easier to handle. We are also interested in incremental aspects in learning: if new utterances are received by a robot, how can the robot update and maintain its grammar so as to be able to parse the new utterances?
Evaluation will be done on standard datasets and on other datasets provided by Sony. Note that even if the topic is related to robots, it does not require to actually work with robots.
We are looking for a candidate with the following skills: - advanced programming skills - previous experience in natural language processing - a good knowledge of machine learning techniques - fluent in English (French would be a plus)
If you are interested, please write to Thierry Poibeau (thierry.poibeau at ens.fr <mailto:thierry.poibeau at ens.fr>) as soon as possible, with a CV, a grade statement and the names and email of two academic referees (who may be contacted later on. We don’t need reference letter at this stage). Interviews will be done in July and / or in August. The successful candidate is expected to begin in October 2016. Salary will be between 1500 and 1600 euros net.
Abstract Meaning Representation http://amr.isi.edu/language.html <http://amr.isi.edu/language.html>
Michael Spranger and Luc Steels. Co-Acquisition of Syntax and Semantics - An Investigation in Spatial Language. In IJCAI'15: Proceedings of the 24st international joint conference on Artificial intelligence, pages 1909-1905, Palo Alto, 2015. AAAI Press. https://sites.google.com/site/michaelspranger/publications <https://sites.google.com/site/michaelspranger/publications>
SEMEVAL 2016 task 8 "Meaning Representation Parsing" http://alt.qcri.org/semeval2016/task8/ <http://alt.qcri.org/semeval2016/task8/>
Thierry Poibeau http://www.lattice.cnrs.fr/Thierry-Poibeau <http://www.lattice.cnrs.fr/Thierry-Poibeau> -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: text/html Size: 4453 bytes Desc: not available URL: <https://mailman.uib.no/public/corpora/attachments/20160708/d5beba70/attachment.txt>