Post-doctoral position in Discourse Parsing, with a focus on
multilingual systems and/ortransfer learning

LORIA, team SyNaLP (Nancy, France)

One postdoc position (1 year) in Natural Language Processing / Machine
Learning is open in the SyNaLP team (http://synalp.loria.fr/) at LORIA
(http://www.loria.fr/en/). This position will be funded by the LUE
(Lorraine Université d'Excellence) Future Leader program and the LORIA.

Information

- Starting date: as early as possible

- Duration: 1 year 

- Deadline for application: open until filled

- Location: Nancy, France

- Salary: around 2,000 euros per month net income

- Informal inquiries can be sent by email to Chloé Braud
  (chloe.braud@loria.fr). The application requires a brief motivation
  letter and a CV.

Topic: discourse parsing

Documents are not just an arbitrary collection of text spans, but rather
an ordered list of structures forming a discourse. Discourse structures
describe the organization of documents in terms of discourse or
rhetorical relations (e.g. Explanation, Contrast ...)  linking the
semantic content of the sentences and clauses. Discourse parsing is an
integral part of understanding information flow and argumentative
structure in documents. Building discourse parsers is currently a major
challenge in Natural Language Processing, but it's an hard task, that
involves several complex and interacting factors, touching upon all
layers of linguistic analysis, from syntax, semantics up to pragmatics.

Previous work have focused on investigating lexico-syntactic features,
with most of the experiments on newswire data in English. The goal of
this postdoc is to investigate novel architectures for discourse parsing
that take into account different layers of analysis (e.g.  syntax,
co-reference, modality, etc) and/or are able to deal with multiple
languages and domains. Another path of research would be to study the
effect of discourse information for downstream applications, such as
question-answering, summarization, sentiment analysis, etc.  The aim is
to explore frameworks such as multi-task learning, to experiment with
out-of-domain and multilingual data, and to investigate representation
learning for clauses, sentences and documents.

Requirements/qualifications

The candidate will have to implement or modify the implementation of
existing discourse or syntactic parsers based on neural networks. S.He
will experiment with several corpora, investigating the effect of
different representations and the robustness of the chosen approach.
S.He will publish in peer-reviewed journals and conferences (ACL, EMNLP,
COLING,...).

The candidate is expected to have:

- a Ph.D. or equivalent in Computer Science, Computational
  Linguistics/NLP, or related fields.
  
- Good programming skills,

- knowledge of current neural network models, and some libraries for
  neural networks (e.g. Tensorflow, Keras, PyTorch, Dynet etc.),

- experience in (discourse or syntactic) parsing is a plus,

- Fluent English. Knowledge of French is NOT a requirement.

Supervision of students is possible, if wanted.

Nancy

To learn more about living in Nancy:
https://www.nancy.fr/nancy-in-english/discover/living-in-nancy-1218.html

Contact

Chloé Braud, LORIA
Équipe SyNaLP
Email: chloe.braud@loria.fr