[Elsnet-list] 2-year postdoctoral job at Prompsit Language Engineering
Mikel L. Forcada
mlf at ua.es
Wed May 1 17:19:19 CEST 2013
Abu-MaTran: automatic building of machine translation
Marie Curie IAPP project FP7-PEOPLE-2012-IAPP
24-month recruitment of a postdoctoral researcher
Prompsit is a research-shaped company created in 2006 inside the
Transducens research group at the Department of Software and
Computing Systems (Universitat d'Alacant - Spain). It's a leading
company in the development of machine translation, specially
linguistically-motivated systems such as Apertium rule-based systems
or linguistically-augmented Moses statistical systems. The company
activity in R&D is intense both as an industry-driven activity or by
participation in public national and international R&D programs.
Abu-MaTran (Automatic Building of Machine Translation) is an
IAPP-FP7 project in which the company is currently involved. The
project aims at increasing the hitherto low industrial adoption of
machine translation by identifying crucial cutting-edge research
techniques (automatic acquisition of corpora and linguistic resources,
pivot techniques, linguistically augmented statistical translation and
diagnostic evaluation) and preparing them to be suitable for commercial
Besides Prompsit as a central node of interaction, the project involves
four top research institutions (Dublin City University - project
coordinator, Universitat d'Alacant, University of Zagreb and
Institute for Language and Speech Processing). At Prompsit, the
project will be led by researcher Sergio Ortiz Rojas, responsible for
most of the code of the Apertium MT platform, Prompsit's
linguistically-augmented Moses system, a modular version of the
Bitextor parallel text collector and other natural language
processing tools for information extraction or opinion analysis.
The position involves research, development and participation in
outreach activities to achieve the goals of the Abu-MaTran project as
well as collaboration with all researchers in the project.
Main Duties and Responsibilities
* Investigate, in collaboration with the partners, better techniques to
- monolingual and bilingual general and domain-focused corpora acquisition
- monolingual and bilingual terminology extraction
- automatic induction of transfer rules
- building of pivot or linguistically-augmented machine translation systems
- machine translation automatic evaluation
* Implement the techniques for each of the previous points
* Carry out experiments to evaluate their performance.
* Release the output as free/open-source tools with appropriate
interfaces to use them.
* Write the appropriate documentation for each of the work lines:
technical documentation for developers, academic-oriented (papers,
posters, etc.) publications, and tutorials or manuals for developers.
* Attend project-related conferences and meetings
* Present the results at relevant conferences and scientific meetings
* Review work plan with the collaborators according to project
intermediate milestones and results.
* Get involved and give support to outreach activities (linguistic
olympiads, FreeRBMT workshop)
Applicants should provide evidence in their applications that they meet
the following criteria.
The staff in charge of this recruitment process, will use a range of
selection methods to measure candidates' abilities in these areas
including reviewing your application, seeking references, inviting
shortlisted candidates to be interviewed, and other forms of assessment
action relevant to the post.
Qualifications (compulsory): PhD in Computer Science (or at least 4
years of full-time research experience) and less than 10 years of
full-time research experience.
International procedure (compulsory): the candidate cannot have worked
or lived for more than 12 months within the last 3 years in Spain.
* Natural language processing, particularly in machine translation
* User-level or developer-level experience in Apertium, Moses,
OpenMaTrEx, Bitextor, FBC and FMC, ccLexExtractor and
* Data acquisition (desirable)
* Terminology extraction (desirable)
* Machine learning (desirable)
* MT evaluation (desirable)
* Creation of user interfaces and software releasing/sharing (desirable)
Programming languages: C++, Python, PHP (compulsory). JAVA (desirable).
Multilingual skills: Good level of English (compulsory). Basic knowledge
of Spanish or Catalan (desirable). Knowledge of the South Slavic
Languages targeted in the project use case -- Croatian, Bosnian, Serbian
or Montenegrin and Slovenian (desirable).
Good writing and communication skills: ability to intercommunicate with
people and to communicate results, ideas, etc. (compulsory).
Collaborative working skills: ability to take and delegate
Experience in free/open-source software development: participation in
free/open-source software development projects as user or, better, as
Experience in transfer of knowledge between the industry and the
academy: interaction between industry and academy in previous positions
is highly valued (desirable).
Creativity and flexibility skills: ability to be open to different ideas
or opinions, to analyse and solve problems and to make decisions
Active research skills: ability to follow state-of-the-art research
lines associated with the project, to learn and acquire new skills
relevant to the project, to write scientific works, and to meet
This post is fixed-term and full-time at Prompsit (Elx/Elche, Spain).
The starting date is January 2014 and duration is 24 months. Splits are
Terms and conditions of employment:
Terms and conditions will be according to the stipulations of the IAPP
program. The recruited candidate will have a full-time contract with
full social security coverage subject to the Spanish laws and taxes.
For an Experience Researcher with less than 10 year experience the
stipulated salary will be a €57,154 per year gross salary corresponding
to living allowance and additionally €683/€977 per month for mobility
allowance (depending on family charges).
Closing date:1st June 2013.
For informal enquiries about this job contact us at info at prompsit.com.
 http://www.abumatran.eu/, more info at
 http://bitextor.sourceforge.net/ (see prompsit branch in source code)
More information about the Elsnet-list