Translating and the Computer 35: 28-29 November 2013

Abstracts

Streamlining your workflow: useful desktop software and mobile

applications for the interpreting and translation industry

V AIiperta

Recent developments in mobile, online and software technologies offer us a range of
solutions to increase productivity, streamlining processes and admin work and also
help interpreters be more efficient and fast thinking in the booth. This presentation aims
to cover some of the best examples of mobile/desktop applications and software for
translation but also booth environment:

- Administration / folder management

- Payments and invoicing

- Communication / follow-up

- Storage, syncing and backup

- Sharing tools / terminology search

- Productivity Despite no CAT tools are currently compatible with mobile devices, plenty
are the new applications that help keep everything under control: seriously well-
structured to-do list systems that sync on all platforms, financial and invoicing
management apps, word-processing and publishing solutions, sharing tools such as
Dropbox, PDF makers, mobile scanners, speech recognition software, audio notes,
Skype or E-fax, As a user of both Mac and Windows-based systems, the speaker will
touch on the apps and software she trusts as a translator. For conference / ad-hoc
interpreting, she relies on a range of apps and online / offline dictionaries, available on
tablet that help minimise the impact of paper-based materials in the booth and
business environments, saving space and time.

Machine Translation Quality Evaluation and Estimation

Lucia Specia

This talk will provide an overview of the most popular approaches to assess machine
translation quality, from manual to automatic metrics, at different levels of granularity,
from word- to document-level. It will focus on recent developments in metrics for
quality prediction, that is, reference-free metrics, and their application to machine
translation system selection, estimation of post-editing effort, and gisting.

New tools for subtitle translation

Yota Georgakopoulou and Lindsay Bywood

Subtitling plays a key role in ensuring audiovisual content is accessible across
languages, With content volumes rising, relying on human translation alone is not a
sustainable model. New technology is needed to enable the industry to grow along
with the market. The SUMAT project aims to explore the integration of Statistical
Machine Translation (SMT) into the subtitle translation workflow and thus provide a tool
that can support the translation processes of subtitling companies and freelance
subtitlers.

We are developing an online subtitle translation service catering for 9 European
languages combined in 14 language pairs, Our baseline SMT systems have already
been built using high-quality parallel subtitle corpora produced by the subtitling
companies of the consortium, A user-based evaluation is currently under way focusing
both on the quality of the MT output as well as the productivity gain achieved through
the use of the system, The final SUMAT systems are expected to provide up to 25%
increase in productivity for subtitle translation and will go live in 2014,

Overview of the XLIFF 2.0 specification

Fredrik Estreen

The presentation will cover the upcoming XLIFF 2,0 standard, the changes since 1.2
and the work done to foster greater interoperability. Starting with a brief discussion on
the successes and shortcomings of the current version of the standard, each of the
major changes will be covered together with an overview of the new standard. The new
modular design of the format is a key difference compared to the existing monolithic
specification. A streamlined core augmented by optional modules catering for process
specific needs is intended to simplify the creation of processing agents. Another key
difference is the inclusion of processing requirements covering the transformation of
XLIFF documents so that the specification covers more than static document
instances, The specification of allowed transformations is intended to promote better
interoperability and allow new processing models to be more easily implemented.
Several new features have been added as modules to provide exchange of common
information that was not covered in the previous versions of the standard. An overview
of all the modules is included in the presentation.