503 research outputs found
Optimized MT Online Learning in Computer Assisted Translation
No full textIn this paper we propose a cascading framework for optimizing online learning in
machine translation for computer assisted translation scenario.
With the use of online learning, one introduces several hyper parameters
associated with the learning algorithm.
Number of iterations of online learning can affect the quality of translation as
well.
We discuss these issues and propose a few approaches that can be used to
optimize the hyper parameters and also to find the number of iterations required
for online learning.
We experimentally show that using optimal number of iterations in online
learning proves to be useful and we get consistent improvement against baseline
results
The Uppsala-FBK systems at WMT 2011
No full textThis paper presents our submissions to the shared translation task at WMT 2011. We created two largely independent systems for English-to-French and Haitian Creole-to-English translation to evaluate different features and components from our ongoing research on these language pairs. Key features of our systems include anaphora resolution, hierarchical lexical reordering, data selection for language modelling, linear transduction grammars for word alignment and syntax-based decoding with monolingual dependency information
FBK’s Machine Translation and Speech Translation Systems for the IWSLT 2014 Evaluation Campaign
No full textThis paper describes the systems submitted by FBK for the MT and SLT tracks of IWSLT 2014. We participated in the English-French and German-English machine translation tasks, as well as the English-French speech translation task. We report improvements in our English-French MT systems over last year’s baselines, largely due to improved techniques of combining translation and language models, and using huge language models. For our German-English system, we experimented with a novel domain adaptation technique. For both language pairs we also applied a novel word trigger-based model which shows slight improvements on English-French and German-English systems. Our English-French SLT system utilizes MT-based punctuation insertion, recasing, and ASR-like synthesized MT training data
Topic Adaptation for Machine Translation of E-Commerce Content
No full textWe describe effort to improve machine translation of item titles found in a large e-commerce inventory through topic modeling and adaptation. Item titles are short texts which typically contain brand names that do not have to be translated, and item attributes whose translation often depends on the context. Both issues call for robust methods to integrate context infor- mation in the machine translation process in order to reduce translation ambiguity. We survey both existing topic adaptation approaches and propose novel methods that augment the standard phrase-table models with sparse features and dense features measuring the topic match between each phrase-pair and the input text. We report extensive experiments on the translation of item titles from English into Brazilian Portuguese, and show the impact of topic adaptation both with and without domain adaptation
Online Multi-User Adaptive Statistical Machine Translation
No full textIn this paper we investigate the problem of adapting a machine translation system to the
feedback provided by multiple post-editors. It is well know that translators might have very
different post-editing styles and that this variability hinders the application of online
learning methods, which indeed assume a homogeneous source of adaptation data.
We hence propose multi-task learning to leverage bias information from each single post-editors in order to constrain the
evolution of the SMT system. A new framework for significance testing with sentence level metrics is described which shows that
Multi-Task learning approaches outperforms existing online learning approaches, with
significant gains of 1.24 and 1.88 TER score over
a strong online adaptive baseline, on a test set of post-edits produced by four translators texts
and on a popular benchmark with multiple references, respectively
Book Ends & Odd Books : Publications Refuting Conventional Form from the Banff Centre Library Collection
No full textMathur explains how he "unselected" nearly 200 works for this exhibition of unconventional publications by international artists and authors, recognizing the influence of Ulises Carrion's article "The New Art of Making Books." The author reflects upon the roles of language and poetics, the distinction between book and text, and how politics and power affect the making and reception of these works. 2 bibl. ref
Book review: paper tiger: law, bureaucracy and the developmental state in Himalayan India by Nayanika Mathur
No full textFollowing eighteen months of intensive fieldwork, in Paper Tiger: Law, Bureaucracy and the Developmental State in Himalayan India author Nayanika Mathur details the everyday absurdities of bureaucracy in the Himalayan borderlands, showing the frequent gulf between ‘real life’ and the abstract workings of the law. Elisabetta Iob highly recommends this accessible, witty and vividly written book as an outstanding and essential example of ethnographic research
Evidence building for policy: Tobacco surveillance/surveys and research in India
No full textIndia is at a crucial juncture relating to tobacco control and implementing the recommendations of the WHO FCTC. Tobacco consumption in the country remains alarmingly high in urban as well as rural areas. Smokeless tobacco consumption is very popular among larger masses, including the youth. Cigarette smoking has declined, but bidi use has increased concomitantly. Youth continue to be lured to initiate tobacco consumption through efficient marketing strategies of tobacco companies. The epidemiology of tobacco consumption is markedly varied across the country, with high rates in 15 States. Progress has been made towards tobacco control by the enactment of laws and regulations and the National Tobacco Control Program. Strengthening their implementation and enforcement is the biggest challenge and requires resource inputs. Evidence generation and its translation and utilisation for policy interventions would be useful
FINANCING COMMUNITY FACILITIES: A CASE STUDY OF THE PARKS AND RECREATIONAL GENERAL OBLIGATION BOND MEASURE OF SAN JOSE, CALIFORNIA
No full textThis study of the City of San Jose’s Parks and Recreation General Obligation (GO) Bond Measure seeks to identify the politics-, management-, and planning-related lessons learned by the City as it developed its community facilities using the GO bonds proceeds. The study finds that these lessons include: be conservative in what you promise the residents; be prepared for changes in economic environment by identifying supplementary funding sources should the primary source not yield adequate funds; make sure that the jurisdiction is organizationally capable of handling the increased workload; and prepare detailed project plans prior to the bond issuance.Community Infrastructure and Services; Municipal Bonds; Public Finance
Non-communicable diseases are the leading cause of mortality in rural Birbhum, West Bengal, India: a sex-stratified analysis of verbal autopsies from a prospective cohort, 2012–2017
No full text- …
