Integrating SDL Trados 2007 and PROMT LSP 9.5 into a translation workflow
Oleg Vigodsky, Head of Argonaut Translation Agency
A detailed description of integrating SDL Trados 2007 and PROMT LSP 9.5 software tools into a translation workflow is provided.
Integrating SDL Trados 2007 and PROMT LSP 9.5 into a translation workflow
Oleg Vigodsky, Head of Argonaut Translation Agency
A detailed description of integrating SDL Trados Studion 2011 and PROMT LSP 9.5 software tools into a translation workflow is provided.
Machine Translation as a translator's tool
Oleg Vigodsky, Head of Argonaut Translation Agency
This presentation covers following issues:
Julia Epiphantseva, Head of Business Development
USA, Seattle, TAUS User Conference 2012
15-16.10.2012
Julia Epiphantseva talks about where PROMT finds data, how they train an MT engine and how they evaluate translation output in the case of Machine Translation User generated content.
PROMT DeepHybrid System for WMT12 Shared Translation Task
Alexander Molchanov, Head of Statistical Research Team
Montreal, Quebec, Canada, Seventh Workshop on Statistical Machine Translation
7-8.06.2012
This paper describes the PROMT submission for the WMT12 shared translation task. We participated in two language pairs: English-French and English-Spanish. The translations were made using the PROMT DeepHybrid engine, which is the first hybrid version of the PROMT system. We report on improvements over our baseline RBMT output both in terms of automatic evaluation metrics and linguistic analysis.
Flexible and efficient management of translation quality
Julia Epiphantseva, Head of Linguistics Research Team
Berlin, Translingual Europe 2010
07.06.2010
One of advantages of our systems is the high level of adaptibility to the different requirements of a customer. It’s because of flexibility of translation pipeline and a variety of tools which help to adjust the translation process for the specific domain and to get high-quality translation. Depending on necessity to process documents with different structure the different format translators and text preprocessors can be used. For example, the preprocessors can be used when the text contains some symbol sequences which should not to be translated or should be translated in some special way, like e-mail addresses, special escape sequences for selecting of the fragments of text.
A Brief Guide to PROMT Machine Translation Technology
This white paper focuses on PROMT machine translation technology. It provides a background of machine translation (MT), the advantages of MT technology used by PROMT, functionality, types of translation demands, and PROMT customization tools.
"MT systems work with natural language - a data set that is infinitely variable, ambiguous, and structurally complex. To translate adequately, an MT system must encode knowledge of hundreds of syntactic patterns, variations, and exceptions, as well as relationships among these patterns. Machine translation software should be equipped with ever-changing vocabulary and specific semantic knowledge about the usage patterns of tens of thousands of words. The system must ensure the accurate identification of parts of speech and the grammatical characteristics of words which may, in different contexts, be nouns, verbs, or adjectives, each having many possible translations. Translation also requires a vast store of knowledge about the world, the intent of the communication, and the subject matter.”
"How the Computer Translates"
Svetlana Sokolova
President of PROMT
Ph.D Computer Science
Machine translation is not as simple as it seems at first glance. It is not just a matter of importing large dictionaries into the machine translation engine. The article explains "How the Computer Translates," how the machine translation engine stores rules, recognizes common collocations, and how the engine can be configured to yield optimal results.
"Machine Translation for Cross-Language Social Media"
Jordi Carrera, Olga Beregovaya, Alex Yanishevsky, 2008
User-generated content available in weblogs and social media a) contains high noise levels, b) is domain independent, c) is generated fast, d) is available in large quantities and d) is inherently focused on information content and knowledge sharing. Thanks to the new Internet culture, which emphasizes accessibility, openness and active participation, communication needs are less stringent but require faster response and must preserve information content. These properties make user-generated content suitable for machine translation and, more specifically, hybrid machine translation, which combines knowledge representation with statistical modeling. In this article we present a qualitative study of data extracted from the Social Media Dataset: we analyze how naturally occurring phenomena can affect machine translation quality and we show how new hybrid approaches may successfully preserve semantics while at the same time achieving near-optimal levels of linguistic fluency.