Wilhelm Weisweber
Open Distributed Machine Translation


This page was translated from German into English with the help of the SYSTRAN MT system. The text has been post-edited by the author. The original result of SYSTRAN can be seen here.

Open Distributed Machine Translation (ODMT) is a combination of concepts of Open Distributed Processing (ODP) and Natural Language Processing. ODMT is understood both as translation service and as working environment for human translators. The special quality consists of the fact that a variety of text types can be potentially processed by the integration of different MT systems and their components.

In ODMT different new and existing MT modules (analysis, transfer, generation, disambiguation), data (lexicons, grammars, mapping rules, background knowledge) and tools (MT systems, terminology data bases, translation memory, alignment) can be integrated with assistance of a MT trading component (meta trader) with attached knowledge representation (translator and system knowledge), which is based on the ODP trader. For that both research and commercial systems and components come into consideration. The latters can be encapsulated if necessary.

The distribution of MT systems and components accounts for the distributed language and domain knowledge competence (translation, linguistics and computer science), which can be adequately used. The openness of an ODMT system leads to an evolutionary development (synergy). Modules and linguistic data of existing MT systems or components are re-usable. An ODMT system can be used as test environment for other components for Natural Language Processing. The use of the design method RM-ODP (ISO standard) can contribute to the development of standards in MT. Beyond that ODMT offers further advantages such as expandability, reliability, robustness and concurrency.


[last change Wed Jun 18 1997 by ww@cs.tu berlin.de]