英语论文网

quality of MT output has not dramatically improved in recent years. This paper explains how factors increasing the utility and applicability of MT have opened the door for market adoption. These factors include: Integration with translation memory, workflow automation, and other technologies resulting in improved output quality. Digitization of enterprise content and the trend towards standard file formats, such as RTF and XML. Increasing tolerance for low-quality text due to the prevalence of email, chat, and other informal, text-based communication.

A general misconception persists that MT is intended to replace human translators. MT output is not equivalent to human translation. But it does add a range of options for saving time and cost, and improving service to customers and employees. Understanding where MT is useful, and why, will help ensure successful deployments. There are three reasons to use MT: To save cost when local language content is required but the expense of human translation is prohibitive. To save time when local language content is required but the delay for human translation is unworkable. To improve customer service when human translation cannot meet the demands of content volume or rate of change.

A number of successes have now been reported by enterprises applying Main commercial operations. By looking at these new, winning applications font, clear trends can be established. Based on lessons learned in these environments, this paper outlines specific criteria to use when evaluating enterprise initiatives for MT. Criteria for success include: Volume, turnaround and quality requirements. Predictability of subject matter and consistency of writing style. Availability of translation memory databases and language professionals to customize terminology on an ongoing basis.

SECTION 1
The utility and applicability of MT
Commercial interest in MT is experiencing a significant rebound, eventhough the quality of machine translation (MT) output has notdramatically improved in recent years. A number of non-linguisticfactors have contributed to elevating the utility of MT, and heightenedits applicability in enterprise environments. These factors have openedthe way for broader adoption of this important technology over thenext 3-5 years, despite the fact that translation quality is not likely tosee large gains in the near future.

Translation quality
The technologies of most MT applications have their roots in the 1960sand ‘70s. Developing each system required many millions of dollars,largely funded by government agencies. Known as “rule based” MT,these systems process translation requests using linguistic rules — rulesthat are necessarily complex and inter-reliant. Improving the outputquality of these established systems is tedious and expensive.

To date, newer approaches utilizing statistics and examples of existingtranslations (Example Based Machine Translation, or EBMT) have yet toprovide better results. MT vendors are currently developing hybridapproaches using both linguistic and statistical methods, which maydeliver a higher translation quality in the future.