《对于机器翻译的感想》由会员分享，可在线阅读，更多相关《对于机器翻译的感想（3页珍藏版）》请在金锄头文库上搜索。
1、 Comments on Trends in Machine Translation1.What I have learned from it:Bright side:Overview of the full text, we can easily find that there are mainly five trends in machine translation as the author has mentioned. And for each point, the author all lists several sub-arguments to fully supplement h
2、is idea. So, the structure of the article is clear enough for us to analysis. However, I favor to use my own words to highly generalize these five trends as modernization, popularization, corporatization, commercialization and reformation. ModernizationIncreasing Acceptance of Machine Translation As
3、 translation turns out to be a huge, expensive, time-consuming commodity with the sharp increase in company files, machine translation is more and more adapted to the need of the times. PopularizationDemystification of the Black Art of MTThe technology of MT has removed its mysterious veil as develo
4、pers have now opened what used to be inaccessible MT black boxes, enabling specialists to tune the engines based on their own style guides, terminology, and domains.CorporatizationEnterprise-Enabling Machine TranslationDevelopers are committed to working on multiple layers of APIs to turn machine tr
5、anslation into a callable service on internal networks, the cloud, and the web. The ultimate goal of this approach is to have translation become the core of a multilingual information platform rather than being a point product. CommercializationMarketization of Machine Translation SystemsMT is findi
6、ng its way in front of the mainstream consumer and business user, which can bring the vendors business opportunities. Many constituencies push MT applications, such as speech-to-speech and speech-to-and-from-text technologies. ReformationThe MT Business Model Continues to MorphThe development of mac
7、hine translation technology stimulates a revolutionary change in the industry business model which allows efficient machine translation to complement human translation. In this way, MT speeds up translation availability, most content will never be translated by humans.Dark side:Although we say MT is
8、 inevitable, feasible, evolvableits not always reliable, appreciable or applicable. According to the authors research, such concerns come from three aspects . Quality ConcernMT still suffers from the reputation earned in the earlier Babelfish days and the age-old IT even satires it “Rubbish in , and
9、 rubbish out”. For this case, to combine with post processing by an LSP or by the crowd, MT can substantially increase usability.Source ConcernUnderstanding the impact of badly written content, planners are beginning to address the issue of improving the production of original materials. They pay mo
10、re attention to the structured content and the asset control.Maintenance ConcernMT Developers Say, “The Day of Free MT Is Drawing to a Close”. Is it true? For a few years, machine translation could be had for free through tools such as the Google Translate API. However, Google shut it down in mid-20
11、11 with the complaint that it was overused and abused. Nonetheless, Google subsequently announced a paid version of the API.2. What I am interest in it:Apart from the prospective that I am a would-be translator, I am also a consumer. It means that I am going to “equipment with” MT technology sooner
12、or later, but I will be a beneficiary of its entertainment functions. Technology such as Apples Siri and Google Voice will raise my expectations for speech interfaces. I am really interested in speech-to-speech technologies, though I have never experienced fast-moving MT integrators from my smartpho
13、ne. Maybe one day I could “talk “to a friend coming from Russia or Iran without language barriers.However, I remember my teacher had played a TED video in class, which indicated such technology had been in motion in assembly room. Wow, I have a strong feeling that I will be unemployed someday. If MT
14、 interpreting takes off, who will silently land? Basic machine interpretation technology will provide interpreters-on-demand to anyone who needs one.3. What I still dont understand:1. Page 16. Machine Translation Engines Interface with AnythingAnything refers to? I dont know its operation mode.2. Page 21.Open source enginesWho are the third-party developers? Are the LSPs and customers their partners? How do they corporate with each other?
