Human Translation Team Information

Team 	   Team Composition (*)		Software	Translation Procedure
ID	Translator  Proofer/Editor	Used?

E01	Ch-dom,15	Ch-dom,13	No	Translator does initial translation, editor
						does qc; disagreement resolved between the two

E02	Ch-dom,3	Ch-dom,10	No	Draft translation by translator, followed by
						editor's proofreading

E03	Ch-dom,10	[same]  	No	Translation by translator alone followed by
						carefully checking dictionaries and searching
						the Internet
E04	Ch-dom,7 	Ch-dom,4	No	Initial translation by translator with frequent
						reference to dictionaries and search engines,
						editor's proofreading

(*) If there is only one individual listed, it means the translator
    does both translation and proofreading/editing.  "Ch-dom" means
    Chinese-dominant bilingual, and "En-dom" means English dominant
    bilingual.

			Inside US	Outside US
Translation Companies	   2		    2
Freelance Translators	   0		    0


Machine Translation System Information

Sys.ID	System Information 	Translation Procedure

E05	Web-based free service	The simple procedure of copying, pasting and saving
				was used.  Each segment was copied into the source box 
				for translation and the result was copied back to 
				replace the source segment. Words not translated could 
				not display properly in the output box, which seems to 
				indicate Unicode encoding used by the system.  Some 
				gibberish symbols remain in the output. Each file was
				saved as ANSI.

E06	Commercial software	Built-in drag-and-drop file translation option was
				used.  Unrecognized Chinese words are preserved in the
				output.  The output is in Unicode. For conversion to ANSI
				code, each output file was opened in Notepad and saved
				as ANSI.  Some files had gibberish symbols in the Unicode
				output and did not make sense either in the ANSI
				version.

E07	Commercial software	Built-in batch translation was used.  Unrecognized
				Chinese words are preserved in the output.  Thus output
				is a mix of plain ASCII and GB.

E08	Commercial software	A newer version of E07, which now can handle sgml 
				documents. Same procedure, but on sgml-formattede source.
				Unrecognized Chinese words are preserved in the output.  
				Thus output is a mix of plain ASCII and GB.

E09	Web-based free service	Web transltion option was used.  The source files
				were placed temporarily on a web server and the link
				to each invididual file was given to the system for
				translation.  The output was then copied to a
				Notepad window and saved using the original file
				name.  The output did not have line breaks - they
				were manually inserted in correspondence with the
				source.

E10	Commercial software	To preserve the original like breaks - which it doesn't
				in normal file translation, each source file was opened
				in a web browser.  The output was then copied into a
				Notepad buffer and then saved as a text file in ANSI.
				The output encoding is also in Unicode and some output
				files also have gibberish symbols.  For almost every
				sentence translated, the system provided an alternative
				translation, which was always removed in the final
				output.

E14	Research System		Result submitted for TIDES MT 2002 Evaluation.
                                Large Track, Chinese, i.e. only data available at LDC were
                                used for training.