Apps & Design

Victor Ortega vor+ at
Sat Jul 1 10:11:59 PDT 2000

Machine translation is, in fact, parallelizable if there is some sort
of "common knowledge" database which processing units can query and
update, and if appropriate references to time and place are maintained
(although at this point we are getting close to the concept of machine
learning).  And in particular, the translation of multiple independent
documents is inherently parallelizable, since it is also possible to
employ a team of people to do such a task adequately.  The finer the
parallelism, the more important the communication between processing
units becomes, but it does not render machine translation


On Sat, 1 Jul 2000, Alan Ward wrote:
> Machine translation is a good example of a non-parallelizable
> task. We (people) are not parallel machines, and don't think
> that way. Instead, our speech is full of cross-references: from
> one part of a sentence to another part, between sentences, 
> between paragraphs, plus all the external references (e.g. cultural)
> you can think of. So to translate a text, you cannot break it up
> into little bits, but must treat it as a whole.
> It is failing to understand this that makes most translation software
> fail miserably.

More information about the Beowulf mailing list