Apps & Design

Gregory R. Warnes warnes at biostat.washington.edu
Sat Jul 1 09:07:18 PDT 2000


Careful.  You are equating "the problem consists of independent pieces"
with parallelizable. 

Stating that natural language processing is inherently non-parallel
ignores the fact that the only system known to correctly process natural
language (the brain) is made up of a lot of independent processing units!

The current tools may not correctly approach the problem correctly, but
that does not mean it isn't possible. 

-Greg

On Sat, 1 Jul 2000, Alan Ward wrote:

  AW>> Machine translation is a good example of a non-parallelizable
  AW>> task. We (people) are not parallel machines, and don't think
  AW>> that way. Instead, our speech is full of cross-references: from
  AW>> one part of a sentence to another part, between sentences, 
  AW>> between paragraphs, plus all the external references (e.g. cultural)
  AW>> you can think of. So to translate a text, you cannot break it up
  AW>> into little bits, but must treat it as a whole.
  AW>> 
  AW>> It is failing to understand this that makes most translation software
  AW>> fail miserably.
  AW>> 
  AW>> Best regards,
  AW>> Alan Ward
  AW>> 
  AW>> 
  AW>> > ----------
  AW>> > De: Kragen Sitaker <kragen at pobox.com>
  AW>> > A: beowulf at beowulf.org
  AW>> > Asunto: Re: Apps & Design
  AW>> > Fecha: divendres, 30 / juny / 2000 20:29
  AW>> 
  AW>> > "Joel" asks:
  AW>> > > 1.  What kind of work has been done in applying Beowulf to machine 
  AW>> > > translation?  Would parallelism help when trying to translate several
  AW>> texts 
  AW>> > > into several different languages in the shortest possible time?  Can 
  AW>> > > existing web-based translation systems (Systran, InterTran, &c) be 
  AW>> > > parallelised?
  AW>> 
  AW>> > Unfortunately, I don't know anything about machine translation, so I
  AW>> > can't answer the first question; but as for the second question,
  AW>> > translating multiple documents is an obviously parallelizable task, as
  AW>> > the results of the translations are independent.  From looking at
  AW>> > SYSTRAN output, it sort of looks like results of translations of
  AW>> > invidual phrases in a document are pretty independent when they're
  AW>> > further than a sentence or two apart.
  AW>> 
  AW>> 
  AW>> 
  AW>> _______________________________________________
  AW>> Beowulf mailing list
  AW>> Beowulf at beowulf.org
  AW>> http://www.beowulf.org/mailman/listinfo/beowulf
  AW>> 





More information about the Beowulf mailing list