From: Barry Caplan (email@example.com)
Date: Fri Sep 27 2002 - 13:22:29 EDT
At 04:26 PM 9/27/2002 +0100, William Overington wrote:
>I had not heard the description "Message catalog" previously, so I can
>search for that too.
>I have previously searched under telegraphic code and language and
>An email correspondent drew my attention to the following list of "numbered
>I have not yet found any example oriented to language translation.
Key Unix libraries have used "message catalogs" as part of the API since time immemorial. Hence any Unix application with even a whiff of a chance of being internationalized is likely to have used those functions.
>not yet found any example oriented to carrying on a complete conversation.
I would look for the earliest references to machine translation int he 1940s and 50s, up to the work with Eliza at MIT in the 60s. I think there is an enormous project whose name I don't recall right now going on in Texas, perhaps Austin, which is spiritually derived from Eliza and focused on sending whole, previous composed sentences back conversational style.
If you want to find the whole of the literature in this area, I suggest searching "Turing Test".
>>A proprietary coding system is a bad idea.
>Well, it depends what one is trying to do. If one wishes to establish a
>system whereby proprietary intellectual property rights exist, then a
>proprietary coding can be a good idea. Various large companies use
>proprietary coding systems for files used with their software packages. If,
>however, one is trying to establish an open system, then you might well be
Or if you want to minimize the amount of reinventing the wheel you do internally. You can easily use a proprietary format outside and XML inside, just as you can use SJIS outside and Unicode for internal processing.
>>Failure to investigate the state of the art, (especially where google is
>>so effortless), means this idea is not pushing any envelope.
>Well, if you have any specific suggestions of what keywords to use in a
>search, that would be very helpful.
I have given you some. Rather than focusing on pseudo-scientific terms like "radiogram", I suggest a starting with a familiarity with the history of computer science, both pure and applied research.
>The keys idea is pushing the envelope.
No it is not.
>As spin off from this discussion,
>maybe the XML people, and the Unicode Technical Committee, will do something
>about having special characters for the XML tags rather than using U+003C
>and thereby help people wanting to place mathematics and software listings
>in the same file as markup. Is using U+003C a legacy from ASCII days?
Why is it not possible to use "<" signs in XML?
>Most of my postings in this thread are in response to people asking me
>specific questions and raising interesting points. That is surely why a
>discussion group exists.
But most of the answers you get are based on a shared technical and educational background which you don't have and/or seem to value. It is difficult to describe but a lot of early computer science research was about how to effectively decompose functionality and data. Sadly, I think a lot of this is being lost. For a more technical starting point, look for the works of Edsger Dijkstra starting in the 1960s. For a less technical point of view, look for "The mythical Man-month" from the mid 60s (recently updated), and its spiritual followups by Ed Yourdon and Tom Demarco.
When I read the responses you get, I have the feeling that the authors have internalized the lessons of these important texts (even if they may not have studied them explicitly). Once you internalize the lessons also, then you will have a better understanding of the points of view you are consistently receiving with friction.
>I am hoping that I can publish some web pages with some comet circumflex
>codes and sentences about asking about the weather conditions and
>temperatures at the message recipients location together with codes and
>sentences for making replies so that hopefully people who might be
>interested in some concept proving experiments can hopefully have a go at
>some fascinating experiments with this technology. Unicode can be used to
>encode many langauges and it will be interesting to find out what can be
>achieved using the comet circumflex system.
That might be an interesting web site in its own right, but the technology is nothing special and has ben done a million times under a million names and ten million times with no name at all.
This archive was generated by hypermail 2.1.5 : Fri Sep 27 2002 - 14:06:01 EDT