AFRO-NETS> Supercourse Newsletter, December 7, 2003

[Ron LaPorte <super1+@pitt.edu>]

Supercourse Newsletter, December 7, 2003
----------------------------------------
 
http://www.pitt.edu/~super1/
 
Friends,
 
We wanted to bring you up to date and to discuss several very in-
teresting issues.
 
Current Status: 
 
Eugene reports that we have an astonishing 1,585 lectures which 
have come from 708 authors. We likely have the largest project on 
global health that has been developed. Pretty impressive stuff.
 
Last week I was at the University Kansas medical school present-
ing to medical students, MPH students in preventive medicine and 
nursing students. Before beginning my talk I asked if people had 
heard about and used the Supercourse. An amazing 50% had heard 
about and used it. Clearly we are now getting wide recognition. 
It would be wonderful if you could ask your classes if they have 
heard of the Supercourse. Let us know what percentage has, and we 
can continue to track the use of the Supercourse world wide.
 
Translation:
 
We wanted to give you a general overview of the issue of transla-
tion of the Supercourse into different languages. When we started 
the Supercourse 5 years ago we had thought to have continuous 
translation of the Supercourse into Multiple Languages.
 
We began to try this out by translating the first lecture into 8 
different languages. Also, many of you offered to translate for 
us, which was fantastic. We soon discovered that this was impos-
sible using traditional translation approaches. One problem was 
that say I wanted to change the first lecture. To update it we 
would have to update 8 different versions, which was caused enor-
mous problems of logistics. With 1,585 lectures we would be 
spending all of our time translating. 
 
A second issue is that we wanted to reach you, the top, elite 
scientists in prevention across the world. To be at your caliber, 
one almost has to read English as most of the top science is pub-
lished in English. Finally, we figured that if we were providing 
lectures in English, it is virtually certain that someone in the 
Universities across the world spoke English. Thus, although we 
would have liked to provide translation into multiple languages, 
especially as the Supercourse is globally, financially, and lo-
gistically it is impossible. We do, however, have Supercourse 
discussions in Japanese with Akira, Russian with Eugene, and Ara-
bic with Abed. Some of you have sent lectures in other languages, 
which is great, however, we all know that if a lecture is on the 
Supercourse in Armenian, it will not reach as broad as audience 
as in English.
 
We wanted to do something different, and had dreamt about "just 
in time" translation. This would be a system whereby all you 
would have to do is to put the URL into a "machine translator" 
and it would take an English language lecture and translate it 
into German or Mandarin. We felt that we would be an ideal test 
bed as if my lecture on type 1 diabetes were translated, the per-
son who wanted to use it would be an teacher somewhat knowledge-
able in epidemiology and diabetes hence they could guess pretty 
well as to our meaning. However, this was not the case as say if 
we had translated the sentence "Childhood diabetes has the lowest 
incidence in June" we would get, "Cliff?s dog was the lion on in-
cense in June". Clearly 5 years ago machine translation was not 
the answer. However, it is exciting as enormous advances have 
been made.
 
Five years ago machine translation was rule-based, and insensi-
tive to context. One thus received rather idiotic translations. 
Now, newer more powerful approaches such as language weaver and 
meaningful machines use a system of statistical translation. In-
stead of relying on grammatical rules, it uses statistical ap-
proaches to identify the best possible word, or sentence in the 
text.
 
This is from PC Mag:
 
Language Weaver's software can simultaneously investigate many 
more word groupings than any human translator could conceive, ex-
perts say the software translates more efficiently than rule-
based systems. Statistical analyses are also deemed to be better 
for distinguishing idioms such as, say, "Don't have a cow" from 
statements like "We don't have a cow."
 
We do not have any connection with these companies but they look 
really cool for what we are doing. If you want to read more about 
this, please take a look at: 
http://www.amtaweb.org/summit/MTSummit/FinalPapers/110-Benjamin-final.pdf 
http://www.pcmag.com/article2/0,4149,1401163,00.asp
 
We are really excited about the just-in-time translation, and are 
looking into this for the Supercourse. We would love to hear any 
of your suggestions.
 
Lecture of the week:
 
While working on the Golden lecture of prevention, our group re-
ceived many questions and queries about the issues related to the 
prevention of HIV/AIDs epidemic. A great number of Supercourse 
members felt that such important issues need to be addressed fur-
ther. Recently, we received a lecture that can serve as a New 
Year?s present to all of you who are interested in this problem. 
Drs. Salim Abdool Karim and Quarraisha Abdool Karim contributed a 
lecture titled The Evolving of HIV Epidemic in South Africa that 
can be accessed at:
http://www.pitt.edu/~super1/lecture/lec12791/index.htm
 
The lecture described HIV/AIDs epidemic in South Africa and gives 
the reader updated epidemiologic data about the morbidity and 
mortality data from this disease. In addition, the authors give 
their suggestions for future directions of HIV/AIDS prevention 
activities in South Africa.
 
If you are interested in contributing more lectures about HIV/ 
AIDS to the Supercourse, please contact us!
 
We may have over 1,600 lectures before Jan. 1, wouldn?t that be 
incredible. We just got 10 inches of snow. There are huge advan-
tages living in warm countries!!
 
Best regards from Pittsburgh,
 
Ron, Faina, Mita, Soni, Akira, Eugene, Julia, Tomoko, Abed, Ra-
nia, Sherine, Wendy, Arin, Denish, Charles
mailto:super1+@pitt.edu

--
To send a message to AFRO-NETS, write to: afro-nets@healthnet.org
To subscribe or unsubscribe, write to: majordomo@healthnet.org
in the body of the message type: subscribe afro-nets OR unsubscribe afro-nets
To contact a person, send a message to: afro-nets-help@healthnet.org
Information and archives: http://www.afronets.org