|
Post by Hannes Vilhjalmsson on Mar 11, 2009 18:52:33 GMT -5
This week we'll be reading about the system I was working on at the USC Information Sciences Institute for 3 years before joining Reykjavik University. It provides a good overview of how AI can be applied to the area of education and training and gives you some of the background for my work here at CADIA. The paper is called "The DARWARS Tactical Language Training System" (and yes, DARWARS is the official name of the DARPA program that funded us). Check it out and write your questions here by midnight on Sunday. I'm looking forward to hearing your thoughts.
|
|
|
Post by halldorrh05 on Mar 12, 2009 14:46:26 GMT -5
How would techniques as used in the rosetta stone (http://www.rosettastone.com/) compare to the ones used in DARWARS?
How quickly on average can an ordinary soldier using this tool learn the language sufficiently to have a rudimentary conversation? What methods do you use to figure out what the soldier said and how appropriate it was in the context?
|
|
|
Post by Jón Trausti on Mar 14, 2009 7:19:45 GMT -5
Very interesting reading.
How perfect is the sound recognition? I read that a lot of pronunciations were rejected, mostly due to saying something slow. Could one "hum" the sound to get it correct? (Kinda like in singstar?).
How well did this project go at the time? And how is it being used today?
|
|
|
Post by Hjalti Kolbeinsson on Mar 14, 2009 8:39:33 GMT -5
When designed, how fast was it meant to be? That is how fast was one supposed to learn a new language using this tool? Was the thinking behind it that if soldier A was going to country B in 6 months he should start using this tool right away and be ready in 6 months or was it more like A was going to country B in 2 days or a week and should learn enough to say what needs to be said in that short amount of time?
Can we see the program in action?
|
|
|
Post by Helgi Páll Helgason on Mar 15, 2009 8:30:16 GMT -5
This crash course approach to language learning is very interesting. Are there any plans to commercialize this technology/methodology for the public?
Has any of Kristinn's work on language turn-taking been used in this project?
What happened after 2004 with regards to this project/technology? Have Iraq-bound soldiers been trained with these tools?
|
|
|
Post by Hólmar Sigmundsson on Mar 15, 2009 10:21:18 GMT -5
Have any languages been added to the training system since 2004?
How flexible is the system when user/learner answeres questions in context, but not the anticipated way?
|
|
|
Post by Birna Íris on Mar 15, 2009 11:38:48 GMT -5
Have the learned language/social skills learned in TLTS been compared to skills learned in a classroom? I have experience in learning social skills for Saudi Arabia in a classroom (not for military purpose though) and think it would have been very interesting to use such a system for that. This also brings up the question about Saya the robot teacher in Japan. Can a human teacher be replaced by a robot? What do we need to accomplish that?
The emotion arousal level in the characters is interesting. Can you describe (point out some resources) the PsychSim multi-agent system and how the agents are realized as POMDP (Partially Observable Markov Decision Processes)?
|
|
|
Post by Jon Gisli Egilsson on Mar 15, 2009 12:42:02 GMT -5
Have you had direct feedback from a user that trained his language skill from scratch with the TLTS and went abroad to use it? What did they say?
Does the army not want to commercialize this program or is there not a big enough market? I for sure would like to be able to try this thing before visiting foreign countries.
|
|
|
Post by olafurgi on Mar 15, 2009 16:11:17 GMT -5
The most interesting part of this project for me is the audio/vocal recognition. What were the difficulties in the audio / vocal recognition ? Were there any concerns of cpu / memory usage regarding audio/vocal recognition ?
Any parts of this project that have been reused in other similar projects ?
|
|
|
Post by Christian Zehetmayer on Mar 15, 2009 18:04:49 GMT -5
Are there any thoughts about commercializing and add commonly taught languages? Is only one-way (English – Arabic) training available?
How many people getting trained with it? Is it a big success? Which artefacts should be improved to get a better language training system?
|
|
|
Post by steinarhugi on Mar 15, 2009 19:17:30 GMT -5
Today even the voice recognition tools for English are barely usable unless you're speaking it perfectly. How did you manage to understand, for example, Arabic spoken by a beginner?
Did you do any research on analyzing body gestures automatically using a camera?
|
|
|
Post by Snorri Jónsson on Mar 15, 2009 20:41:17 GMT -5
How many people are using this system to day ?
Any recent changes to it?
|
|
|
Post by hordur08 on Mar 16, 2009 8:25:26 GMT -5
Very interesting approach in language teaching !!
1) Has it been compared to other methods of learning a language?
2) Has this technique been used in teaching other subjects than language?
|
|