spread thin
Jun. 24th, 2007 01:10 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
guslI have so many projects and I'm spread really thin. Yet, I seem to be doing ok... I wonder if I'm doing something wrong. (More than one professor has advised me to focus on a single topic)
Besides my 60%-time job, I am trying to publish in the short term in:
* causal graphical models (trying to run the study before I go to LA)
* text learning: applied to analogies (making progress today), classifying political orientation of bloggers (waiting for the annotators to give us some data)
In the long-term:
I have a project idea about Knowledge Management / User Modeling with Scone: have the system figure out what kinds of things you already know ("impedance matching"), in order to communicate with each person in a custom way (like this, but for non-math).
cogsci: I've been flirting with cogsci for several years now. Now that I'm going to GSS2007, it's about time I started doing some actual work there. Friday, I went to an excellent, AI-flavoured talk about methodologies for integrating ACT-R with other systems (e.g. Leabra), higher vs lower level, tight vs loose integration, etc. Some people proposed integrating with general AI systems (Cyc, WordNet), at which point I raised the question of cognitive plausibility. I explained my idea about "cognitive compilers", and guess what: it's been done before!
This week was the PSLC Summer School, of which I only saw a couple of hours. I realized that technically, I didn't have that much to learn from the CTAT course. But it was interesting to see how people were using the software to realize their ideas. Maybe next year, they'll let me teach.
I keep meaning to play around with argumentation-based dialog systems (i.e. declarative, rather than procedural tutors), but authoring these tutors still takes way too long.
Besides my 60%-time job, I am trying to publish in the short term in:
* causal graphical models (trying to run the study before I go to LA)
* text learning: applied to analogies (making progress today), classifying political orientation of bloggers (waiting for the annotators to give us some data)
In the long-term:
I have a project idea about Knowledge Management / User Modeling with Scone: have the system figure out what kinds of things you already know ("impedance matching"), in order to communicate with each person in a custom way (like this, but for non-math).
cogsci: I've been flirting with cogsci for several years now. Now that I'm going to GSS2007, it's about time I started doing some actual work there. Friday, I went to an excellent, AI-flavoured talk about methodologies for integrating ACT-R with other systems (e.g. Leabra), higher vs lower level, tight vs loose integration, etc. Some people proposed integrating with general AI systems (Cyc, WordNet), at which point I raised the question of cognitive plausibility. I explained my idea about "cognitive compilers", and guess what: it's been done before!
This week was the PSLC Summer School, of which I only saw a couple of hours. I realized that technically, I didn't have that much to learn from the CTAT course. But it was interesting to see how people were using the software to realize their ideas. Maybe next year, they'll let me teach.
I keep meaning to play around with argumentation-based dialog systems (i.e. declarative, rather than procedural tutors), but authoring these tutors still takes way too long.
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
(no subject)
Date: 2007-06-24 06:59 pm (UTC)(no subject)
Date: 2007-06-24 07:12 pm (UTC)Publishing by itself is not very hard: publishing in a top journal is, and the rewards reflect this. I think this is why the professors advised me as they did.