As the mentor for this PhD Session noted, the three talks given really show the eclectic mix that can be found in computer science. This was the first time I attended these forums, and I tried my best to fill in the feedback forms as best I could with useful comments. All three presenters did a really good job and were really well prepared, so my comments were really only of small things!
Warehousing Markovian Streams
Imagine that you have an RFID tag attached to you, and that several sensors record your movement around a building with time stamps. You might want to ask questions like "when did Bob enter the coffee room?" The only problem is that you can't be 100% sure where exactly someone is based on the RFID sensors, since there are overlapping signals, etc. Instead, there are a bunch of probabilities of Bob physically being somewhere, probably based on how close the signal is to the sensor.
Julie's research was all about having a database of all these probabilities stored as Markovian streams (I think). The key question was how to make it more efficient, and the main points of the answer centre around indexing and approximation. The Lahar database developed is efficient enough to run in real time as the data is streamed.
There are some cool applications of Markov streaming, so making use of this kind of data is definitely desirable. Some examples include using tracked information for diaries, health monitoring and fitness assessments. Markov streams can also be used to process audio streams, which may be very useful for sound search.
Classroom Resources and Impact on Learning
Margaret A Dickey-Kurdziolek
The big question for Margaret is whether there is worth in having technology in classrooms. I think this is a very interesting question indeed. After all, it's easy to try and bring in all the newest and coolest tech, but are kids actually learning more because of it?
Margaret focused on SimCalc. She found that when it came to test scores on standardized tests, the use of this technology didn't improve student results all that much. But when it came to the students' abilities to learn advanced math skills, the technology made a huge difference. This brings up a whole other issue about standardize tests hurting more than helping, but that's another blog post for another day.
The research focused on a selection of teachers from Texas who used SimCalc is various setups, from all students using it in the computer lab and having their own computer to the teacher just projecting one computer in the classroom. I actually don't recall the results between these setups, but Margaret did mention that the students who shared often faced problems, but that learning to share was highly valued by teachers.
I think this sort of research will be very useful in shaping the future of technology in the classroom, and am looking forward to seeing more of it as time goes on.
Augmenting Biographical Memory
The goal of this research is to help people remember the details and events of their lives. For instance, have you ever wondered "when did I meet this person and what did we talk about" after a day at Grace Hopper? Wouldn't it be great to have some easy way to recall these little details?
Current solutions for this are what we might call 'male-oriented'. It's kind of like someone noticed some cool tech out there and wanted to figure out a way to use it. Instead, Andrea took a more human approach and used cognitive science to figure out how people and memory work. She found out about the differences between memory cues and the memories themselves; the older methods of life blogging and semantic desktop don't really differentiate these things.
The highlight of this talk for me was the idea that computer science can benefit so much from so called 'softer' sciences (especially psychology). I completely agree with this, and I wish more computer scientists could be exposed to these ideas, even if they don't have to work with them directly.