NC State’s Tiffany Barnes on the Big Graph


Week 6 Voices from the field. Interview with Tiffany Barnes


“One of our biggest advances in my lab was the idea of taking individual student experiences, their data from interaction…from procedural problem-solving environment, and putting that into a large graph of all of the students’ experiences–all of the steps they take when…solving problems. …then using those steps in that graph to then look at a new student who comes along, match them to someone who’s done the problem-solving…exactly the same way, and then use that as a source of domain knowledge that we could use to give hints to new students.”


I liked Tiffany the moment she described putting all student data from her project into a large graph. I don’t recall any single moment which has produced more excitement for me in my graphic design career than those problems which found solution not in simplification, but by growth. It’s very rare outside of Big Data to find opportunities to look at information at these scales. More importantly, thanks in part to DALMOOC, I now recognize what putting data into a graph entails.


Of course, our interest today is the Learning Analytics context. Mrs. Barnes continues by describing her project’s intervention as introducing targeted hints in line with each student’s learning profile. A conclusion demonstrated by the experiment says the machine generated hints have the effect to reduce the completion time for the student activity by an estimated 45 percent. We’ve talked a lot in this course about informing instructors how they can better understand learner’s activities. This is an example of taking the measurements of student activities one step further using the intelligence from the analysis within another activity. While the metric is performance, the project has discovered machine learning makes it possible for instructors to map each student’s individual pathway to knowledge.


You might recall a few years ago the Khan Academy was garnering a great deal of public attention. Thoughtful people were asking difficult questions about the future of education in America. Listen to Salman Khan describe his vision in this 2011 Ted Talk. Khan asks us to question the traditional learning activity where an instructor lectures to students on a topic, and for homework, student take home all the problems. In other words, the actual moment where the student is in greatest need of personal assistance, is at home when they’re by themselves. Khan’s web site suggested this relationship could be reversed. Homework could be watching video lectures in front of the computer at home, and school work could focus on working with teachers and fellow students on applying these new concepts in the classroom setting.

Following this reasoning, the most important contribution of teachers to the learning process is their ability to coordinate individual student’s learning needs–giving individualized help to students. Is machine learning’s capacity to track each and every student’s learning pathway a threat to teachers? That’s the fear and promise. No single instructor can cater to every student in the classroom, or compare one student’s profile to any number of students from previous years. Try comparing television to video games. DALMOOC has already accomplished a lot to provide a baseline practical understanding of these technologies. What is certain, the information revolution continues to change every aspect of our society.