The play with algorithms last week was thoroughly enjoyable. It put into perspective how I choose the content I consume. I spent the week playing around with YouTube’s watch time algorithm and managed to test the codes ability to recommend videos based on the length of time a video is watched. I am a frequent user of YouTube (I also have my own channel based on travel) and I more often than not watch videos based around consumer electronics and travel. Based on this fact I am constantly given suggestions on similar content and never really redirected away from these specific topics or genres. For example, I enjoy technology and travel, so a lot of my videos recommended revolve around these two themes. Now I would not consider myself an expert on these two topics but I feel my knowledge base increases the more videos I watch. The creators of the videos are similar to teachers who are conveying content to students (viewers).
So a question we were posed this week was how can we use these recommendation algorithms in education. These algorithms need to work in the same way in which videos are recommended on YouTube for the consumer. However instead of suggesting videos the algorithms would suggest courses learners could enroll in. These optimization algorithms , also termed personalization algorithms (Rauch, Andrelczyk and Kusiak, 2007), collect user information and analyze the data so that it may be relayed to the user at specific moments (Venugopal, Srinivasa and Patnaik, 2009). For example, when I am finished watching a video on YouTube or movie on Netflix I am surely presented with a list of recommendations on the genres I have just consumed. This idea works similar with personalization algorithms that would be able to recommend courses or avenues of learning based on the learners prior knowledge or courses completed.
An example of a personalized learning algorithm I came across was PERCEPOLIS (Hurson and Sedigh, 2010). This particular algorithm adapts various pieces of information and takes the individual pieces to produce a recommendation for the user. The information analyzed was data from a learner profile, teacher profile and environment profile. These are all known as learning analytics. I can definitely see the real world applications of such a tool especially when students are going into colleges blindly not knowing what they want to study. This could help alleviate some of that pressure and make the student’s learning more personalized. I know I could have used assistance like this when I was in my undergrad. Would have saved myself a lot of headaches and tears.
Searching this topic of personalized education further I found that perhaps the term personalization is misused and what the algorithm is doing is not personalizing the learning at all. Instead the algorithm is taking the onus away from the learner. Trent Baton (2015) believes that personalization of learning is still the responsibility of the learner. He does admit though that it is impossible to deny algorithms and the importance they play in our lives but one needs to understand that technology is there to enrich or enhance human lives, not to run it. This is his biggest fear with regard to recommendation algorithms in education. I tend to agree with him. Ultimate control still needs to be with the learner and not just left to an equation. He goes on to say that, “Algorithms, like good teachers, need to be guides on the side and not new sages on the stage.” His thoughts here are very explicit. We need to maintain control of our own learning.
It was an extremely eye-opening week and on to the next one.
Em
References:
Baston, Trent. (2015). “Personalized Learning: It’s Not the Algorithm, It’s the Learner – AAEEBL.” Personalized Learning: It’s Not the Algorithm, It’s the Learner – AAEEBL. Association for Authentic, Experiential and Evidence-based Learning, 17 Feb. 2015. Web. 15 Mar. 2015. <http://www.aaeebl.org/blogpost/1008436/208980/Personalized-Learning-It-s-not-the-Algorithm-It-s-the-Learner>
Hurson, A., Sedigh, S. (2010) PERCEPOLIS: Pervasive Cyberinfrastructure for Personalized Learning and Instructional Support. Intelligent Information Management, Vol. 2 No. 10, 2010, pp. 586-596.
Rauch, L., Andrelczyk, K., Kusiak J. (2007) Agent-based Algorithm Dedicated to Personalization of e-Learning Courses. In: 20th EADTU, Lisbon
Venugopal, K.R., Srinivasa, K.G., Patnaik, L.M. (2009). Algorithms for Web Personlisation. Studies in Computational Intelligence Volume 190, 2009, pp 217-230