Last week, my colleague and supervisor Jan van Bruggen and I conducted a workshop on recommendation strategies at the PLE conference in Berlin. The conference itself was also very interesting but I will blog about it in a separate post. In this post, I will describe the workshop.
Our workshop focussed on recommendation strategies for social networks supporting Personal Learning Networks. Below you find my presentation.
The reasoning behind the research we conducted and the topic of the workshop is that current recommendation systems look for navigating through the abundance of people and resources out there. To minimize the choices, they are therefore created to look for items or people similar to the user. However, for learning, this may not be the most relevant strategy. For learning purposes, you would actually like to look at people or items that are related to whatever you are interested in, but that are different in some significant way. This difference creates an interest and a learning opportunity. In our workshop, we presented two methodologies to match people on similarity and on dissimilarity.
The workshop was very interesting from my point of view, for various reasons:
- It was very interesting to see that even in the select group of participants of our workshop, we could see very different types of matches emerging, by using the different methodologies.
- As we asked the participants to basically do the computer’s work, it was funny to see how quickly they brought in other, more intelligent steps to improve recommendations. We had to be quite strict in forcing them to follow the computer’s rules. As a designer, it certainly showed me the limitations of the technical side and, of course, our natural intelligence 🙂
- What I also liked was how quickly the participants had seemed to make up their mind about whom of the other participants to connect with. Although we asked them to choose just one or two, most of them revealed to us afterwards that such a choice was difficult to make (although some did indicate initial preferences). Also, it was clear that such choices were dependent on a whole range of things, from professional to personal to other.
My reflections on the PLE conference will follow soon.