from Academic Recommendation Systems to the Fishscale

A bit more than a year ago I made the acquaintance of Jenny Delasalle, who also had a chapter published in the Only Connect book (I actually met her at the launch of that book). And last week she asked me in an email whether the Fishscale was still working in practice. As probably a lot of you, I’m at the point of the year when I think about what has worked well last term, and what maybe didn’t work quite so well (and yes, partly this is a very constructive way of procrastinating in between marking), so this was quite a well-timed enquiry.

Last term I changed my delivery of the Fishscale slightly. Some of the feedback that I get on this activity always included bafflement as to where the fish are coming from. Students seem to have trouble understanding that I’m not bothered about the fish, it is the concept of provenance that I care about, and that I think they should care about. But some students appear to feel slightly patronised by the format of this which is written in the form of a children’s book (and that is sort of on purpose). So I thought maybe it would “breed” even more ownership of the concept, if students were developing their own ranking systems in class. So for some of my modules last term, I assigned watching the Fishscale stand-alone presentation as homework, and then got students to discuss this concept and develop their own ranking systems in small groups, which they were then expected to apply to their literature for the rest of the term.

One of the things that I was hoping to get them to think about is one of the things my students, who are all studying creative studio-based disciplines, often struggle with and that the Fishscale (that is conceived for more generic use) doesn’t take into consideration: where do technical, how-to instructions and creative, visual, inspiration type sources fit in? So I wanted them to pay particular attention to this when they developed these systems for their own use, or adapted the Fishscale (which is an option they of course also had).

What happened was quite interesting. For a start there are of course students who don’t do homework, who found themselves in a session talking about fish and secondary sources and had no idea what was going on. But overall the groups came up with some really good ideas for their own systems, ranging from sandwiches to different magic tricks to levels in computer games. There were also some groups who really didn’t want to go visual and went for numerical ranking (values 1-10 or letters like in grading). But the really interesting thing was that pretty much every group complete and utterly ignored the stipulation that this was not about the content of a source, but rather its type, and they all insisted that relevance (of content, not of type of source) should be the most important thing for them to look at in order to judge whether this was a useful source or not. At this stage of their research – these are all first year students – this means that they are basically sabotaging themselves. They seem to have something in their head about what information is relevant, and they discount everything that doesn’t fit, meaning that in their literature search they never really find the things that are new to them or that would allow them to broaden their horizons.

I mentioned this to Jenny, and also that I wasn’t quite sure what to do with this information, and she replied that one of the things she looks at with the students in the information ethics course that she teaches are recommendation systems, the most know probably the recommendations you get from amazon. She said that they “are partly based on relevance but also try to incorporate elements of serendipity and broader themes. Some theorists warn against too high a relevance in fact, because it makes all that spying and user profiling that they do really obvious and undermines their trustworthiness!” She then suggested getting the students to look at how different services recommend movies, as broken down here. I have to say, I absolutely adore this idea, and I love this particular link. I didn’t realise that there were so many different movie recommendation sites out there – and that they all worked so differently!

So in future I definitely want to try to look at recommendation systems before thinking about our own system – and the Fishscale can still be the example we talk through in detail, before we critique it.

It also made me think about the ‘recommendation’ systems that we have in place in academia, the reading lists, the bibliographies of the sources we already have, the shelves in a university library that basically present us with related books, but add that serendipitious element. And while, as far as I know, there are no algorithms putting all these things together, it might be a good idea of thinking of these elements as recommendation systems we are now getting more and more familiar with in our (digital) lives.

 

Inspired by our email conversation Jenny also wrote a blog post on recommendation systems, which you can find here – we might even start a blog conversation!