Card sorting your way to a better you
October 7, 2008
I’m trying something new in trying to get a handle on the topic of open source and what research people have done on it so far: a card sort.
Card sorting your way to a better you
I’m trying something new in trying to get a handle on the topic of open source and what research people have done on it so far: a card sort. I’m just using the most basic idea from the card sorting technique to try to estimate, looking at titles and abstracts, what areas researchers in open source have been looking at regarding open source, how much they looked at them (how many papers), and when they looked at them (what years).
Card sorting is a technique from information design and usability in which you’re looking for how to organize a set of information so that the people who need to find that information . You write concepts on cards (index cards, typically), give them to someone (typically either a domain expert, or a person who would have to use the information) and ask them to organize those cards into groups as they see fit. Repeat for several similar people and correlate the results, and you’ll get an idea about how most people expect to find things. A lot of the literature on card sorting involves correlating the different sorts that each person produces for the same set of information and generating a final organization from that.
A typical use of card sorting is to define a menu system for a website, and you’re trying to answer questions like: will people look for pitchforks under “gardening” or “tools” (or both?).
Card sorting open source papers
I first went to the Web of Science database via http://libraries.claremont.edu, searched for “open source” and downloaded all the papers it found into Endnote: 337 papers between 1998 and 2008.
I then wrote a Python program that parsed those citations and printed each one onto little cards, assigning each card a unique number.
My plan is to sort those 337 papers into broad groups, then sort the papers in each of those groups into smaller groups, and so on until I’ve got the papers organized into coherent and cohesive topics. I’ll enter the numbers of each of the papers in each group into a spreadsheet and analyze the data in the spreadsheet to generate a subject hierarchy. I should also be able to see when papers in each topic were published, so that I can see if a topic is dead, living or just being born.
I hope to get something like this diagram I built for e-mail literature, but more useful. I built that diagram by looking at titles of thousands of papers on electronic mail, but my flaw in that process was that I didn’t record which papers belonged to each topic. As time went on, I had trouble remembering what I’d meant by each topic, since I couldn’t refer back to example papers.
