Presentation slides (.ppt)
Davis, Ian. “Why Tagging Is Expensive.†Silkworm Blog. 2005. 18 September
2006 http://blogs.talis.com/panlibus/archives/2005/09/why_tagging_is.php.
Del.icio.us. http://del.icio.us.
Deli.ckoma. http://deli.ckoma.net/stats. Accessed September, 2006.
Mathes, Adam. “Folksonomies – Cooperative Classification and Communication Through Shared Metadata.†Adammathes.com. 2004. 18 September
2006. http://www.adammathes.com/academic/computer-mediated-communication
/folksonomies.html.
Shirky, Clay. “Ontology is Overrated: Categories, Links and Tags.†Shirky.com. 2005. 18 September, 2006. http://www.shirky.com
/writings/ontology_overrated.html.
Sinha, Rashmi. “A cognitive analysis of tagging.†Rashmi Sinha. 2005. 18 September 2006 http://www.rashmisinha.com
/archives/05_09/tagging-cognitive.html.
Vander Wal, Thomas. “Folksonomy Definition and Wikipedia.†Vanderwal.net. 2005. 18 September 2006 http://www.vanderwal.net
/random/entrysel.php?blog=1750.
Vander Wal, Thomas. “Tagging for Fun and Finding.†OK/Cancel. 2005. 18
September 2006. http://www.ok-cancel.com
/archives/article/2005/07/tagging-for-fun-and-finding.html.
Chapter four of Rosenfeld and Morville begins part two of the book, “Basic Principles of Information Architecture,” and introduces the reader to these principles. For the librarian or library science student, the concepts in these chapters are almost second nature, but the practical considerations the authors illustrate make it a very worthwhile read.
Worth noting are the explanations of “top-down” and “bottom-up” information architecture. “Top-down” information architecture refers to the process of identifying, predicting and anticipating the set of questions most likely to be in the minds of users of a site. Once an information architect has a grasp on these questions, she can set about structuring the site in such a way as to make the path to the answers as short and intuitive as possible.
“Bottom-up” information architecture refers to structuring content into meaningful and well-documented chunks in order to best support searching and browsing of a site. The idea here is that if the atoms of information that make up your site are accessible to search engines and have robust metadata, then users can discover answers to their questions outside of any navigational structure or page hierarchy you may have created. This concept strikes me as increasingly important as users become primarily searchers rather than browsers. Google is a powerful tool for directing searchers deep into a site, but improper “bottom-up” IA can easily make for search results which reflect old or inaccurate content, which may actually be worse than having a site which is opaque to search engines.
The key components of a site’s information architecture, according to Rosenfeld and Morville, are Organizational Systems, Labeling Systems, Navigation Systems, and Searching Systems. The first two, the subjects of chapters five and six, are old hat to librarians, although understanding the notion of a system like DDC or LC subject headings is quite different from actually creating an organizational or labeling system.
In fact, the timing of my readings couldn’t have been better this week, because I had the opportunity to attend an event directly related to the creation of organizational and labeling systems. Sponsored by the local chapter of the Usability Professionals Association and hosted by the Target Corporation’s user experience research team, the topic of the event was “Card Sorting from the Bottom Up (and the Top Down).” I’m going to forego a detailed recap of the rest of my reading in favor of sharing some thoughts about this event.
The members of the Target group gave a quick walkthrough of their card sorting process, which they use to generate taxonomies primarily for use in navigation on their internal corporate sites. Card sorting is an exercise in which people who have been recruited because they match the profile of site users are given a set of cards, each of which has a term or phrase on it, and are asked to sort them into groups of like cards. The terms on the cards are drawn from the content to be organized, and should be a representative sample of user destinations or tasks.
The “top down” and “bottom up” in the title of the event should not be confused with Rosenfeld and Morville’s distinction mentioned earlier. Rather, they refer to two types of card sorts, which the book calls open and closed, respectively. In top down, or open, card sorts, users are asked to create groups in which to place their cards, and to label those groups. These sorts have the potential for greatest use because they hint at potential structural and labeling schemes for content. Bottom up, or closed, card sorts ask users to file the cards into a predetermined and pre-labeled set of groups. These card sorts are primarily used as a means of validating results from earlier top down sorts.
Reaction from some of the other usability professionals in the session seemed to indicate that the Target group was somewhat unique in that they preferred to run their card sorting exercises with small groups of users, rather than single users. The value they see in using groups of 3-5 users is in the users’ conversation, which can often give a glimpse into users’ thought processes and give the researchers more data than is reflected in the piles of cards at the end of an exercise. The tradeoff, of course, is that group dynamics can override individual users’ behavior. One strong-willed user can unduly influence a group’s results, and anyway, when was the last time you browsed a web site with 2-4 peers?
The Target team said that they often find that patterns emerge quickly, which was borne out when the event attendees did their own card sorting exercise in small groups. The same terms and concepts were echoed by group after group. Ideally, the Target team performs top down card sorts with multiple groups in each user demographic that they identify for a given site. The results from these exercises are then tested with bottom up card sorts. It strikes me that bottom up card sorts may also be a useful opportunity to test any terms or other organizational schemes that may be second nature to the site’s creators, but obscure to the site’s users, such as internal organizational charts.
Lastly, the team shared some of their techniques for data capture and analysis, which they warned was not particularly pleasant work. The tool of the trade seems to be the Excel spreadsheet, which allows for easy sorting and resorting of data by card and by group.
It appears that Rosenfeld and Morville go into greater depth about card sorting later in the book, but I was glad to have the opportunity to get an introduction to the topic from some skilled practitioners.
