louisrosenfeld.com logotype

Home > Bloug Archive

May 18, 2005: Experimenting with Tagging

Do you use del.icio.us? And are you interested in enterprise information architecture? Then consider tagging your EIA bookmarks with the tag "enterprise_ia". If at least a few of us start doing this, then, as the tag gods suggest, we'll all benefit from each other's research by monitoring the tag at "http://del.icio.us/tag/enterprise_ia". Please spread the word to anyone who you think might be interested.

I'm hoping others do this for more than the purely selfish reason of being interested in enterprise IA. My RSS aggregator currently includes tag feeds for "information architecture," "user experience," and many of their common variants. I've found that monitoring these tags is a sort of useful way to keep up with these two areas. But only sort of. I'd really prefer to track more specific topics (like enterprise IA or search log analysis), as I assume I'd be exposed to fewer, more specialized, and therefore more useful results. Unfortunately I haven't found many sufficiently precise tags in del.icio.us land.

One of my concerns with folksonomies is that many taggers tend to tag a bit too broadly. How do we solve this problem? I'm hoping that a little coordination among people with a shared interest--as I'm suggesting with "enterprise_ia"--might go a long way (assuming we generally agree on what EIA actually is). But even if this somehow succeeds, it doesn't strike me as a scalable long-term solution. What are other ways of encouraging precision in folksonomic tagging?

email this entry

Comment: donna maurer (May 19, 2005)

I think a simple improvement for something like del.icio.us would be to expose suggested tags when someone tags a page. If the list of tags that other people have used was available at this point, it would be easy to select an existing tag, or use your own. Best of both worlds.

Comment: Chris Armstrong (May 19, 2005)

I share your concerns with regard to the choices of the folk using folksonomies (see http://i-a-l.blogspot.com/2005/03/folksonomies.html and article in eLucidate 2(3) shortly to appear as "Outguessing yourself!" at http://www.ukeig.org.uk/content/public/activities/newsletter/index.html).

The good news is that Connotea have indicated that they may be prepared to consider the possibility of suggesting tags. That is to say if I type in "publishing" it could use some sort of stemming algorithm to suggest that 'most other taggers use "publishers"'.

I suppose the other possibility is for an advance in RSS readers so that a variety of tags can be collected in one feed. That's the choice: pre-coordinate or post-coordinate!

Comment: Per Axbom (May 19, 2005)

The best way to encourage change in behavior is to provide good examples and success stories. And your idea with "enterprise_ia" has potential for evolving into a best practice :)

Comment: Joshua Porter (May 19, 2005)

Lou, what makes folksonomies powerful is that this sort of thing doesn't happen. What is powerful about them is that people tag things how they want to, and then we aggregate those things to see patterns and trends.

It leaves open room for much more subtlety in meaning. The way folksonomies can address the issue you're bringing up is to do pattern matching across users and resources. In other words, look at what resources were tagged in similar ways.

So, lets say we have a resource...a great article...about information architecture on corporate intranets. You and others tag it "enterprise_ia" and I and others tag it "intranet development". Then, we built interfaces to show this relationship. In other words, we look at all the things that this resources has been tagged, and pay attention to those tags that were used repeatedly. We'll discover that these two things are related...but different. *Related, but different is the key!* If we both went ahead and tried to force them into one tagging scheme, we would only know about "enterprise_ia", and perhaps not find the subtle difference with intranet development.

You seem to be advocating the creation of a controlled vocabulary within del.icio.us. I think this is a bad idea. Folksonomies are interesting because that's exactly what they are NOT.

Comment: James Mason (May 19, 2005)

To echo some of what Chris and Joshua have pointed out, tagging doesn't work like traditional categorization. The benefits come *directly* from the lack of centralized control. There's a great article by Clay Shirky that explains this: http://shirky.com/writings/ontology_overrated.html

To quote:

"Similarly, the idea that the categorization is done after things are tagged is incredibly foreign to cataloguers. Much of the expense of existing catalogue systems is in trying to prevent one-off categories. With tagging, what you say is 'As long as a lot of people are tagging any given link, the rare tags can be used or ignored, as the user likes. We won't even have to expend the cost to prevent people from using them. We'll just help other users ignore them if they want to.'"

Comment: Rishi Khanna (May 19, 2005)

James & Joshua, I think Lou is asking a different question.

He is asking is there any way to get precision from folksonomies. Allowing anyone to tag anything any way they like doesn't help with the findability of very specific interests; not if people refer to the same thing differently, and they do.

Folksonomies are fine for "browse" searching. But if your business demands precision and completeness, as mine does, folksonomies are severely flawed by themselves.

The question I've been struggling with is, are folksonomies, when combined with taxonomies, something more powerful that can solve the problem better? How would that combination work? Is that just suggested terms and some mapping from there?

Comment: Joshua Porter (May 19, 2005)

Rishi, the findability of very specific interests is exactly what I think folksonomies protect. In the example that I gave above, there is meaning lost if the group throws many things under the same label, especially if those users would have tagged the item something different on their own. Instead of being about both enterprise ia and corporate intranets, it's only about one of those things. What if somebody searches on corporate intranets? What to do then?

Comment: Lou (May 19, 2005)

Rishi, you beat me to it; thanks. Joshua, the issue I've raised isn't about meaningful relationships between terms (which are a good thing), nor is it about controlled vocabularies (I'm not sure why you think I'm suggesting them here; I certainly didn't mention them, and I'm honestly not sure how or why one would impose them within a decentralized system like a folksonomy).

The issue I'm interested in is precision, and my goal here is to explore how to introduce greater precision to increase the utility of systems like deli.cio.us.

In other words, I'm posing my question as an interested but slightly frustrated user who wants del.icio.us to work better, not a hierarchically-fixated, authoritarian, old-school, librarian with an irrational, blind hatred for anything that threatens my lucrative metadata franchise. ;-)

Comment: Michael (May 19, 2005)

We've been running an experiment in our Taxonomy class with using IMT530 to tag materials referenced in class or used for projects in class. http://del.icio.us/tag/imt530 We've subscribed to the main RSS feed and easily get updates. The way I explained it was using a controlled vocabulary in a social tagging tool. I have also been doing this with some friends that are adopting and we have been able to share adoption sites by subscribing to each others tags (instead of the overall adoption tag.) It wouldn't work for the world to use this approach, but it will work for a small group (like IA's!).

Last week we had gentleman from Microsoft speak that was explaining the process of tagging content and the challenges. He fell for my trick question as to how many people would he like to have tagging content to reduce the personal bias and how many he has. He wanted more than he had. Ah ha! Isn't this where we can start implementing social tagging with the tagging the professionals are doing? This gives us the mixed benefit of many taggers and professional tagging. Neither one is perfect by the admission of all involved.

I'm wondering if "enterprise_ia" is going to be a little broad of a topic for many contributors? Let's find out.

Also, I'm looking forward to you speaking to our class on the 27th!

Comment: Joshua Porter (May 19, 2005)

Lou, thanks for the clarification. You're right to call me out on my over-zealousness. However, I'm suspicious of techniques to homogenize tags (even if it is a precise homogenization) in systems like del.icio.us that, in my opinion, are interesting because they place no restrictions whatever on what people are tagging things as.

I think tags are special: they're our "blink" reaction, to borrow from Gladwell. They're the first words that we think of to relate to something. If we're concerned with what others are tagging, then I think we negate that initial reaction. So, if I'm wondering if I should tag something "enterprise ia", then I might not tag it what I would have originally tagged it. If that happens, is meaning lost? I wonder.

Taken to the extreme (and I admit that I'm pushing here) it becomes a controlled vocabulary. In other words, it's pigeon-holing content into a category, and not looking for natural patterns in the self-categorization, which I think is the powerful bit. And though I think there might be utility in the way that you described, I think it's a subtle shift away from the spirit of the tool.

Comment: vanderwal (May 19, 2005)

Precision in a folksonomy comes from people tagging for their own self retrieval. When people start tagging for others we get into metacrap territory.

Joshua, it is a good thing that del.icio.us does not limit what one person can call something. The tool has the means to block individuals you do not want included into the grouping you have an interest in.

Comment: Lou (May 19, 2005)

Joshua, thanks for the comments. I'm still not sure I'd see my suggestion even remotely as a form of controlled vocabulary. "Enterprise_ia" wouldn't be a required or even preferred term for anything. Nor would it be an exclusive term; users could continue to tag with as many additional terms as they normally would.

*If* others were to use the tag, it'd simply be the foundation of a mutual, non-exclusive pact to share information by an informal group who opted into such an arrangement.

In a way, discouraging this sort of emergent communal activity in favor of a "right way" to tag--namely, the individual tagging bookmarks "blink"-style (assuming that this is how people really do their tagging)--seems to me completely contrary to the unorthodox, anti-authoritarian spirit of del.icio.us. Look at it this way: if ten of us agree to tag bookmarks as "enterprise_ia" and we don't force our tag on other users, then deli.cio.us has made the world a better place for us. OTOH, if this activity is to be discouraged, then the value of del.icio.us becomes cloudy. In this latter scenario, the goal of del.icio.us would almost seem to be the development of a folksonomy for its own sake, or as an interesting exercise, or to prove a point, rather than deliver value to end users.

Comment: James Mason (May 20, 2005)


I think you and Joshua might be talking past each other here. I see two distinct purposes to tagging, but I donít know that Iíve seen them explicitly separated before.

First, the purpose that most people (including Joshua) seem to be abuzz over, is using tags for personal findability. Adding precision in this case would destroy much of the usefulness of the folksonomy that develops, since you would lose the richness that comes from the diversity of tags. I think this loss is what Joshua is cautioning against.

The second purpose is the one I think youíre talking about: using tags for collaborative feed building. Michael pointed out two good examples of this use, and I know Iíve seen many others (especially in relation to flickr). For this purpose, precision is a necessity. It wouldnít work if everyone didnít agree on a single term and the meaning for that term.

The more I think about this, the more the second purpose seems like a form of ad-hoc, open access mailing list. This seems extremely useful to me, but as soon as I bring up the term ďemailĒ I immediately jump to the term ďspam,Ē and I think ďHey, I could bypass Louís spam filter by tagging my junk with enterprise_ia.Ē I donít know that anyone has gone there yet, but itís probably inevitable.

What really interested me, from the perspective of the second purpose, is using it internally within an organization. It seems like an incredibly simple way for a large number of people within an enterprise to create an ad-hoc community for collaboration. Something like the way technorati uses tags, with the added benefit of being able bring items to the attention of the group that you didnít write yourself. A single email to start the ball rolling and the IT department doesnít need to do *anything* to support the effort.

Thanks for the inspiration, Lou!

Comment: Lou (May 23, 2005)

James, thanks for the comments. I guess I don't see those two purposes as being mutually exclusive. Why can't I "blink" tag a bookmark *and* tag it with a more precise, collaboratively-determined tag like "enterprise_ia"? I might be missing something, but I just don't see why these two purposes can't co-exist.

Of course, the spam issue is a big one, as you point out. (Damn!) But maybe tags managed by loose collaborative groups will be difficult for spammers to identify precisely because they're so precise (pun intended), esoteric, and therefor unlikely to be advertised widely?

Comment: Joshua Porter (May 23, 2005)

I think James is right about the two things going on. And I think Lou is right in that they aren't necessarily mutually exclusive.

My concern is that from a contextual standpoint that they might as well be mutually exclusive. If I'm tagging something for a group purpose then I probably won't be tagging it for personal purposes, and vice versa. I'm speaking from my own experience, but I'm also interested in learning if del.icio.us can tell us something about that in it's own aggregation of data.

Add a Comment:



URL (optional, but must include http://)

Required: Name, email, and comment.
Want to mention a linked URL? Include http:// before the address.
Want to include bold or italics? Sorry; just use *asterisks* instead.

DAYENU ); } else { // so comments are closed on this entry... print(<<< I_SAID_DAYENU
Comments are now closed for this entry.

Comment spam has forced me to close comment functionality for older entries. However, if you have something vital to add concerning this entry (or its associated comments), please email your sage insights to me (lou [at] louisrosenfeld dot com). I'll make sure your comments are added to the conversation. Sorry for the inconvenience.