Adaptive Path Blog

This week I joined 1100 other folks at the Semantic Technology Conference in San Jose. I attended this conference back in 2007 and I’m happy to say I really see a difference in the past two years. Back in 2007, everything about the conference was about the technology. What was the code that made this stuff go? I tried to keep up in a number of sessions where they kept flashing XML up on the screen. I’m happy to report that my experience was much different this year.

From the moment of the first keynote, folks were talking about the user experience. Yay! Our message is finally getting out there. It seems to me that they have finally gotten the technical bits mostly figured out on how to make this semantic web thing go. Now it’s time for the fun stuff: using it to power things that make people’s lives better.

There seemed to be two big uses for semantic technologies that I heard at the conference. First were the groups of folks talking about plug-ins and snippets of code that anyone can drop into their browser or onto their web pages to make an enhanced experience. Yahoo!’s Search Monkey and Google’s Rich Snippets are both examples of simple XML bits that you can add to your pages to enhance your results listing on their engines. Adaptive Blue is a Firefox plug-in that let’s you see your friends’ reviews of books, movies and other stuff as you look at these items on different sites.

The other use is more like what I traditionally think of when I think semantics. There were lots of examples of vendors who can create ontologies and connections by parsing the corpus of unstructured text you give it, whether that text be email, Wikipedia or the Bible. These tools let you see what topics occur in given populations (such as football and the Longhorns in Enron internal email) as well as moving through those related topics. The guys at The New York Times talked about how they use semantic tools to publish their topic pages as well as their news alerts, widgets, RSS feeds and to automate their editorial process.

It was a fun 2.5 days. I learned a lot and am eager to update my personal sites with Rich Snippets, RDFa and microformats.

Join me on Wednesday, June 24 for my virtual seminar on the semantic web. I’ll explain the basics of how this stuff works and why user experience folks need to be involved.

About a year ago, Jesse came to me and suggested I change my title from Information Architect to User Experience Designer. He gave a number of reasons, but none of them resonated with me. I clearly remember commiserating with some dear friends at the IA Summit 2008 about this proposed change in title.

I didn’t want to give up the title. I considered myself an information architect first and foremost. I’ve called myself an IA for nine years now. I was proud of the name. It was who I was. So I didn’t change it.

In Memphis this past weekend, at the IA Summit 2009, I spent a lot of time talking with first time attendees and those new to the field of information architecture. I hosted a round table at lunch for those new to IA. They were a great table, with tons of questions.

One of the things they really wanted to know was how to become a great IA. My answers surprised me. I didn’t tell them that they had to master multi-faceted classification or be able to generate thesauri and controlled vocabularies from scratch. I didn’t tell them about stencils and templates for making better wireframes.

I told them how important it was to listen to the customers of the organizations they would be working for and to deeply understand their behaviors and motivations. I told them to be champions for the user. I told them to listen to the pain of their clients, and think about how their designs could ease it. I told them not to go in shouting about CVs and classification and indexing and how their clients were doing it all wrong. Be subtle, I said. Listen for their needs. Present classifications and metadata and all that cool stuff as the way to get your designs implemented, not as an end in and of itself.

And I realized… I wasn’t telling them how to do good information architecture. I was telling them how to do good user experience design. I realized while I love IA, and it is my core competency, it is also only a small part of what I do.

For that reason, I am taking on the title of User Experience Designer.

I’ve been looking for a metaphor or a model that I could use to describe how the Information Architecture day of UX Intensive is structured. The day is focused on metadata, controlled vocabularies, classification schemes and search. They sort of build on each other, but not in a simple, neatly stacked way. I was thinking about this while in Copenhagen a few weeks ago, when the answer hit me: Stone Soup!

Do you remember the story of Stone Soup? It’s a Grimm Brothers’ tale about returning soldiers and their guise to get a selfish, starving town to learn the lesson of cooperation and its benefits. They can make soup from their stone, but it will be a more tasty and filling soup if they get the whole town to pitch in and add ingredients.

Information architecture is like Stone Soup. You can make a website without explicitly thinking about the IA. You don’t have to use metadata or control your vocabularies or develop thesauri. You don’t have to tweak your search engine and play with recall and precision to improve your results.

But it will be better if you do.

Putting structure into your unstructured data allows you to make your site that much better. It’s a way to “plus” it. A way to add some “BAM” to your site, to borrow a phrase from Emeril Lagasse. Because it’s easier to slice and dice and do interesting things with structured data than it is when your data is a big, undifferentiated mass.

IA from a stone? Fancy that.

This weekend I attended the BlogHer conference in San Francisco. There was lots of talk about traffic to blogs, and what you can do to increase readership, and generally promote your blog. Most of advice made sense, but there was one thing mentioned that got my blood boiling.

I was in a session on DIY Content Syndication and Promotion, and one of the audience members asked how you could use tags to help with promotion. One of the speakers, I don’t remember which one, advised that the best way to use tags is think of the most general topic you post is about and tag with that. Also, if you are commenting on someone else’s post or video, you should copy all of their tags and add a few of your own.

Um… excuse me? Sure, that’s best practice if you want to add tag spam, water down results and piss off people when they come to your post only to find that you are tangentially related to the topic they are interested in. Remember, it was this broad spectrum, shotgun approach to tagging that taught search engines they couldn’t rely on the keywords metadata field..

The rules for tagging are very simple:

1. Tag only significant mentions.
2. Tag at the level the item is about.
3. Use one tag per concept.

I like to use this rule of thumb to check to see if the tags I’ve chosen are accurate: If I did a search for the tag I’m considering, would I be happy getting this post/image/content item? If the answer is no, I drop that tag.

Following these rules to tagging insure that your tags are appropriate for your post/image/content item. Targeted tags help ensure that recall as well as precision are high. You want the folks who are interested in your specific topic to find your thing. If you are talking about the giant green dinosaur with glowing red eyes in South Dakota, you don’t want to tag your post with general terms like “United States” or “statues.” Better choices would be “Wall South Dakota,” “dinosaur statue,” “glowing red eyes,” and “kitschy roadside attractions.”

Recently on projects I’ve found myself designing a number of search results pages. While each project has its own set of requirements and nuances, I think there are a handful of elements that should be included in most all result page interfaces. If you start out with this list, and then tweak as your situation requires, I think you’ll end up with a pretty good page.

Here are the items on my checklist, in no particular order:

• Highlight the query term in the results.
• Restate the query on the results page.
• Show the number of results that were found.
• Include next and previous buttons, as well as links to additional pages, to move through results. These should be smartly linked; no link on previous if you are on the first page and so on.
• Include a query box so the user can search again.
• Don’t show the URLs of the result pages, unless your audience is techy enough to derive meaning from the URL.
• Have meaningful page titles and descriptions for each result.
• The page title should be the link to the result.
• Allow sorting and refinement tools if appropriate for your users and content.
• Indicate if a result is not a regular page (e.g., a PDF file).

What items do you have on your checklist?

I’m a big fan of indexes. There are many a content-focused website whose content could be made more findable with an A-Z index of the site content. There are lots of places on the web that talk about the merits of such indexes. Most will tell you to put a row of the alphabet at the top of the index, and have the letters be quick jumps (i.e., anchor links) to that section of the index.

But what about those letters that don’t have any entries? Do you show the letter and have it link to a message saying there’s no entries? Do you show the letter but have it not hyperlinked? Do you just remove the letter all together? Which option gives the better user experience?

This very question came up recently on a project I’m working on. My gut told me to show the letters but not make them links. But why? I looked high and low on the web for someplace that told me which was the better way, but I couldn’t find anything. Looking at examples of indexes wasn’t overly helpful for I saw sites doing it in all kinds of different ways.

So I turned to the wisdom of the crowds. I asked the question on Twitter and Plurk, as well as the Argus Associates Alumni. (Thanks to everyone who responded!) The overwhelming response was to show the links not hyperlinked, AND grayed out.

I also got multiple reasons for why this is the better approach:

• It preserves the pattern of the alphabet and makes glancing easier.
• Users don’t wonder why some letters are missing, which can make it look broken.
• It saves users from getting to a destination just to find out there’s nothing there for them.
• Some back-end work is saved if content is added in those areas in the future.

So there you have it. Be sure to include your unused letters at the top of your index, but gray them out and don’t make them links.

Like most folks in the office, I’ve joined the Twitter bandwagon. I find it’s an easy way to keep in touch with folks I don’t see on a regular basis. And learn new things about those a I do see.

Just now I checked out Twist, a site that lets you chart how much folks are twittering about a topic and plots them against each other, over time. It’s a neat way to see how topics ebb and flow. I thought it would be neat to see how much the five senses are talked about on Twitter.

I think it’s fascinating that folks Twitter the most about things they see so much more than any other sense. I would have thought that smell would have ranked higher. I wonder if it’s because we are used to sharing things we saw with our cell phone cameras. Or if it’s just part of human storytelling. “You’ll never believe what I saw on the way home today…”

Dolores Blog showed thousands of colors to people and asked them to name the colors they saw. They then plotted those names on a color Wheel, printed in the color. They have a blog post describing the project. The resulting image is beautiful. They then added a filter so you can search for different color names and see where it is on the wheel. It’s based on a study to test the universality of language.

When I first saw this, I thought it looked like a type of tag cloud. I like how their filter let’s you expand and contract the colors that appear on the wheel. It certainly helps to illustrate how ambiguous language is. I love that there are at least four different colors all called “chocolate.”

It also started me thinking about what other types of visualizations could be done. There are certainly lots of things that could be done intersecting it with other data, depending upon what you are interested in. Being able to see the color names along with if the namer was colorblind, their gender, native language and other demographic data would be interesting. I found myself wanting to click on a color name to get more information like how many times that name was used for this color.

What ideas for visualizations do you have?

I’ll admit it, I’m stubborn. Perhaps it’s my pedantic librarian roots, but I like words to mean what they mean. It drives me nuts when vendors take perfectly good terms and corrupt them for their own purposes. For example, it took me years to accept the use of the term “taxonomy” in a non-scientific classification context. Now it’s something that I use all the time, but I’m sure to be clear what definition I am using. Language is ambiguous enough as it is without vendors and marketers muddying the waters further.

“Ontology” is one of those words that up until yesterday, I felt protective of. It drove me nuts when folks would use it as a synonym for taxonomy or thesaurus. “We have perfectly good words to describe these things,” I’d tell myself. “We don’t need to use ontology as well.” Perhaps it was my own failure to truly grasp even the original meaning of the word that led to this stance. I mean, “the metaphysical study of the nature of being and existence” is a bit much to work into casual conversation. But ontology has just made a huge jump to the top of the list of words I use on a regular basis. Why is that?

This week I’m attending the Semantic Technology Conference in San Jose, CA. I’m finding myself surrounded by big brains who think and dream in OWL standards and RDF triples. It’s been a long time since I’ve been to a conference where they actually show lines of code up on the screen. But this emersion has been great. I’m making all kinds of strange and wonderful connections between what I know and what I sort of thought I knew but now realize I had no idea about but oh gosh am I excited about it now.

The best example of this is “ontology.” I finally understand what those database folks mean when they say it. Ontologies aren’t synonyms for taxonomies — they work together. Ontologies are the rule sets — they set up the structure of how things are related to each other. They take the “related terms” of a thesaurus and completely blow it out of the water with the complexity and depth of relationships that are possible. With an ontology you set up how concepts and information relate together, and then use that as a blueprint to build out the instances, or what I would call the taxonomy or thesaurus.

But the ontologies go one step further than just being glorified vocabulary maintenance tools. Because of their close relationship to data modeling and database content, they allow connections, learnings, and inferences to become clear. And they, along with some other technology that I don’t fully understand yet, make those connections happen. I’ve said for years that building a thesaurus will help improve the results in your enterprise search system, or that you can relate terms in your thesaurus together to auto-populate related content or highlight cross-sell and up-sell opportunities. But I now see that this was only one part of the solution. The semantic relationships within the ontology are what enable the computer build out those connections. It’s the magic fairy dust that makes this stuff happen. And oh boy is that cool!

Founder Indi Young was interviewed by IA Voice, an IA Podcast channel in Europe about her mental modeling process. The postcast of the interview is available on the IA Voice site. Indi describes it as “kind of like a whole course compressed into a few minutes.”