Faceted Search: An Interview with Tito Sierra

I met Tito Sierra at the IA Summit in Miami, and we talked soon afterwards about his experiences with faceted search. Previously at Amazon.com, Tito has spent the past few years at NCSU Libraries, working with a great team to transform the library catalog and identify best practices for search design.

In addition to sharing lessons learned via Endeca at the NCSU Libraries, they've extended their successful model to the Triangle Research Libraries, and created a research testbed for faceted search and navigation.

Highlights (notes not quotes) from my conversation with Tito include:

• We went overboard at first by exposing twelve facets. Our studies showed users suffering from "facet fatique." The new design has a smaller facet footprint, and we removed the prominent LCC browser.
• We're also employing collapsible facets, quickfilter checkboxes, and stacking breadcrumbs to use space wisely.
• Facets are ordered by frequency of use (e.g., subjects are most popular) and grouped by type (e.g., exploratory versus known item search).
• We've created virtual hierarchies (e.g., formats under Videos and DVDs).
• We've designed "facet triggers." For instance, upon selecting an institution (e.g., Duke), users are shown individual libraries (e.g., Law).

I also asked Tito about the future of search. As massive digitization projects (e.g., Google, OCA) mature, he's excited by the prospect of discovery interfaces that leverage both algorithms and structured metadata. Tito also sees potential in personal search and the use of past queries to inform future results.

Finally, Tito is committed to advancing our shared understanding of search. His testbed is designed to make it easy for researchers to experiment with and gather data on a variety of faceted search and navigation interfaces. If you're interested in learning more, please contact Tito.

Strange Connections

Check out The Noisy Channel by Daniel Tunkelang, chief scientist at Endeca.

Karl Fast told me about Evernote. Ready to build a search engine for your life?

Spime Search

Inspired by this mesmerizing Bruce Sterling vimeo, I've added more links and examples to spime search, but I'm plagued by the plethora of even better examples I haven't yet found. If you discover one, please let me know!

Bruce Sterling from Innovationsforum on Vimeo.

Strange Connections

My Search Patterns talk is now a Slidecast (with audio).

Catriona Cornett's inspireUX is worth a visit.

EuroIA is headed to Amsterdam.

Information Architecture Summit

I had a great time at this year's IA Summit. My workshop went well, and though there's room for improvement, I was happy with my first Search Patterns talk.

| View | Upload your own

The quality of the conference sessions was high. My favorites were Content Page Design by Luke Wroblewski, Edge Interfaces by Stephen Anderson, and Placemaking by Dennis Schleicher. I also missed some great ones, as usual.

Other highlights included:

• Running to Miami Beach for a swim and a walk along Ocean Avenue.
• Encountering a huge, green, wild iguana (and living to tell the tale).
• The Argus dinner and the eBay dinner and my big lunch in Little Havana.
• Spending time with old and new friends (and diving into the twitter stream).

Thanks to everyone (especially these folks) who made this year's summit so wonderful. See y'all next year (February 18 to 22) in Memphis!

Strange Connections

For more vicarious fun, see Crowdvine, Flickr, SlideShare, de.licio.us, Twitter, Blogsearch, and stay tuned to Boxes & Arrows for the podcasts.

I'm returning to Florida in four weeks to keynote Endeca Discover 08.

The IA Institute's annual report is available. Five years and going strong!

Brian Goodman: Interview

I talked with Brian Goodman about Search Patterns. As manager of IBM's WebAhead Lab, Brian identifies and develops emerging technologies (6 to 18 month time horizon) that can be applied to solve real business problems.

During the past year, Brian's group has focused on building social spaces (e.g., blogs, wikis) and platforms (e.g., tools for bookmarking, tagging, rating, and reputation management) that support integration. For instance, they designed an enterprise tagging system with widgets that can be integrated into any service.

This platform approach has enabled the integration of social content into enterprise search. The first step was a portlet (on the side of result pages) featuring blog posts and bookmarked articles. Subsequent adoption and satisfaction were higher than expected.

The team then went further by integrating social data (e.g., ratings, tag frequency) directly into the algorithms for enterprise search. A page that's been bookmarked, for instance, receives a boost. In similar fashion to Google's PageRank, this approach to socially influenced search improves both result relevance and user satisfaction.

Other highlights (notes not quotes) from my conversation with Brian include:

• Created a recommender system that leverages folksonomies and taxonomies (which are deconstructed into tags). Users can subscribe to tags and taggers (people). The recommendations are surprisingly good.
• Created a reputation system built on peer review and feedback that became a fantastic source of data for expertise search (which is also integrated into enterprise search).
• These tools help people identify content and individuals they wouldn't normally find by searching.

Looking ahead in search, Brian sees more personalization, and a profound need for new interaction models that make complex queries simple.

Strange Connections

If you can't go to the IA Summit, or if you're like me and can't wait to get there, check out Meet Your Peers by Jorge Arango. See y'all soon!

Behavior & Design

I've added a collection to Search Patterns that includes concept visuals, behavior patterns, and design patterns (e.g., faceted navigation).

Also, with the help of my Blackberry Curve (I dropped my trusty Treo one too many times), I've added several new examples to mobile search. I'd love your feedback and suggestions for new patterns. Thanks!

Strange Connections

Going to the IA Summit? Then, create a personal schedule using Crowdvine.

Will Evans: Interview

I talked with Will Evans about Search Patterns. As an information architect and interaction designer, he's designed search for such sites as Kayak and Gather.

Will is an amazing source of insight and examples. He's already contributed to the pattern library but was willing to share more, including Volkswagen UK (faceted navigation), and the Encyclopedia of Life (taxonomy visualization).

Will noted we must look beyond our immediate disciplines and vocabularies for sources of innovation. In that spirit, he recommends the following:

• Scented Widgets
• Dynamic Taxonomies
• Semantic Search

Will is now with CrowdSprout, a startup that will aggregate supply and demand by using the power of social networking and e-commerce. In other words, it's a search system that helps buyers and sellers find deals. Good luck Will!

Jason Blackwell: Interview

I talked with Jason Blackwell about Search Patterns. Jason, a user experience specialist at IBM, has led an enterprise social search project that uses bookmark and tag data from Dogear to improve search and social networking.

The team wanted to keep the main intranet search results (official data) front and center while offering social data as an extra feature. The initial design adopted the Google model with filter tabs for each content type (e.g., blogs, wikis, forums, news, people). But users didn't visit the tabs very often.

The new design adopted the Ask model with a customizable sidebar (to the right of results) that by default presents three preview results for each content type. Click-throughs skyrocketed. A nice example of how seemingly small design details can make a big difference. Other highlights (notes not quotes) include:

• We review the top 300 queries (which account for 30-40% of all searches each week) and manually adjust the Best Bets as necessary.
• Tags and bookmarks impact the ranking of the main search results.
• We may add personalization and narrow/broaden capabilities.

Jason noted that designers no longer own the search user experience. At IBM, folks use Greasemonkey to add new sources to the sidebar. And, on the Web, StumbleUpon injects social data into Google results. By providing APIs and RSS feeds, we can open search to support further customization and innovation.

WineM

My favorite contribution to Search Patterns this week is WineM.

Here's an excerpt from last year's press release:

The smart wine rack uses radio frequency identification (RFID) to track individual bottles in the rack and identifies ones that fit the users' wine selection criteria. Collectors and restaurants can use WineM racks to search collections, track specific bottles, update information about wine in real time, and manage inventory visually...the wine in a collection can now be dynamically reorganized by any combination of year, region, price, or any other information axis that interests the collector or sommelier.

A handheld device accepts queries, and full-color LED lights transform the elegant wine rack and the bottles themselves into a search results interface. The system even supports faceted navigation. Very cool!

Strange Connections

I'm happy my disturbing hole has wormed its way into the minds of futurists.

If you're coming to the IA Summit in Miami, please consider the IA Institute's Leadership Seminar. An important topic and some great speakers!

Keith Instone: Interview

I talked with Keith Instone about Search Patterns. Keith is a lead information architect at IBM where he's working with the UX and IT teams on the next generation of search for ibm.com. Here are some highlights (notes not quotes):

• One goal is to make search more contextual. Where is the user when they start to search? We can deliver context-sensitive results. But what if our educated guess is wrong? We must let the user drill down or go sideways.
• We're conceiving of search as a separate space with unique layout and navigation for result pages.
• Simply defining a shared vocabulary is a challenge. What's the difference between a filter and a facet? Filters are visible (e.g., link, tab, checkbox) whereas facets are conceptual and "behind the scenes."
• The user interface is not the biggest challenge. First, search is a massive IT project. Given millions of documents, it's not easy to index all and only the right stuff efficiently. Second, getting the content tagged with high-quality metadata (e.g., language, part numbers) is difficult in such a decentralized organization.
• While search is a project, it's also a process. We employ a mix of tools and methods (e.g., the search elsewhere test) to solicit feedback and drive continuous improvement.

Keith predicted that trying to make site search work as well as web search will remain an ongoing challenge. This made me wonder whether it would ever make sense for Google to license domain-specific PageRank data as an input into site search algorithms. Probably not, since Data is the Intel Inside, but it's an interesting idea. Thanks Keith for a thought-provoking conversation!

Strange Connections

The Externalities of Search 2.0 by Michael Zimmer examines privacy threats emerging where search meets Web 2.0.

Charles Knight has posted a list of the Top 100 Alternative Search Engines. Be sure to try them all.