Search & Serendipity: Finding More When You Know Less
Today, Just Behave welcomes the second of our guest columnists, Larry Cornett, Vice President of Search Experience at Yahoo.—Gord Hotchkiss Where do you go to get your daily fix of fresh information? If you are like millions of other users out there, you hit sites like Digg, Delicious, Techmeme, or just open your favorite RSS […]
Where do you go to get your daily fix of fresh information? If you are like millions of other users out there, you hit sites like Digg, Delicious, Techmeme, or just open your favorite RSS reader. Or, you tap into your Stumbleupon toolbar to see what your fellow Stumblers have found interesting lately. Most people don’t automatically think of search as a place for serendipitous exploration. Search is best when you have something specific in mind, right?
Actually, search has been evolving a lot lately. And, as we seek new and better means of disambiguating intent, the search engines are increasingly offering richer media and information that feels a lot more dynamic and exploratory.
As I mentioned earlier this year in an interview with Gord Hotchkiss, disambiguating intent is one of the biggest challenges that search engines face today. It is difficult to understand what users really want when they just enter a few words into a search box.
Communicating intent to the search engine is a challenging experience for many users as well. search is great when you know exactly what you are looking for and you know how to ask a search engine for it. Things can quickly become frustrating when you don’t.
So, how can search engines help users discover great results when they may not even know how to express their query? How do we weave the power of serendipity into the search experience?
Classes of search intent
For some search use cases, there is a pretty good understanding of intent. Navigational queries, for example, have become an experience in efficiency. Enter a company or site name, hit return, and the official site is typically one of the first few results. However, Andrei Broder, Research Fellow and Vice President of Emerging Search Technology for Yahoo, has found that navigational queries only account for a fraction of overall web search intent.
The other two major classes of search intent he describes in A Taxonomy of Web Search are Informational (e.g., the user is seeking information to learn more about something) and Transactional (e.g., the user is seeking a site where further interaction will occur, such as a purchase). Disambiguating user intent for these classes is quite challenging without specific multi-term queries.
Consider the typical query for a musical group such as “3 Doors Down”. What is the user actually looking for? He might want to read about the band, find their tour dates, see photos, watch a music video, or find lyrics to a song. Maybe even all of the above.
Transactional queries can be just as challenging. What type of transaction does the user have in mind for a broad query such as “ipod”? Is she looking to make a decision about what type of iPod to buy? Is she ready to buy an iPod or is she looking for support for the iPod she already owns?
Current explorations in search
Go visit any of the major search engines today and you will encounter their current attempts to disambiguate these types of queries. You can definitely see a testing of the waters of serendipity and discovery, ranging from offering query refinements to presenting richer media on the results page itself.
Ask.com offers a federated approach to their search results with “Ask 3D“. For the query “3 doors down“, Ask provides multiple ways for the user to disambiguate the query: Query refinement links, a shortcut, an images module, a list of popular songs, etc. This approach definitely offers a variety of information on the page to allow users to discover much more than they initially asked for.
Google began offering a Universal search experience a few months ago. They take a different approach to disambiguating a similar query by offering a shortcut at the top of the results, a blended video result, some related searches, and an images module at the bottom of the page.
Yahoo! search has also been offering a richer results presentation with its latest set of Shortcuts to allow users to discover more than they expected. They provide a rich content shortcut for this artist query at the top of the results which offers a link to her official site, albums, lyrics, photos, videos, and the ability to listen to popular songs or watch a music video right on the search page.
Finding more when you know less
These are just a few examples of how search engines are helping users discover more when they may not even know how to formulate all of the queries that would generate these types of rich results. They also illustrate the power of disambiguating intent via content, instead of just relying on text link suggestions.
I think we are just now seeing the early steps of expanding search beyond a pure focus on relevance and precision to offer an opportunity for serendipity. As the millions of people who use Digg, Delicious, and Stumbleupon can attest, there are times that you want to explore and discover. There are times that you want to see the interesting things that others have found. There are times that you don’t know exactly what it is you are looking for, but you will recognize it when you see it.
New on Search Engine Land