Search: how to clarify user’s intent? Kosmix comes with an impressive solution


 

At Kelkoo, we observed that, out of the top 100 most frequent queries per month, about 60 to 70 of those queries are brand names. Yes, people searching for “samsung” or “apple” and expecting to get relevant results for their query. But what are the users’ intent when searching for a brand name? In May / June, we also got an impressive number of people searching for “nintendo wii“… Were those users looking for a game console or videogames or accessories for this console? Again, how to understand the user’s intent?

As Peter Morville, a father of the information architecture field, recently explained on his blog:

“[…] we should distinguish between clarify and refine. First, we must clarify the meaning or context. Are we in the right ballpark regarding the searcher’s intent? Clarify is all about disambiguation. Then, we’re ready to refine or narrow. […] Refine is about increasing specificity. […] It’s a subtle distinction, but from a designer’s perspective, I think it’s a valuable way to frame the search process. Clarify, then refine. What do you think?”

Clarifying the user’s intent is one of the areas where, I think, Google currently fails; and where a lot of companies try to innovate. Ask launched Ask3D in 2007; Yahoo! launched Search Assist and more recently Glue Pages. And tons of startups are trying to find some solutions to this problem, and Kosmix is one of them. As described on TechCrunch:

Kosmix, until now a vertical search engine for information about health, automobiles and travel, transformed itself into a universal search engine for all subjects earlier today during a general redesign. […] Now when users conduct keyword searches on Kosmix (much as they would on a traditional search engine like Google or Ask), they are presented with mashups of results from a variety of sources. Unlike Mahalo, Kosmix itself doesn’t publish any of the content it displays. Rather, it pulls it all from services like Flickr, Google, Wikipedia, TheFind, Yahoo Answers, Amazon, Truveo, and YouTube.

 Canon eos

For shopping-enabled queries, Kosmix aggregates content from Shopping.com, TheFind, Yahoo! Shopping, eBay, Amazon… Quite an impressive list of data sources for a big mash-up… But where Kosmix brings value is in its “Related in the Kosmos” module.

Search for “Samsung” on Kosmix:

Kosmix understands that “Samsung” is a brand focused on semiconductors, so it suggests related companies (like Sanyo or Nec). It also highlights the areas where Samsung is successful/popular: consumer electronics, electronics, and – getting into the details – mobile phones. For mobile phones, we get a list of popular mobile phones produced by Samsung, but also some generic searches regarding mobile.

Search for “Apple” on Kosmix:

For “Apple”, we got the Apple company and a list of products / technologies from this company; but also some grocery items related to the “apple” fruit (but we miss the Apple record company from the Beatles ;) )

Search for “Nintendo wii” on Kosmix:

For “nintendo wii”, we have a list of Nintendo consoles, video games for the Wiii, generic items regarding video games industry…

I believe Kosmix comes with an impressive solution to clarify user’s intent when searching, and I hope they will expose those data through APIs. That could be a clever way for english-based CSEs or retailers to improve the findability on their sites.