Tuesday, March 3, 2009

Microsoft's Kumo Prototype Wrestles Powerset Search Features into Live Search

While the concept of the content organization features found in the Powerset search application was always compelling, the original content in the demo application set up for the early version of Powerset was not the most powerful presentation of its strengths. Now in the hands of its acquirer Microsoft, the Powerset features appear to be ready to take on a much-improved content set and interface in the guise of an internal project at Microsoft labeled "Kumo." As revealed by Kara Swisher at All Things Digital, an internal Microsoft memo is encouraging staff to play with the prototype search engine to get some initial feedback.

In spite of some scathing negative reviews from the search engine intelligentia, the screen grabs provided by ATD of the Kumo interface look to be pretty competent. Gone is the over-busy Powerset interface, replaced by and interface that is at once Google-esque and yet unique. The top five web results are followed by results that match different facets of a search term. For example, results for the recording artist Taylor Swift return groupings of content available for her songs, her lyrics, her bio and her music downloads and her albums. On the left are possible searches by related artists and categories, as well as the ability to initiate new searches in video collections, bios and so on.

It's unclear at this point whether Kumo will be just a project name - it's apparently a word that means both "cloud" and "spider" in Japanese - or whether it's just an internal marker that may disappear at its features get absorbed into Microsoft's Live Search engine. For that matter, it's unclear that the features will make their way into production at all, though they are certainly useful enough. What is clear, though, is that Microsoft is going to continue to search for new ways to make alternatives to Google palatable in a way that might appeal to both enterprise and media audiences. I don't think that too many people harbor illusions about the ability to crack Google's dominant market share in search any time soon, but competition is good for the breed, they say.

I suppose the most intriguing aspect of Google's success that challenges the challengers such as Kumo is how Google has attained its success without explicit content categorization features. One can go to dozens of knowledge management and search conferences every year and hear about how important good content categorization features are for the success of search engines - and then look at the nearly naked search results on Google to contemplate just how true that may be. The assumption that categorization specialists have is that having categories makes it easier to browse content collections. Well, that may very well be true if you are in fact interested in browsing relatively finite and well-organized collections of content, but in general search engines have become less about browsing and more about delivering specific answers for most people. The average searcher seems to be trained now to refine their own searches via the "white box" rather than to traverse through browsing categories.

This isn't to say that content categorization isn't useful: it's more a matter of where it turns out to be most useful. Where it does seem to help most is in portal solutions where someone has come to a specific page of content and may want to explore that site or database from different facets. Where people understand that there's a finite, well-curated collection at their disposal, categorization seems to do quite well. Where it's a matter of sifting through billions of pages for the needle in the haystack, most folks are getting used to typing in the best search string that they can think of. With that said, the features in Kumo do provide an interesting and engaging alternative to Google search results, but they'd probably be better off either in specific content portals that need enrichment or in creating an on-demand portal from its results sets, so that it will be a more browsable set of content in its own right - and then, perhaps, attract a higher breed of advertising, if that's the goal. Instead of trying to out-Google Google, perhaps challengers such as Kumo need to think about how to out-aggregate the aggregators to build better revenue margins for smaller search operations. Something to wrestle with, perhaps.

Reblog this post [with Zemanta]
Post a Comment