Sign up for our daily recaps of the ever-changing search marketing landscape.
Anything Else? Vivisimo’s Remix Clustering Surfaces Subtler Results
You know that aggravating feeling you get when you’ve done a search and your results are OK, but you’re sure there’s something better if only the “obvious” results weren’t totally dominant? Or if you could somehow come up with a cleverer way of expressing your query without wasting a bunch of time diddling with variants of your search terms? Vivisimo’s new remix clustering feature helps solve both problems.
Vivisimo was a pioneer in introducing on-the-fly clustering, grouping results into topic folders. Do a search on Vivisimo’s Clusty meta search engine and you’ll see these cluster folders on the left side of the search result page. Clustering makes it easy to quickly drill down into a topic without spending a lot of time coming up with careful, precise queries.
For example, a search on “senators” returns clusters including “Congress,” “Clinton, Obama,” the “Ottawa Senators” hockey team, and so on. Click a cluster and you’ll see results about that topic, minus results from the other clusters.
Remix clustering lets you simply click a link to quickly answer the question, “What other, subtler topics are there?” It works by looking at the same set of search results but ignoring the first batch of clusters, pulling up results that may be more ambiguous or that have less powerful signals for the relevance algorithms.
It’s an interesting approach when you’re feeling less-than-satisfied with the search results you’re getting, or when you’re simply in the mood to see what else is out there. Vivisimo CEO Raul Valdes-Perez put it aptly in a blog post: “I’ve been asked whether remix clustering is only for when none of the folder topics look interesting. Not at all! When I select a book off the bookshelf, it doesn’t mean that every prior book I saw was uninteresting. Instead, it just means that I want to see what else there is. Same thing here: What other topics are there?”
Remix clustering appears to work best on short queries with words that have multiple or ambiguous meanings. Give it a spin over on Clusty.