October 16th, 2009 by Collarity
Interesting article (Google’s vision for future of search) in the Telegraph where Alfred Spector, Google’s VP of Reasearch and Special Initiatives, describes Google’s current R&D direction. According to Spector, relevance evolution involves hybrid intelligence, where computers learn from the search habits of people. Hmmm… that sounds familiar. Some excerpts:
He also predicted a future in which “hybrid intelligence” – the principle of computers learning from the search and browsing habits of users to refine search results and serve up the most useful pages first – would be a key technology underpinning web search and results delivery.
“If people click on a result, then that’s an indication we probably got it right,” he said. “We can reinforce the behaviour of the system to weight it towards that pattern. It’s a really powerful, data-driven approach.”
The vision of relevance moving from the link-driven-back-end (link topology/pagerank) to the user-driven-front-end (searcher implicit feedback) sparked the launch of Collarity in 2005. The idea of search migrating from a library model (the books you see on the shelves are the same as the ones I see, organized by a linkogopoly) to a more personalized model, where your search results are driven by your choices and help from like-minded communities of other searchers.
This blog, starting from our introduction to this post, chronicles the evolution of Collarity’s people-driven relevance approach. Here is a group of selected posts illustrating many of the ideas.