Yahoo's Search Future Found In 'Web Of Objects'
The next evolution of engines will converge search, a URL produced by an algorithm, with portals. It stores and serves content from articles to images to videos laid out on a Web page that looks something similar to an interactive newspaper.
At least for Yahoo, an algorithmic approach to serving up content on portal pages by ranking non-URL content similar to search continues to improve results, revenue and click-through rates, according to Raghu Ramakrishnan, chief scientist for search and cloud computing at Yahoo, and a fellow at Yahoo Labs.
The concept isn't unique to Yahoo. Google and Microsoft view the evolving landscape to combine search and portals similarly, but Yahoo believes it holds the advantage because the concept plays into the company's strength as a curator and content aggregator. The differences defining "search" and "portal" will dissolve, he says.
Numbers from comScore suggest that Yahoo sites were the most visited in December 2010. Yahoo served content to 181 million unique visitors, followed by Google with 179 million and Microsoft with 177 million, according to comScore.
While content today on Yahoo's front home page differs from Yahoo Entertainment, the future search experience will combine content found in search results and on portals. Yahoo engineers realized this and began to make the change in late 2009, transitioning the back-end infrastructure on Yahoo's front page, a portal, to an algorithm that picks and serves editorial content. Yahoo plans to embed the algorithm on all its portals to better optimize pixel ratio and revenue, enhance user engagement and personalize every page.
The combination gets a bit tricky, considering that Bing powers parts of Yahoo's back-end search results. The Bing-Yahoo deal allows Yahoo to analyze click logs, Web crawls and add Yahoo search technologies across the sites. Since the organic search results influence the type of paid-search ads that appear, Yahoo continues to monitor changes to determine whether the additional technology enhances its partnership with Bing to power the back-end.
"Organic search results become a blended experience, with a chunk owned and created by Yahoo, and the other chunk -- URLs -- coming from Bing," Ramakrishnan says. "How do we ensure the paid-search results are constant? It hasn't proved to be a problem because the keywords capture user intent."
Ramakrishnan says the feud between Google and Bing that produces the list of ranked URLs doesn't directly affect Yahoo either, although the company is a "consumer" of those URLs from Microsoft.
The future of search on Yahoo looks like a "Web of objects."
Ramakrishnan defines it as a database of recognizable concepts, such as Julia Roberts or the city of Cupertino, Calif. Related information is pulled in from across the Web and hung on these concepts to create a spider effect. From this gigantic index of information, content gets served up on the portal using semantic concepts, rather than keywords.
The percentage of data pulled from Yahoo, compared with Bing, continues to grow, according to Ramakrishnan. All media assets come from Yahoo, for example -- content Microsoft doesn't provide. When site visitors search for content not contained in the 10 blue links, Yahoo pulls the content from its database and delivers it in search results. "Today, it's a non-trivial percentage and growing rapidly," he says, pointing to local as an example.
Microsoft provides a list of 10 URLs, but Yahoo crawls the Web to classify, extract, and analyze and "makes sense" of the content before giving it to those searching the portals and sites.
"The big story around how I'm going to understand what you want, and then give you highly relevant semantic content, falls apart if you can't more precisely infer user intent," Ramakrishnan says.
"The game has changed. None of us -- Google, Microsoft or Yahoo --really understand how to deliver the next immersive experience. We think Yahoo has a real opportunity to win through technical excellence and this Web of objects."