OneRiot, a social search engine, announced today that its search results pages now update in real time with content from Twitter, Digg and the wider social web.
Results are prioritized based on an algorithm of about 26 factors, filtered for spam, and unduplicated if links are shared through multiple URL-shortening services. There are two search modes: Users can browse real-time results or (in “pulse” mode) see links ranked by social relevance. We spoke with a caffeinated and exuberant Tobias Peggs, general manager at OneRiot, about 20 minutes before the new release went live at 9 a.m. “We’re trying to get a sense of current social relevance; what are people talking about right now,” he said. And more than any competing product currently available, OneRiot succeeds.
Earlier this week, we wrote in detail about indexing links from the social web. “The true power of real-time search lies deeper than seeing a conversation stream around a keyword,” said Peggs in an early press release last week. “By digging, tweeting or sharing links to web pages, people are signaling that the content on those pages is relevant to them right now. OneRiot takes those signals, indexes the web page content and returns that information in our search results in real time.”
“One factor in the algorithm is velocity, the hotness factor: How many times a link has been shared on the social web within the past minute?” said Peggs. “We can get a sense of whether a story is increasing or decreasing in popularity from one minute to the next.”
Users’ online reputations also influence the score and help provide a means of ranking social relevance, kind of like Google’s PageRank system. “The trouble is,” said Peggs, “who’s to say what’s more important: A link shared by Scoble or a link shared by me? It doesn’t necessarily mean the links he shares are more important.”
A link’s relative popularity (how many times has a link been retweeted or dugg) is also normalized so potentially relevant results don’t get dwarfed by stories on CNN or major outlets.
New features of the current OneRiot release include searching within specific domains and blogs to determine which articles have the most current social buzz, searching for a specific URL to see how many shares it’s received, identifying the first user to share the link, and expanding results to see every single share along with the user and any surrounding text.
So if OneRiot results are intended to be the freshest, most current links available, when do results disappear? “Because the results change in real time, users can come back and keep searching to stay on top of the latest information,” said Peggs. In other words, top results will constantly fluctuate, and the maximum time a link will appear without refreshed bouts of sharing is a few days, meaning constant vigilance would be needed for social SEO in real time.
Also, because of the real-time nature of the product, Peggs noted, “Users tend to search more times per day per query than on a traditional search engine. Ultimately that’s good for monetization because we have the opportunity to reach those users more times per day.” Monetization plans at this point primarily consist of serving display ads.
When breaking down search behavior, Peggs stated that twenty percent of search queries are navigational (users seeking a specific URL) and 40 percent are queries for static information such as recipes or contact information.
“Google does an amazing job on that 60 percent,” said Peggs, “but the remaining forty percent of users are looking for what’s going on at this particular moment. What are people saying right now? Because of the way Google indexes the web and the amount of time it takes to index and rank pages, they are never, ever going to serve those socially relevant, very fresh search results. OneRiot is not a Google killer; it’s not going to find your dentist’s phone number. It’s a completely different experience.
“What we’re doing is showing people the real, true potential of real-time search,” he continued. “It’s way deeper than seeing a string of conversation around a keyword. We’re uncovering the web content – blogs, videos, news stories. It’s very different from what traditional search can do. Hopefully, people will start to see the potential of real-time search.”