A mobile search engine relying on social signals, location-based services, machine learning and natural language processing to rank and serve queries could turn into a gold mine for marketers and advertisers.
The four cofounders -- from a former Apple intern to a venture capitalist -- include CEO Corey Reese. The co-founders refer to the "Likeness" engine as Ness, which combines search and recommendation in a mobile service.
The engine aims to answer subjective questions by understanding the likes and the dislikes of the person making the query. The difference in factual, and subjective queries reside in questions such as "Which Italian restaurants are in Los Angeles?" vs. "Which Italian restaurants would I most enjoy in Los Angeles?"
Although in the conceptual stage since 2009, Ness Computing took the "talented" team 18 months to develop. Co-founder Paul Twohey, a Stanford University graduate who interned at Apple between 2000 and 2002, said Ness will launch imminently. As a budding developer during his last summer at Apple, he designed, wrote and implemented the first version of a cluster-wide remote software upgrade for the Mac OS X server in less than a month.
Designing a good mobile experience requires different thinking, so some of that "talent" came from Apple. Twohey -- along with other company executives -- recruited Scott Goodson, an original member of and senior engineer on Apple's iOS team, and appointed him director of mobile engineering at Ness.
Reese was formerly in venture capital at Alsop Louie Partners, where he initiated the company's investments in Justin.tv and Gowalla and managed its program that discovers up-and-coming technical talent at leading universities. Reese founded the company with Nikhil Raghavan, a founding member of Yahoo's Structured Web Search team; Twohey; and Steven Schlansker, a multiple-time hackathon winner and programmer at UC Berkeley.
Together, the team will build out the data pipeline to serve up queries on the mobile search engine.
Search queries will pull data from crawlers and licensed content as well as social networks, when the searcher opts in, providing Ness with access to information in social streams and sites, such as Foursquare.
Ness will serve up more personalized results through social signals and location-based services, depending on the quantity and quality of information that searchers choose to share. Twohey declined to detail the back-end of the mobile search engine, such as content ranking factors and the integration of paid-search and rich media ads, and instead focused on personalization.
The search model will also likely support both audio and text search queries from historic searches, supported by machine learning and natural language processing.
The concept took years to craft because the Web lacked social data that people were prepared to share, Twohey said. "Today, we're only using the shallow signals, such as check-ins to the Lady Gaga concert," he said, which might reveal the person arrived late or bought premiere parking passes for $25 to eliminate the hassle of finding a place near the stadium.
Aside from preference messages in comments or likes that express sentiment, there are other signals that could provide useful targeting options. These signals in aggregate -- a bonanza for marketers and advertisers -- would suggest the value of a potential audience segment, for example.
Twohey continues to carry the experience from Apple with him. "When I worked at Apple, we thought of the user as the teacher -- someone who wants to get their job done -- and your product is not their primary purpose, just a tool that helps them get to their real goal," he said. "That's the biggest message I learned from working at Apple."
Ness might not have any direct competitors, but engines such as Google and Microsoft -- banking on mobile search and advertising -- already have a head start. Research firm IDC estimates that Google's Android operating system will take 38.9% of the worldwide smartphone market, compared with Apple at 18.2%.
Ness on Wednesday announced raising $5 million in Series A financing in November 2010. The round, led by Vinod Khosla and Ramy Adeeb of Khosla Ventures, also had participation from Reese's group at Alsop Louie Partners, as well as TomorrowVentures, Bullpen Capital, a co-founder of Palantir Technologies and several angel investors.
Ness plans to start as a mobile service and add search for computers and possibly tablets and other mobile devices in the future. The company's team of three machine-learning PhDs is led by Dr. Jeremy Schiff, who was formerly president and co-founder of Fotoflexer.
The company has operated in stealth mode since it was founded in October of 2009, when it received seed funding and was developed by Alsop Louie Partners.